On-demand session here – registration required.

Rachel Stephens joins Ned Bellavance and Kyler Middleton on their podcast, Day Two DevOps, to discuss the state of DevOps and the impact of AI. They explore the distinction between developer productivity and development productivity, underlined by a DORA report finding that while AI dramatically boosts individual developer productivity, it often fails to improve overall development productivity. They also debate the necessity of a “human in the loop” to maintain software stability and organizational performance.

Listen here!

Links

• LinkedIn: Zach Lloyd
• X: @zachlloydtweets

Transcript

Rachel Stephens (00:12)
Hi, welcome to RedMonk Conversations. I’m Rachel Stephens and with me today is Zach Lloyd from Warp. Zach is the CEO of Warp. It’s a company that I first heard about back in 2022. And at the time they were thinking about the product in terms of re-imagining the terminal experience. In particular, trying to make it easier for a broader group of developers to use. So it wasn’t just a tool for power users, but it was something that everyone could be more effective with and made it easier to learn, less of a steep learning curve.

I was impressed at the time and since then the market in DevTools has shifted so very substantially that it has pivoted, I don’t know if pivot’s the right word, but it’s definitely expanded its scope in some really interesting ways. Zach, I’m so excited to have you here. Tell me about your journey building

Zach Lloyd (00:58)
Yeah, so I’m Zach. My background, I’ve been an engineer for a really long time. The mission of Warp has always been the same. It’s to empower developers to ship better software more quickly. Your sort of preamble on Warp is pretty accurate. we set out with a product vision of trying to reimagine the terminal with the idea being like this is a everyday tool for most developers.

It’s something that from like a product experience, I don’t like very much. Like, just think the stock terminal people are wasting a ton of time, experiencing a lot of frustration in terms of like how hard it is to use. And conversely, if you get good at using it, you can get a lot more done. the current version of warp is quite, is like similar and different. Like we, basically have taken all the work that we did to make it an incredible terminal experience and made it work for agentic development. And so.

We don’t even call warp a terminal at this point. It’s really, we call it an agentic development environment. And the idea is like, you can have a single interface for telling your computer what to do. You can tell it to do that in terminal commands if you want, and that works really well in warp, or you can tell it in natural language, in English. And if you tell it in English, then warp launches an agent to achieve that task.

And because of like where the terminal sits in the development stack, can do almost any kind of tasks. So Warp’s really, really good at coding. We’re one of the top agents on Sweet Bench and Terminal Bench, but we’re also really good at like setting up projects or debugging in production or doing stuff with Docker or your clouds. And so it’s like a really, really versatile, powerful spot in the stack to be. And like you said, that’s like transformed our business, which has been awesome.

I would say we were like kind of half right in our original thesis and just like, you know, we, I went into this being like the terminal is like an overlooked piece of real estate. It’s kind of like how I would imagine it. And like our original business model was going to be building collaboration features on top of that. And like that kind of sort of worked actually. And like, we have traction with that, but the, you know, we got pretty lucky. Like this transformational technology came along like, you know, a couple of years into the business, which applies super well.

to what we’re working on. now, again, we couldn’t have really predicted this, but the terminal is sort of like the center of the agentic development universe. Things like Claude code or Gemini CLI or warp are like the preferred interface for doing agentic development. So it’s been really good for us.

Rachel Stephens (03:16)
Mm-hmm.

Yeah, I have so many questions. So when you were originally thinking about collaborative terminal, that was like helping team members basically, if you have terminal commands, you can share them across your team. Was that the original version?

Zach Lloyd (03:36)
Yeah, the easiest way I can see of it is like Postman for your terminal. So Postman is like this API collaboration platform. A lot of that same type of thing applies in the terminal where it’s like, you might be executing complicated commands or you might want to put your on-call playbooks in a place where you can get to from the terminal. We even built like real-time, you know, multiplayer in the terminal, which is very cool, but like not that frequently used. So that was like…

Rachel Stephens (03:40)
Okay.

Okay.

Yeah.

Zach Lloyd (04:04)
And that kind of goes to my background. Although I was principal engineer at Google, I used to run engineering on the Google Docs Suite and I built a lot of Google Sheets. so, as a founder, think you kind of like do what you know. And that was a thing I knew how to do. But I think we’ve been pretty flexible and adapted well to the fact that market has changed. overall, it’s been a really positive change in the developer tools market. It’s gone from like…

A market where it’s really hard to make money, where it’s like you have some of the fastest growing startups in history are in this market now, which is cool.

Rachel Stephens (04:35)
Absolutely. It’s been a profound shift in the I won’t pay for any developer tools ever, too. I will pay $200 a month for some developer tools. It’s been…

Zach Lloyd (04:41)
Well, so

here’s the thing. It’s like, don’t even think of the market as like developer tools. I think that’s the wrong framing for it. It’s like, I think it’s the market for automating the production of software. that’s like, that’s where this is going. Because like,

Rachel Stephens (04:52)
Okay, how do you think about it?

That’s fair. Okay, yeah.

Mm-hmm.

Zach Lloyd (05:03)
Like, look, there is a developer tools aspect in terms of like a lot of what these tools do, Warp included, is they help your developers be more productive. But I think the really, really big opportunity is like, if you look at like the market for the production of software, and that could be enterprise software, consumer software. And I think the more interesting part is enterprise, honestly, not like the long tail of like five coded websites, but like…

Rachel Stephens (05:17)
Mm-hmm.

Zach Lloyd (05:26)
That’s a huge, huge cost to companies and also like a huge economic driver. And we’re getting to the point where big pieces of that can be, you know, can be automated. Whether it’s like, you know, you still need developers in the loop. So it’s not like total automation, but pieces of it are going to be totally automated. And that’s just like a gigantic economic opportunity.

Rachel Stephens (05:48)
Absolutely. So when we’re thinking about where you envisioned collaboration kind of being human to human collaboration, team collaboration, how did that kind of translate into building agentic tooling? Because it’s still collaboration, but like in a very different flavor of collaboration. Like, are there parallels or no?

Zach Lloyd (05:54)
Yep. Yeah, so actually the pieces, there are like, so the thing that we built on the collaboration side, which we call warp drive is like this knowledge store. And the original concept was like, you and your teammates can put organizational knowledge into it. That’s useful for using the terminal and like now all that knowledge is just available to our agents. So that’s, that’s like one very obvious thing, which is really cool.

So if you have this corpus of commands or notebooks or environment variables, or these days like shared prompts or whatever, the agent now, like we set it up so you can do rag over that. And the agent just understands your context way better. So we have that in addition to MCP and just like having the ability also to like share MCP, we have the ability to like share rules. So all of these things that make agents not just work for individuals and work for companies as part of our collaboration story, but.

I think what you’re asking is like, you actually as a developer now, the way you develop software is you kind of collaborate with an agent and the stuff that’s actually more useful for that that we’ve built has to do with like how we change the user experience in the terminal itself. And so in warp, like we have the terminal session isn’t just like a stream of text. We’ve essentially made it something that’s much more like.

you know, like a chat experience where it’s like the user says something, the agent responds, the agent does tool calls. Those tool calls can actually have like rich renderings, meaning like you can see like an actual diff. If you want to go and edit that diff, you can edit it. And so just the fact that we put all this effort into making the terminal interface, something that we own gives us this huge advantage when it comes to other tools that run in the terminal, which are just like,

plain old text-based apps and you can’t do any of these things with them. So that’s what I think a very differentiated offering in this space compared to everything else out there, which is either like a VS Code fork or it’s a pure text-based app and warp is like, no, we’re like an actual like native GUI app that has like some of the best parts of the terminal. Like it has that interface, that form factor, but we can also do things like code editing or code review or whatever we want with the UI and the terminal.

Rachel Stephens (08:11)
So that sounds kind of like a lightweight IDE embedded in the terminal. Is that their way to think about it?

Zach Lloyd (08:16)
So we call it an ADE. We are attempting to create this category of a Gentic development environment where our position on this is like neither the IDE nor the regular terminal is the right interface for doing a Gentic development. The way that development is happening is fundamentally changed. And what you want is something that is designed from the ground up to support that workflow. And like it’s going to have aspects of the terminal for sure. Like, but it also has aspects of the IDE.

Rachel Stephens (08:18)
Okay.

Zach Lloyd (08:43)
but it’s not an IDE because in the IDE, like the primary interface that you have looks like Microsoft Word for editing your coding files. And like it’s really geared towards hand editing. I mean, no shade to Microsoft Word, like hand editing, handwriting code is not the primary thing that developers are going to be doing. So having an entire interface dedicated to that doesn’t make sense. It’s like the secondary action. And so the primary action should be prompting an agent.

Rachel Stephens (09:05)
Right.

Zach Lloyd (09:11)
I think like the secondary actions that you really want to support are like reviewing the agent’s code. So actually if you’re like, I think code review becomes much more important and not like code review, in GitHub or whatever, but code review in the inner loop as the agent is writing code. And then if you need to having the ability to go and edit that code, but that’s, like I said, that’s, that should be your fallback. That shouldn’t be what the primary interface is geared towards.

Rachel Stephens (09:33)
Interesting. So when you’re thinking about your competitive set then, are you kind of thinking more of like the AI CLI tools or is it like the AI kind of IDE tools or is it a mix of both? It’s all of them?

Zach Lloyd (09:44)
It’s all of them because like, it’s like whatever tool someone’s adopting to build software by prompt or to build software automatically. And you know, there’s different interface approaches to it. Like there are the AI CLIs like cloud code and there’s the AI IDEs like cursor. I would just think of the competitive set in terms of like what you can do with the tool, not what the interface is. And so in warp, can, you can code. Like I said, it’s really, really good coding.

Rachel Stephens (09:45)
Okay.

Zach Lloyd (10:09)
We also, again, one thing I think we’re much better at or better positioned for than like the AI IDEs is doing all this other stuff that goes on in the software lifecycle, figuring out why a server is crashing or like setting up Terraform or like, you know, interacting with your AWS, you would never want to do that through an IDE. So that is our competitive set, but we are like, in a sense, even broader in terms of what functions you can do beyond just coding.

Rachel Stephens (10:31)
Yeah. Interesting. I like the concept of what the code can or what the tool can do versus the form factor of the tool. I think that’s going to be increasingly shifting in the next little bit. So talk to me about how you are building context for warp. I know you mentioned that you can do rag over warp drive, but like what are the primary places in which warp is kind of deriving all of its information?

Zach Lloyd (10:40)
Yeah, so there’s a lot of them. there’s code-based context is one thing I think is important. Actually, not every tool invests in this, but we vectorize the code bases and do embeddings on them and have empirically found that that actually increases efficiency by finding things more quickly in search for some use cases than just like grabbing over the code. So code-based context is one. Another big one is just like…

The fact that we’re in the CLI and CLI tool calling is such a wide surface of things that you can bring into the context window by executing a CLI command. So that’s probably our number one way of doing it is like, OK, if you want to understand your G Cloud setup, just have warp run a G Cloud command and ask about the servers. So there’s all this context gathering on demand for that. There is MCP. So we’re, you know,

fully supporting MCP and actually trying to make the MCP experience and work really nice and work not just for individuals but work for teams. Because MCP has this activation energy which is annoying. There’s stuff like rules, which is another big one. So you can do that on a per project basis or you can do it through our collaboration platform and get it globally. And that’s like, that kind of gets you most of everything.

You know, like through those tools, like again, the warp is super flexible. You can do web searches, you can attach screenshots. So like that’s other stuff that we’re looking at too is just like, do you, what are all the tools that you can bring in context? But it’s like a pretty rich set of things that you can bring into the agent. And then the other thing that we do a bunch of work on is like, with all that context, how do you present it efficiently so that the agent is looking at just the right stuff at the right time?

And there’s a lot of work that goes into the harness there, whether it’s summarizing context, doing subagents, truncating long context, using different models for different types of things. So that’s another place where we spend a lot of time working.

Rachel Stephens (12:50)
Interesting. there’s so many ways we could go with that, I do, I don’t want to do, I don’t want to derail us too much. I’m curious how you’re thinking about the vision of just agendic development overall. What do you see coming together here?

Zach Lloyd (12:53)
Yep. Go for wherever you want.

Yeah. So I think that there’s like three phases of development, roughly is how I see it. So there’s what I’ve done for most of my career, which I conceive of as developing by hand and that like in that flow, I’m like opening a code editor and typing code and then opening a terminal and running commands. I think we’re transitioning from that into developing by prompt. and like, I think we’re really early into this, despite all the hype, like the, the,

use cases where development by prompts is most successful for is what you would think of as like vibe coding. And it’s really like, you know, building websites or web apps. It’s kind of like the successor to Squarespace or WordPress, in my opinion. And like for professional context, we’re still very early in agentic development, but in agentic development, it’s like, you’re opening a tool like warp. are as a developer, telling the computer what you want to do. And that could be building a feature or fixing a bug.

And then I think there’s the third phase, is like automated development. And we’re interested in that as well. With that, like there’s different approaches. So there’s like the approach that we’re doing, which is basically trying to provide developers with programmable versions of our agents so that they can, you know, have the agent do some amount of their, their work on their own. that would be like, like we just did a hackathon in this. And for instance,

having an agent that automatically updates your documentation based on every code change is like, think a canonical example. And so we’re providing a version of warp where you don’t have to sit there in front of warp, typing into it and like watch it run locally. Instead, you can program it to handle triggers or run on crons or respond to system events and do some portion of either code writing or just honestly, it could be any like developer task.

Rachel Stephens (14:40)
Thanks

Zach Lloyd (14:45)
All of the stuff that we have in Warp’s coding agent is really useful. So yeah, those are the three phases by hand. It’s these interactive agents and then it’s like automating parts of the development workflow using agents.

Rachel Stephens (14:55)
And so like the concept of the asynchronous background running agent I think is compelling to me because in a lot of my experiences trying to play with agent workflows Like I end up kind of being like the XKCD comic about the compiler It’s like I’m just waiting for the agent to run better go have like a sword fight in the hall It’s I don’t want to lose my flow But I also like so I end up just waiting and that is clearly not the efficient way to do it So it’s like is that we’ve got to keep moving to that that next phase. It really is

Zach Lloyd (14:59)
Yeah.

Yeah.

Totally.

It’s the new compiling.

Rachel Stephens (15:25)
So for me, it’s fun to just kind of hear people’s visions on how this all comes together, because I think, as you said, we’re still really early in this phase of trying to figure out how all these workflows get incorporated.

Zach Lloyd (15:35)
Totally, yeah. from a business perspective, think automation is the best. It’s just like productivity improvements are hard to measure, but if you could just be like, sell automations or sell a piece of developer infrastructure where they can build these automations, which is how we’re thinking about it at Warp, then it’s more outcome oriented and you can charge in a different way. And it’s just easier to measure the value. So I do think that’s where it’s going.

Rachel Stephens (15:57)
Can you talk to us a little bit about your pricing? Cause I know you’re not doing tokens, right?

Zach Lloyd (16:01)
We do a credit system. the pricing for all these tools is very complicated, like for us, get, there’s our basic pricing plan as you buy a fixed number of credits and that recurs every month and that’s a subscription. And if you exhaust the number of credits, have the option to go onto a pay as you go model where you can just like buy credits all the cart.

Rachel Stephens (16:02)
Okay.

It really is.

Zach Lloyd (16:26)
That’s okay. It’s like, I don’t know the right, like I would be lying if I said exactly, I knew the right way to do this. Token pricing I’m like a little bit against from a business model perspective, just because it makes it very hard for us as we improve the efficiency of the model to capture any of those gains. If we’re just like, if it’s just like raw tokens that we’re charging for. Like if we do something, like I think we should be incentivized as a company to make the interactions more token efficient.

And if we’re just like passing through token costs or not. And so that that’s the reason I don’t like that model very much. I do think some sort of consumption based pricing is like pretty important. Otherwise it’s just like, what’s that? Yeah. You just end up spending so much money. It’s very, it’s very hard to price the fixed request plans.

Rachel Stephens (17:05)
Otherwise you get eaten alive. Yes. Absolutely.

Zach Lloyd (17:15)
in a way that’s incentive aligned without some sort of consumption based pricing, because otherwise like we end up just like losing more money, the more money, the more people use our AI and what we want to be in a position is like, we make more money as they use it more. So that’s like a consumption model. But yeah, it’s like, it’s messy for sure.

Rachel Stephens (17:19)
Thank I just I love hearing people’s general philosophies on all these things where we’re going how we price it like all of these things are all still so up in the air right now that it’s just interesting to hear how people are approaching it and what they’re finding works and doesn’t Wonderful well is there anything I didn’t think to ask about so like Like I don’t want to make you pitch but like if somebody is looking at warp Versus like a cursor like why would someone want to use warp versus like one of the AI tools that they’re already using What’s their switch point?

Zach Lloyd (17:43)
Totally.

Yeah, so great question. the, way that we’re trying to differentiate is really on the developer experience, meaning like when you go in to try to build something with agents, we want warp to be the easiest to use. We want our agents to be the highest quality. We want to give you the features that really let you see what the agent is doing and control it and review it. We’re really focused on working on real code bases. Like we’re not.

positioning ourselves as a vibe coding tool. And so if you like using Claude code or Gemini CLR, if you like that way of working, you should strongly consider a warp because it’s like, it’s a similar style of working, but with what I consider a much better developer experience. And like similar, it’s not more expensive really. It’s like maybe like it’s in the, it’s comparable cost wise, but just like, from a product and usability standpoint, that’s like my DNA, what I really care about. And so when you come in, you should just be able to get more done and have like more fun and doing it and work.

Rachel Stephens (19:01)
Well, Zach, thank you so much for carving time out of what I’m sure is a very busy schedule as a CEO to talk to me. But I appreciate it, though. This has been great. And if you have anyone who wants to look up Warp or look up you, where should they go?

Zach Lloyd (19:05)
I’m very busy. Yeah, but this is fine. This is cool. For sure. Just go to warp.dev. You don’t need to look me up. If you want, I’m on LinkedIn. I’m writing my thoughts on LinkedIn, that’s like the social channel that I’m most on. I’m not very good at Twitter, but just go to warp.dev. Check out the product. if you’re, we have a preview built too for people who want to get the early access to stuff. So, but just go check it out.

Rachel Stephens (19:20)
Okay, fair enough.

Thank you very much. Have a good day.

Zach Lloyd (19:38)
Thanks, Rachel.

Links

• LinkedIn: Jamie Dobson
• Jamie Dobson, Visionaries, Rebels, and Machines: The Story of Humanity’s Extraordinary Journey from Electrification to Cloudification (SRA Books: Bristol, 2025).

Transcript

James Governor (00:12)
Hey, it’s James Governor, founder of Redmonk, and I’m here with another founder, a co-founder, Jamie Dobson, co-founder of Container Solutions. Welcome, Jamie.

Jamie Dobson (00:21)
Thank you James, thanks for having me. You will notice my attire.

James Governor (00:25)
Not at all. I know that is amazing. There is a great conference called MonkiGras. And it will be on March 19th and 20th next year, 2026. The theme will be prepping craft. The idea is being prepared in software and in life. And so basically a bit of resilience and basically thinking about the things we need to do to be prepared. Cause it seems like…

Systems are just ready to give us a slap in the face at the moment. So But we don’t want to talk about that What we want to talk about today is Jamie Dobson’s work He has created a great book That well, I’ve read I’ve enjoyed it a lot. I think most people that have read this book have found it to be You know a bit of a bit of a cracker it’s been called a page-turner

It’s been called a murder mystery. Actually, maybe it wasn’t murder mystery, but it’s a good, a heist, a heist, not a murder mystery. Sorry, it’s a heist. And the book is called Visionaries, Rebels and Machines. And it’s got some lessons for us in tech. Ask some big questions, some tough questions. But yeah, Jamie, first question. What’s the book about?

Jamie Dobson (01:28)
I’m high. Well, the book, well, the title, Visionaries, Rebels and Machines, only gives a little bit away. The subtitle, How the World Went from Electrification to Cloudification, goes a little bit further. The idea was there are lessons in the past. Fornication, electrification, cloudification.

James Governor (02:03)
Did you say fornication?

Did you say-

sorry! Sorry, cloudification, sorry. I didn’t hear you there, Jamie. I thought it was the-

Jamie Dobson (02:11)
Well, no worries. The whole stupid project started when somebody said, what’s the cloud? And in trying to answer that question, I found myself going back further and further in time. then when you go back in time, so basically as any writer or historian ever sort of drew their bow back in 1799 and let loose an arrow that went all the way through Edison’s work benches, through Bell Labs, the invention of the PC at Xerox PARC.

to the cloud. It’s never been done. There’s a book about PARC and there’s a book about the cloud and there’s a book about Edison, but they’re all very much deep dives. So where is that long journey? It’s a relay race of

James Governor (02:50)
Mm-hmm.

Jamie Dobson (02:52)
stories. And the thing is I’m a cheeky, can I swear Jim? I don’t think I can swear on your…

James Governor (02:57)
Well, I mean, I just, I mean, I just totally derailed it by talking about fornication. So I think, I think we better. I mean, I, yeah. Right. Let’s, so let’s jump into some swears then. Yes, absolutely.

Jamie Dobson (03:02)
Californication. So I forgot what I was going to say now.

James Governor (03:12)
Well, the arrow had shot all the way to the beginning. You were saying that lots and lots of books about systems thinkers and yet there was not a systems view of the emergence of all of these systems.

Jamie Dobson (03:23)
So I was a cheeky bastard I’ve been trying to tell people for years computers are stupid They just do things very fast. You shouldn’t be scared of them But people don’t believe me to say you’re only saying that because you’re clever and I’m like I’m not that clever right if I was clever I’d have done physics at school right instead of computer science so the The point is I was a cheeky bastard because basically in chapter one I said this is a light bulb and then in chapter two I said well This is a this is a light bulb with a wire in it and it amplifies a signal

James Governor (03:39)
Mm-hmm.

Jamie Dobson (03:51)
And then a few chapters there I say, this very small thing does exactly what the vacuum tube does. It’s called a transistor. And then before you know it, you’ve got microchips, microprocessors, computers in the cloud. And the idea was to leave the reader feeling clever and engaged. Wow, now I know a f**k ton of concepts about computing that I would have never have learned if I had to start with the cloud or with a computer. built it up very cheekily. That was a point.

James Governor (04:16)
In that case then, so who is your reader Jamie? mean is it for nerds, it for fake nerds, real nerds, business people that might want to learn, you know, who is your reader? If you need to trick them into enjoying the technological explanations, who do reckon your reader is?

Jamie Dobson (04:28)
It wasn’t meant to It wasn’t meant to be the nerds. It wasn’t meant to be me or you or our friends in the community. But our friends in the community read it and liked it because I think they saw themselves in the story. The original plan, Complete Mea Culpa, was around content marketing from our work at Container Solutions. We have this great book, Cloud Native Transformation, Patterns of Change. All the middle managers and the team leads love it.

It’s technical, dense, it’s done well. People read it and they give us a call. But then from a business strategy, it’s like, if only I could speak directly to executives at the airport. So I need to write a more general purpose book. That was how it started. But then those crazy kids at the DevOps meet up in London asked me to do a talk when I was almost finished with the book, not quite, but almost. And there were lots of young people there. And they asked me at the end to, you know,

speak to them about the olden days so obviously I was feeling dead chucked I was there and there was a bunch of people and all excited and I was excited because I don’t get out much and these people said can we speak to you about the olden days so I started thinking yeah cool mainframes no problem and the microprocessor and they stopped me and they said no sorry we mean in the 90s just before we were born so that’s what they wanted to they wanted to talk about where does the cloud come from and I was like oh my god

James Governor (05:55)
Yeah, yeah, yeah, yeah, yeah.

Jamie Dobson (06:01)
So there was another reader, the young recent graduate, know, a computer programmer or designers or people who were moving into tech and actually didn’t know how they got from there to here. So anyway, they became the next reader.

James Governor (06:09)
Thank Okay, yeah, I mean, think that that just comes through so strongly is, I mean, you talk about that sort of that arrow shooting through time is there are just lots of different points at which you can stop and enjoy the sort of the innovation journey that we’ve been on. How did you, I think from my perspective.

I mean, he had to do a really good job of like saying no to ideas, right? So to write this book and like, so for me, Apple barely gets a mention, right? Apple is sort of, yeah, we’ve done the Xerox PARC and then we’re like, well, surely now, surely now we’re going to find out how Apple became this dominant company and changed the world. And Jamie’s like, now we’re not going to talk about that.

Jamie Dobson (06:54)
F**k them. Yeah. It was difficult, James. It was difficult. And I upset a lot of people on the way. for example, back at ARPA in the late seventies, when, when Steve Jobs was running off with the PC, Vint Cerf came up with the TCP/IP protocols. He only gets a one sentence mention. my friend, Adrian Moa read a few drafts and was saying, where’s this person? Where’s that person? And he wasn’t, he wasn’t wrong. And unbelievably, Claude Shannon doesn’t get a mention. Does not get a

Claude Shannon who it was to software what the transistor was to the hardware element so it 50 % was just making very difficult decisions and I’ve actually got chapters about Shannon and about other people that just are not in the book So there was a there was definitely a taking away element that was hard However, I helped myself by doing the following the three ways you can look at an electrical circuit. There’s the there’s the way

the physicists, there’s one there’s one view of the circuit. It’s the physicists view electrons and currents moving around. Number one. Number two, there’s the gadget view of a circuit. You put something on it, a light bulb, a receiver, and it becomes a thing. But the third way of looking at electrical circuit is informational that when you close a circuit, a signal can be sent.

James Governor (08:06)
yeah, yeah, yeah. Well that’s, yeah.

Jamie Dobson (08:28)
all of the things I talk about to do with information processing. So I don’t talk about the radio because that’s more on the gadget end of the spectrum. I don’t mention that in the book, but the main point is we go in from Edison’s workbench and the arrow is the arrow of the informational sciences. That’s where we go from Edison’s effects through to cloud computing. That allowed me to frame it. So that meant that Marconi is not in. It means that John Logie Baird television is not in.

But so that was easy. However, choosing which of the visionary rebels and machines to keep in around computing was definitely challenging. And I think if there’s going to be a version two, so to make up for the Claude Shannon issue, I am going to dedicate a full two part episode to Shannon on the podcast, which comes along as a complimentary sort of set of content to go along with the book.

James Governor (09:22)
Okay, and so partly, so Amazon works, they claim to work backward from the customer. Because you were working backward from Amazon, you had to say, no, you’re not allowed in. I’m not gonna, but there is gonna be second book, is there? Is that, that’s a thing.

Jamie Dobson (09:39)
Well no, so what I’m doing right now, I’m having a lot of fun. hope my my listener and my readers having a lot of fun. I’m now on substack and every time there’s a podcast, a supporting podcast, welcome to Visionary, Rebels and Machines, my name is Jamie That type of thing. And the first chapter, the first…

James Governor (09:51)
Mm-hmm.

I think you should always speak like that Jamie,

I want you do the rest of the podcast like that.

Jamie Dobson (10:02)
So the first episode was about Volta and his frog’s legs and that’s absolutely brilliant because basically people say, well he invented the battery so what? And I’m like what the fuck? He invented the battery for the very first time humanity had a continuous current. It never been done before and at the same time people were thinking if that electricity can move a frog’s leg maybe we don’t have a soul and maybe there’s not a god so everything changed with the battery. I didn’t have time to put that into the book. So episode one is an exploration of Volta, Mary Shelley, Frankenstein, it’s all connected. And then the idea is that…

James Governor (10:39)
Yeah. Have you seen the new film yet? Have you engaged? I mean, it’s out now. I mean, you know, you’ve got to, maybe we should arrange a viewing together, Jamie, because I really want to see it.

Jamie Dobson (10:44)
I can’t wait for that to… When is it out? When is it out? I saw it was coming out on Netflix and I thought I’ve got to see Frankenstein and anyway, so I don’t know if there’s gonna be a book too, but there’s definitely short articles on substack the philosophy of the battery the machine itself a little exposé of Volta it links back to the podcast and the idea is that we’re going on a rampage from 1799 right the way back to the future and it’s all gonna end on my 50th birthday next August that’s the idea and

If the episodes are any good, if the sort of what we record is any good, that will form the basis for either an ebook or a sequel. But the most important thing is to finish the podcast and finish the substack. And I’m desperate, James, to fight against the enshitification of the internet. All of the content is coming from me. It’s all curated. There’ll be links to books, to papers and bread, bread trails. I’m going to do a little bread trail and say, okay, you’re a leader. This is where you can go to learn about psychological safety.

James Governor (11:26)
Okay.

Jamie Dobson (11:55)
you’re a techie this is how you can learn about that so it’s my gift slash bit of fun between now and when I become 50 next August

James Governor (12:05)
Okay, so psychological safety. When I was reading your book, it was quite interesting because I, in fact, my book was published yesterday. Thank you very much. So it’s a book about what we call progressive delivery. It was written with Kimberley Harrison, Heidi Waterhouse.

Jamie Dobson (12:17)
Congratulations.

James Governor (12:34)
and Adam Zimman. And it’s really about, we just felt that there was some missing aspects. When you think about CI/CD and where we’ve come on this long journey through agile, there were some things that you could do that are based on sort of cloud and cloud abundance and cloud automation that you couldn’t frankly do.

Or it was very expensive and hard to do when, you know, the, the, the, the, the giants that put, you know, what sort of, you, if you think about the kind of work, the people like Martin Fowler or Jez Humble were doing, I mean, it it was, if you were going to have two instances, if you were going to do a blue green deployment, you know, Fowler talking about a blue green deployment, right? It was literally two.

Massive systems that you needed to have next to each other, each with expensive Oracle database licenses, expensive networking gear, yeah, just Cisco stuff. Probably, I don’t know, whatever it was, the customer was using EMC storage and yet to two of them. That you would deploy the application to in order to see how this, you know, obviously some sophisticated.

sort of, you know, routing there, but you had to have two entire infrastructures if you wanted to do a blue green deployment. That’s really expensive and really hard. So without virtualization, super hard, virtualization change things a bit, but the cloud and cloud abundance meant that suddenly it’s like, why would you only have two of things? You, you could have, you can have a bunch of things and see which one works best. That’s not really thinking about the user and

the, how the user feels about the application that you’re rolling out. And we’re seeing that in AI now, actually. I mean, we’re seeing people, they’ll have eight different agents spin up a different version of the application that you might deploy. It’s abundance thinking. It’s a very different way of thinking about the world. And so we’ve just tried to capture all of that. Now, what I was actually going to say is that one of the, in the book, we had to talk about the notion of psychological safety and what we’re trying to.

developers and organization in order that we can experiment and in order that we can roll out applications confidently and safely on a Friday afternoon and the sort of culture you need to get there. And so I was really, well, a couple of things happened. A, your book is annoyingly good, Jamie. was, I was really fell in love with the author, authorial voice. I thought it was excellent. And I, I, I had to beat myself up a lot.

and just feel like here are some things that I and my co-authors could have done but psychological safety, what’s that and why is that so important?

Jamie Dobson (15:34)
Psychological safety, feels like a modern term. It is important, but in a nutshell, it’s about having a tolerance for failure. many companies come to me and ask, we want cloud, we want multiple experimentations, but they’re terrified of failure. Failure is something that other people do. So psychological safety, as we know it today, the Amy Edmondson version of psychological safety is that it’s a group level phenomena.

where people feel safe to fail. They feel that if they try something and it goes wrong, they won’t be criticised or humiliated or punished in any other way. It’s a group condition. That’s what Edmondson figured out. So if we were working in a team together, James, and I wanted to try something a little bit mad and a bit random, there would be no fear in the back of my mind that later in the pub or at the end of the week in retrospective, I would be told off or punished or told that was a mistake.

It’s such a difficult thing to build. It’s easy to lose once you’ve got it, but to build that, leader, the person in charge needs to model that behavior. They need to own up to mistakes. They need to control the urge when something goes wrong to move into that punitive mode. Everybody does that. If you spend 50 grand on an experiment, it’s blown up. And if you’re a reasonably small company, of course your first reaction is to go, well, could we’ve done anything different? That’s actually not a bad question, but you know,

or what have you done there? Yada, yada, yada. So without psychological safety, it becomes impossible to experiment. And all of your work into progressive delivery and experimenting our way forward or growing through failure into the future requires that psychological safety to be there. So you cannot, and this happens to me all the time, it appears in the book, you cannot say, I want to do experiments and I want to be cloud native and I want to be like Netflix and…

I’m not going to do the psychological safety thing. I mean, just kidding yourself. This is the foundation of that and not the other way around.

James Governor (17:36)
Right. And I think that comes through really strongly in the book. And interestingly, because it’s a book about technology, but because it’s a book about innovation, it is a book about culture. yeah, particularly this organizations that succeed in delivering sort of breakthrough, I’ll call them digital,

products and services today, but certainly these breakthroughs, that’s what comes through. You’ve got to be able to fail. That’s fundamental.

Jamie Dobson (18:13)
You’ve got So the great thing about psychological safety and the culture of teens is it proves a really important point. Human nature doesn’t change, technologies do. So the exact same environment that gave rise to the incandescent light bulb at Menlo PARC, which was Edison’s little research laboratory, the same culture that produced those artifacts is exactly the same culture that they have at Netflix. Of course, the technologies in between have changed. How do you get that? There’s two very simple rules to create an environment where…

psychological safety is there and that will produce innovations and breakthroughs. Low bureaucracy, just get rid of it, it completely kills the flow of information and of course lots of information. Thomas Edison, people think of him as an inventor, he was primarily a fantastic manager. He knew how to deal with people, all these German snobs who came to Menlo PARC, he used to haze them, he’s tried to bring them down to the American pragmatist level, not because he was a sadist.

James Governor (18:52)
Thank

Jamie Dobson (19:14)
because you know they were building stuff there we couldn’t we didn’t care about people’s history we cared about people’s future Robert Oppenheimer my absolute I love I call him Robert it’s weird that I know I call him Oppie that was a nickname he picked up when he was at university in Leiden I didn’t mean to become an expert on Oppenheimer

James Governor (19:32)
So you must have loved it when the film came out because you were already, or you might hate the book, but you were deep in your love of Oppenheimer and then this sort of amazing view of it comes out as a film. That must have been wild for you.

Jamie Dobson (19:36)
The problem is… No, no, I the film. I love the film. Go. The thing is, I’ve got it on my desk, I’ve got it in every… I’ve got multiple copies of this book. So this is Kai Bird and Martin Sherwin’s American Prometheus. The film adapts it so well, and I know this book so well that actually nothing appeared in the film that I didn’t understand. Some of the lines I’d memorised. It’s a fantastic film, but I didn’t appreciate it like others did because I knew the story back to front. Oppie was a clever man, really clever, charming…

James Governor (20:15)
Right.

Jamie Dobson (20:18)
But he was a brilliant manager. What does he do in Los Alamos? What does he do? Brings people in, smashes the hierarchy, has a fight with General Groves who wants to compartmentalize exactly like we do in big businesses. And Oppenheimer’s like, we need to break down bureaucracy. Information must flow. He said security might cost us the race, but a lack of information definitely will. So Oppenheimer, more than anything, was cast in the role of manager, which of course,

feel like you know a bit beneath him and I’m like why don’t am I mad am I the only person who sees Oppie as a fantastic manager and Edison as a fantastic manager that’s what they were and my book is meant to inspire other managers that hey there’s nothing wrong with management it’s a noble profession and you can do a lot of good work with it

James Governor (21:08)
I mean, that’s a huge question on our industry, isn’t it? Are you an IC? Are you a manager? You know, the work of people like Charity Majors that talks about this a lot.

Jamie Dobson (21:18)
I love Charity Majors’ work by the way. I love her stuff on leadership. I love Charity Majors, brilliant stuff.

James Governor (21:22)
amazing. Yeah, absolutely. You can learn so much from every blog post she writes. Yeah, Charity is just amazing. I mean, it was funny, again, thinking about your work on what it means to have effective management in tech. I’ve just started reading Sarah Drasner’s book on engineering management. yeah, questions about trust.

how you get a team to work together, the importance of values, the fact that different people may have different values, you need a way of navigating that so they can collaborate together. This stuff is super hard. But I want to ask actually a question about Edison. Did you feel any, I mean, did you feel some sort of responsibility to reclaim Edison? I think these days everyone wants to be like, no, Edison was the bad guy.

know, Edison just was, he was just out there like, you know, killing elephants and stuff like that. And, you know, we should not admire Edison. know, there’s an alternative history in which, you know, Tesla, who had not been screwed over by Edison is able to make a better and brighter future for us all. I mean, was there an aspect of reclaiming Edison?

Jamie Dobson (22:26)
Yeah. 100 % and and and a lot of that I started to fight for Edison after the first drafts came out because people Said to me you you’ve committed a crime one of my friends. He said you really Committed a crime because he was a bastard right and and you know, you’ve not you should have be said he one sentence would have done it I just want to And I’m like, well, okay, but I didn’t just make it up. I’m surrounded by papers and books about Edison and

This bastard view of Edison is not shared by historians and biographers who are way more talented than I am. And so the question is, where does this come from? Was it jealousy? That seems like a sensible place to start. He was very, very successful, both in terms of his output and in terms of his wealth creation. He became very rich. Is it jealousy? Is this what people said about him? If you succeed long enough, because there’s lots of people out in the world that don’t like me either.

and that’s because for 10 years I was a leader and I wasn’t afraid to lead, I made decisions. Some of my decisions benefited the group but not universally. Some people did not benefit from my decisions so it’s easy to say, Jamie’s a bastard. Well, you reverse that and put them in my shoes what decisions would they have made. So if you lead

James Governor (24:01)
very effective operators are bastards, Jeremy. mean, if we’re, you know, we’ve got a rather rich chap in the world at large right now who clearly is capable of getting some stuff done. But also, I mean, it’ll be quite interesting to see how our history looks back on Elon Musk, right?

Jamie Dobson (24:04)
That was it. and He will not be in the book of visionaries rebels and machines part two. He won’t make it. He won’t make the cut. He won’t.

James Governor (24:27)
Right, he’s not… I don’t think he does much on psychological safety, Jamie.

Jamie Dobson (24:33)
No, but I don’t like Elon Musk. The comparison between him and Edison is really unfair. Edison, so this is the rough way this goes. Elon Musk, he’s not really, invents anything, he’s a cuckoo. He goes in there, pushes the founders out, and he claims Tesla for himself. It didn’t actually do any work. And therefore he’s like Edison, because Edison did that. What? Really? Did Edison really do that? I thought Edison invented the incandescent light bulb and the phonograph.

Andy spent 10 years trying to do some mining that didn’t work and he invented the film industry as we know it today. He did all of that. He didn’t steal that. At one point he was so famous if you were an inventor and it’s very hard to be an inventor and an innovator and a businessman, would say to people, listen, I’ll be able to reverse engineer that and come up with something better in a pretty short amount of time. Why don’t you sell me the patents, we’ll stick the Edison brand on it.

and we’ll both get rich. That was a tactic of his, towards the end of his life when he was very, very famous. He gave people a choice. This is where the comparison between Edison and Musk comes up is that tiny thing. But that was only one part of Edison’s work. So was he a bastard? I don’t think he was. The Nikola Tesla thing, the current wars came after he invented the incandescent light bulb. So don’t forget, I said this was an arrow shot in the spirit of computation.

James Governor (25:36)
Right.

Jamie Dobson (25:58)
all of the conflict with Tesla came later. Now, the next thing where it went wrong, where Edison gets cast as being a bastard is because of what he did with the film industry. He invented the camera, I can’t remember what it was called, Edison’s CineMotion, it had a strange name. And he had this idea that you could only license this stuff from him. This was a mistake. So people were like, no, we’re gonna make an alternate version.

And actually Edison controlled a lot of the places you could film in New York and in any case, the light was bad. So people ran away from Edison because they didn’t want to pay the subscription fees and to use this patented stuff. And that’s how Hollywood began. So Hollywood began because it had great light and Thomas Edison wasn’t there. So yes, he annoyed a lot of people when the birth of the film industry arrived. But I think people knee-jerk that he’s an awful person without really thinking about it.

James Governor (26:38)
Okay.

Jamie Dobson (26:53)
I don’t see any evidence in there to suggest he was particularly evil. He wasn’t a very good first husband. So Mary Stilwell eventually died, officially of complications because of an illness, but she was heavily self-medicating on drugs. So it could be suicide in small increments, but either way she died a neglected and lonely woman looking after two children that Thomas Edison barely saw.

James Governor (27:21)
That needs to be on the account. And it’s, again, it’s not ideal if you’re doing psychological safety at work and you’re not giving your family psychological safety. That is not a trade off that I think either of us would be willing to make

Jamie Dobson (27:41)
me earlier about the people I left out the book. I have a genuine sadness that I couldn’t find the space to mention the women behind the men I talk about. And I also genuinely regret not being able to unearth the women, visionaries and rebels in the stories of visionaries, rebels and machines. It’s not that they didn’t exist. They’ve been airbrushed out of history.

So there’s a mental note somewhere in there that I’m going to go dig in and try to reclaim some of the lost figures of our history.

James Governor (28:15)
So I don’t need to bash you for that. Cause that was one of the things that sort of struck me in the book. was, you know, I was definitely waiting.

Jamie Dobson (28:21)
Do you know what? Do you know what it came to me really late? I worked really hard on the research I really really did it came to me late in the in the day so there were two Women that spring to mind there Lynn Conway. Who invented I think a process for designing Microchips and then of course a woman a famous woman from England. You maybe have to help me

James Governor (28:45)
By the way, you’ve got to see some comics that my friend Jim Bolton makes. Jim makes comics of figures in tech history. You’ve got to meet Jim. He’s an amazing storyteller. He’s brilliant. You two will get along. But also, these comics are amazing and he’s got a comic of Lynn Conway that is really nicely done.

Jamie Dobson (28:49)
Yeah.

Cool.

James Governor (29:13)
So yeah, so he, luckily enough, I think, you know, sort of, or at least in that, so he speaks to the people, like, let’s make sure that we capture some of this in their own words and their lovely comics. Brilliant history. And I think the two of you, maybe that’s, I don’t know, maybe that’s another podcast, but we’ll all have a pint together first.

Jamie Dobson (29:25)
Yeah. I’d love to speak to him. So Lynn Conway was a trans woman who left a remarkable footprint on the world that we live in. She started work at Xerox PARC and then went on to develop this chip design process, the Mead Conway design revolution it’s called. But it all came to me at the last moment. And if you think that this was easy for me, I spent one day looking pale and I said to my editor,

I’ve committed a crime, I’ve contributed to the airbrush and I’ve contributed to a great historical crime. then of course, Steve Sherley was on my radar. Steve Sherley is actually Stephanie Sherley, but she told everybody and it was Steve because nobody wanted to deal with a woman. And, you know, there’s loads of these stories and it came to me late and I was close to exhaustion at the end of the book. And my editor Beth said, you’re gonna have to do it in the next edition. You’ve got a ship.

James Governor (30:16)
Yeah.

Yeah, amazing.

Jamie Dobson (30:35)
and in the end that’s just what did.

James Governor (30:39)
When I read it, it was noticeable. I don’t think it’s unforgivable. You’re aware of it. You’re going to do the work, as they say, with later editions. And it is harder because if we think about invisible figures, if you’re reading history, it is hard to know, actually. There is a layer of history that we’re lacking because that airbrushing has been a thing, unfortunately.

Jamie Dobson (31:07)
Well, because they’re invisible, you don’t stumble across them. So it was only at the end of the book when my digging was so deep, did I actually come to these women. So I’m not using that as an excuse. And then even the Lovelace thing, Ada Lovelace is a powerful figure in her own right. And recently she has become recognized as such. But for most of the last century, she was Charles Babbage’s muse.

Now when you start reading about Lovelace, this is utter nonsense, she was the powerhouse in that relationship, he was another devil for an idiot. And then you get to that weird Victorian period where women don’t contribute as much because it was such a break of gender roles, you get Alexander Graham Bell, another bumbling idiot, I mean I really can’t stand him. And you get his poor wife, who basically propped him up, her job was to manage his emotions and then motivate this unmotivated and unmovable Scott.

James Governor (31:37)
Absolutely.

Jamie Dobson (32:02)
She’s the real hero in that story. So in a way I’ve contributed to it, have, and that’s on me and now I need to do something about that.

James Governor (32:14)
Well, I mean, you know, first episode, I you’ve to own it. It’s better than saying it’s not a thing. So, but yeah, we don’t, I’m not, well, let’s, think we had a slight slight gear change. But, but yeah, I think that that’s something I look forward to seeing in your future works.

Some again, and I didn’t know actually when we, mean, obviously I’m an arch and inveterate planner. I plan everything down to the very last detail. So I knew exactly how this conversation would go before we started. I was interested in this notion of sort of innovation and how it works. And in

described some of the characteristics. So first of all ask, what are the characteristics of successful innovation engines, successful organizations that innovate? So I’ll ask you that first and then I’m gonna ask a second question. yeah, what are the, you looked at archetypes in the book. Like what do you see as the archetypes of successful. organizations that can innovate and create new and breakthrough digital products and services.

Jamie Dobson (33:24)
Okay so innovation is a bit like pornography it’s a bit hard to describe but you know it when you see it. Most people don’t understand innovation is quite mysterious nobody you know there’s books about this does it come from flashes of inspiration however there are patterns there are broad brushstrokes of patterns the first thing is that so so innovation is slow it tends to bubble up after Volta discovered

The battery, and I say discovered because it was waiting to be discovered, he unearthed it like an archaeologist. For 30 years it took for somebody to take the battery and put it into the telegraph. So the process of doing something sensible with battery took a very long time. That was number one. Number two, all innovation is recombinations of existing things. So if you look at the battery, the disks and the wires coming out, and then look at the telegraph.

Physically, it’s almost identical where the wires keep coming and then at the end of the wires There’s a there’s a there’s a hammer to tap out my dots and dashes on a piece of paper So you think well, why was the telegraph earth-shattering it went viral it changed the world It changed how we speak to each other It changed everything the Associated Press came out of the telegraph the Western Union did why why did the battery not have such an effect? Well, that’s because the telegraph was the hardware. I just described we combined with

very ancient technology, language. The dots and dashes and the small communication channel between you and I that is bound by words. So it was two ideas melted in the telegraph, the hardware and then the notion of language. So innovation is quite mysterious, but it’s always recombatant. Is that even right? It sounds like something from Mortal Kombat.

James Governor (35:12)
recombinant, recombinant, recombinant, recombinant. I mean, I think we’re in the right ballpark, Jamie. I mean, you know, I think people know what you mean.

Jamie Dobson (35:15)
What type of right am I? I don’t claim to have the best vocabulary in the world. So, so anyways, so those environments that have done well allow these ideas for want of a better word to meet each other, marry, have sex and the consummation of those marriages and new ideas. So Edison’s light bulb is just that battery again, but instead of having a sound on the end, like in the telephone, he stuck a light bulb on the end. So these ideas have got to swirl around. Ideas have sex in between the ears of, you know, human beings.

James Governor (35:51)
We’re obsessed with it today by the way Jamie. started I shouldn’t have said fornication at the beginning I mean, I’ve got you know, I’ve really I’ve really set you off. I Blimey

Jamie Dobson (35:56)
You started this. started this. And so then if you look at Container Solutions, just to use as an example, we always, we designed the company to be flat, to be open. We didn’t have any Google 50 % rule, but we’re always looking for ways to create space. And so if you look at the external secrets operator, it’s a tiny, tiny example. We didn’t start it, but in the mode, when we first got our hands on it, we were component builders.

And ESO, it’s a secret operator for Kubernetes, whatever. So we were building on existing technology, but then the next thing was to be like, well, how do we find use cases for this? All of this requires space and time. In my role back then as the chief executive, I’d have said, what is the return on investment? is, when are we going to see a payoff from this? In other words, I’d have paralysed the whole operation by bringing bureaucracy into it. ESO would have never gone off the ground and then later nabbed with the CCS.

So Xerox PARC, Container Solutions, Netflix, they create space for ideas to bubble up, to meet, and then to become the next thing. Of course, all of that requires psychological safety, which we spoke about earlier. And it’s challenging because many managers, it all starts with the manager. How tolerant are you for failure? I’ve got a high tolerance for failure. And I trust people. I’ve built a lot of big computer systems. And if one of my engineers says, I’ve got this hunch, I just say, follow it.

I don’t care about the 20, 30, 40 grand that you know we’re going to spend in terms of salaries and licenses just just go for it and let’s talk again in four months. Not many managers can do that and in fact when I explain this

James Governor (37:35)
No, can this only happen at the beginning of the process? Because to me, you know, I look at, so if we think about the names you’re mentioning, and right now, you know, if you you talk to or listen to people that have are working at or have worked at OpenAI right now, for example, there is an incredible freedom to just fix problems and build things.

These organizations, at least within their engineering staffs, are finding ways to enable great innovation or new stuff. mean, we both would agree that AI is moving pretty fast. There’s a lot of new cool stuff happening. Google’s had a good year. 2025 has been amazing for Gemini. They’ve had some breakthrough consumer products, Nano Banana crossed over, Notebook LM continues to gain plaudits Gemini is

Jamie Dobson (38:22)
Saw that. They had their first billion dollar quarter.

James Governor (38:32)
So the very good year for Gemini. But you know, some of the, I think my question was when I, it’s pretty hard to have psychological safety if everyone’s afraid that they’re gonna get laid off. And to me, so many of the big firms at the moment, how are they, how can they possibly be sort of inculcating a feeling that you can fail when the management decisions are so predicated on we’re firing people, show me the money right now.

Jamie Dobson (39:09)
Well, there is something called creative desperation. When you think, when your back’s against the wall, you come up with creative ideas, it’s not very sustainable. My feeling is, I don’t know what goes on inside these big companies. Your question is, can this only happen in the beginning? Not necessarily, but obviously, look at the telephone company, they’ve got a product, and then once it’s working, it’s about scaling, driving down costs, and then broadening reach. Once you’re in that cost-cutting operational mode, then of course you’re not gonna

If your absolute goal is to drive profit margins up by 1%, you’re not going to sanction a 50 grand spend on innovation. A lot of companies, it’s very sad. A lot of companies’ innovation strategy to buy smaller companies and not because it’s a, it’s a bad system. It’s just because they know that trying to get a company that is so used to driving down costs, increasing revenues, turning the handle and existing machinery, getting the type of people who run such a machine.

to them all of a sudden innovate over there. It’s just impossible. Psychologically, the people drawn to a large organization have a certain set of character traits. They don’t like risk-seeking, they don’t like novelty, and they have no tolerance for failure whatsoever. That’s why they work there and they do a fantastic job. The challenge for big businesses is they come to me and say, we want to be GenAI and we want to do a bit of cloud computing with the same staff. And that’s not possible. Now, it’s impossible to innovate in a dangerous environment, I think.

James Governor (40:41)
Yeah, and I look around and I think orgs, some of them are gonna struggle because there’s just too much fear. It’s really hard to do your best work if you’re, and I mean, guess that’s, I mean, is that something, does innovation, maybe there is a bit of a, like when the economy is growing, it’s easier or I wonder what the correlation sort of there is, but at the moment, yeah, it’s challenging, not much safety around. So.

Jamie Dobson (41:08)
But it’s not to say that large organizations can’t innovate. So Bell innovated remarkably well in the last century through Bell Labs, but under immense pressure from the Department of Justice, it has to be said, they innovated because their profits went above a certain threshold. It would trigger anti-monopolistic court cases. So they had an incentive to burn their money through Bell Labs, which benefited all of us. And then of course you get this weird situation.

James Governor (41:17)
Mm-hmm.

Jamie Dobson (41:38)
Xerox could innovate as well. Xerox was a remarkable company. And then they came up with the Palo Alto Research Center that invented the personal computer. So you can’t say that they couldn’t innovate, but then what they failed to do with the PC was to monetize it, capitalize on the investment.

James Governor (41:55)
Right. See, I would say the innovation is partly the money. think of the invention, at least this is my definition, the innovation is actually the scaling and the making money from it. The invention, they did the invention, but they didn’t take it to market. At least that’s my…

Jamie Dobson (42:10)
Yeah, that’s a Some people do agree with that. I think it’s a spectrum. when does an invention become an innovation? So yes, indeed, an invention is like you imagine a penitent tinkering, they come up with a component, but it takes somebody else who can drive all the inefficiencies in that component, then scale it for wider societal use, driving down the cost so you can stick it in the hands of tons of people.

Xerox definitely failed to do that. So the moment has to be right and then thereafter you have to succeed in the scaling. So maybe they did some invention and then they did the first bit of the innovation cycle but then unfortunately the last bit of the innovation cycle was Microsoft’s and Apple’s in that case. They were the big winners.

James Governor (43:00)
So… On that, I the other thing, and I guess we have to get to this, and some of it’s AI being blamed, but I think, well, mean, given the subject of the book and the subject of the conversation, lightbulb moments is probably a transition here, and…

AI is very real. Another book I’ve been reading, Gene Kim and Steve Yegge’s, I don’t know if it’s Yegg or Yeggie, I need to find out. Their book on vibe coding. And it’s interesting because…

I mean, Yegge is certainly, like he’s not someone that reflexively jumps on new technology. As they say in the book, like he was, you know, handwriting, refused to use an IDE, handwriting Java code for the longest time, refused to learn Git. And yet 20 years of his life was building systems at Amazon and Google that probably many of us use on a daily basis, but skeptical.

a conservative in terms of technology adoption. But that moment when AI, we realized bloody hell, this is very real. I I think that’s interesting to me because in your book, it’s that gear change where we go to, bloody hell. We’ve come through cloud, we’ve got the compute that has enabled us to do these things with transformers. We’re into AI now and shit, what have we done? You’ve got this.

ear change in the book where, yeah, sort of the horrific downer later in the book where you just start to ask questions about will anyone have jobs? Will we use AI to sort of create viruses that kill us all? You turn into a doomer at the end. And I mean, I’ve got a bit to say about, but yeah, tell me how you felt sort of, the book is quite an optimistic book until

Suddenly you’re like, no, no, I’m not optimistic anymore. And I mean, you don’t, and it’s funny, because one of the things you talk about is not cars, not being able to drive properly. Wait till you go to San Francisco and get in a Waymo, because, or when they come to London, Waymo is, they feel safer than normal drivers. That’s, that’s any cab driver I’ve ever had, a Waymo feels safer. So tell me about that sort of gear change where you’re like, actually things could get bad here, folks.

Jamie Dobson (45:32)
I think things can get bad. History’s arrow makes you think that the way things happen is how they had to happen. So the reason I can be optimistic and joyous about what’s been in the past is because I’ve got the benefit of hindsight. There’s a few things that mark my thinking. Some of them are personal. Some of them are practical. think people get excited about killer robots. I wouldn’t worry about that. They’re going to come. Not right now. The real thing to look at, and if you were depressed after reading section five of the book, you’re to be even worse when I start talking now. The original quote that said one percent of all the wealth is held by 99 percent of all the people within this country. What year do you think that quote’s from?

James Governor (46:14)
Yep.

It’s got worse, hasn’t it?

Jamie Dobson (46:17)
Well it has got worse, in the last five years it’s got much worse. But what year do you think that quote was made?

James Governor (46:21)
Much worse.

Not sure.

Jamie Dobson (46:23)
1906, 1906 in a book called Poverty. as the last century as the last century turned

James Governor (46:28)
Bloody hell. Okay. Did we fix it a bit and then it’s gone bad again?

Jamie Dobson (46:33)
Well, yes, but we required a first world war and then the Holocaust for it to be fixed. And it was only a temporary fix. So if you really want to get upset about all of this stuff, at the turn of the last century, the big trust, the big monopolies, AT &T, Rockefeller and all of these people were hoovering up so much vast wealth that the seeds for social discord were sown. We could not stop that train once it was in motion. Starting with the great crash of 1908.

James Governor (46:59)
Mm-hmm.

Jamie Dobson (47:03)
First World War, all of that building up to the big big crash in 28, which inevitably gave Hitler that platform in the 30s. All of the markers that came before the Second World War are now presently within our societies. The financial crash of 2008 was the equivalent to the stock market crash of the Great Depression of the 1930s. That’s the most terrifying bit of this whole story. And it’s systematic because a computer system or a business based on a computer system

James Governor (47:18)
Mm-hmm.

Jamie Dobson (47:33)
consolidates gains and hoovers them up into a very tiny percentage of the population. So it’s history that is making me think what comes next might be difficult challenges. But I do say in the book, we’ve been here before and there’s great hope in that. We have seen this before. We understand what happens when you impoverish billions of people and what they do at the polling station. So we know this. So I think there’s a fight back. So this was part of my look.

James Governor (48:03)
Yeah, we had a weird, I mean, you know, and obviously, I mean, podcasts shouldn’t be necessarily time specific, but, you know, talk about the news. We had a, the Democrats had a pretty good day yesterday. You know, New York’s gonna have a new mayor. London and New York are gonna have Muslim mayors. There were all sorts of interesting wins by Democrats, actually across the country.

Jamie Dobson (48:15)
Yeah,

Yeah.

James Governor (48:31)
So to your point about optimism, yeah, occasionally humans do take a step. At Colorado, passed a bill, hire people on higher incomes in order to do free meals in public schools. So sometimes we do the right thing.

Jamie Dobson (48:52)
Exactly and the right thing so back in the turn of the last century the Paris underground and the municipal transport systems were put together not by left-wing do-gooders but by politicians who were terrified that if working people didn’t see some of the gains of the know increases in wealth there would be social unrest. So we know this history teaches this common sense teaches this as well so there is some hope we have been through these scenarios before we may not fall into them again.

but I don’t think we should be complacent. And this is very difficult for me because if you ever read a book about really deep dark material, whether it’s type of abuse or anything like that, you can only put so much darkness on the reader for so long before they just completely switch off. And I think that’s normal, that’s human nature. But I’m here trying to say, I’d like us to talk about these dark things. I think we’ve all done well in technology. It’s been brilliant for me. I’ve had a great career. I basically…

James Governor (49:38)
Mm-hmm.

Jamie Dobson (49:49)
fuck around for a living these days, it’s great, right? I read, I talk, I help people, I play with computers. I used to do that for fun before I got a job and then the day after I got a job I got paid for it. I think we need to not shy away from these conversations and chapter five or section five is about trying to hold the industry to account, hold myself to account and start to say, listen, if we can’t predict the future, we won’t be able to shape it. And those who can’t construct the past cannot predict the future.

And if we zone out now, the future will belong to the Elon Musk’s of the world. And we can’t let that happen. So this is part of the fight back. And then I think the second part is that I came from a deindustrialized city and I witnessed the mining strike on TV. This wasn’t abstract. This was happening in my community. Government policy, when it goes wrong, leads to the childhood I had and the city I was raised in. So technology change will always come. There will be winners and losers.

but policy choices by governments will dictate what happens to most of us thereafter. So as a part of section five is advice, policy advice. Well, this is happening. The potential for people to become very rich is happening. What can we do to make sure we all get to use these technologies and we all get to be paid fairly and to benefit? It’s never going to be 100 % equal, but how can we all win right now? Because if we all win, we might intercept social unrest that potentially could follow up if we don’t solve that equation.

James Governor (51:23)
through AI. And I was at a conference yesterday. No, no, I mean, I think people have to read the book. It’s a great book, Section 5, Nightmare. But you’ve got to read it. We’ve got to be clear sighted. For me, at least. I care about people. So, yep, this is stuff you need to think about. I have a family.

Jamie Dobson (51:23)
I didn’t mean to depress you and set you free.

James Governor (51:47)
Um, you know, we want them to be safe. Um, you know, if you think about, talk about Maslow’s hierarchy of needs in the book and self-actualization. mean, you have to start with safety, know, psychological safety is one thing in an organization, but if you don’t have a roof over your head, uh, food on the table, um, you know, that stuff is super important. And obviously I want, you know, me, you, uh, our kids, our families, everyone on this podcast and the wider world to have, uh, those, those base levels.

And yeah, I think we do have to fight for them. But I did yesterday, was at an AI conference and you know, I’m very used to it. And there were loads of amazing people there, like building amazing stuff. there’s like vision models and voice models and there’s all sorts of startups building this cool stuff. And it’s like, this is amazing. at this thing they built. Robots and all sorts of amazing things that may or may not take jobs. And you know, they’re basically, they’re pretty clear goal to make money.

and I, I, I got, I went to have lunch and I just started talking to this chap. His name is Bruno Suarez. And he, I asked him what he did. He said, I’m an oncologist. And I was like, what’s an oncologist like doing at an AI conference in Lisbon? And he said, well, I am an oncologist. And, one of the really important things is how,

Jamie Dobson (52:57)
Right.

James Governor (53:12)
like people report on their symptoms, particularly not the symptoms of the cancer, but the side effects of the medicines we give them. Because the medicines we use are tremendously invasive. They really hurt people. And if you, if they take too much of them, it can shorten their life and it can certainly ruin their quality of life in any life that they have. And we’re really not very good at this stuff. There are global standards to lay out reporting for symptoms.

He kept on using the example of diarrhea and I was trying to my lunch and I’m like, dude, but he’s an oncologist and he just came, he didn’t even think about that as a thing, but you know, the sort of seven levels of, and what does it mean? And how do we, how do we have a consistent way of reporting on our symptoms? And there is a French healthcare company, they’ve got one which is quite rigid in classification. And he was actually just pushing back and saying that his organization and the patients that they were dealing with,

They should build their own system using, and he was building a proof of concept. He had built a proof of concept. It was LLM based. Immediately someone else came, you should use an SLM, use small language models. But basically he built it so that people, the patients could use WhatsApp. They could talk to the machine and it would turn that into standard reporting. And he obviously was quite worried about making sure there weren’t hallucinations. into standard reporting on

the effects of the drugs on their lives. And for me, you know, as I say, it’s, it’s, I mean, there’s people out there that say AI doesn’t work. There are people out there that say it’s all bad. I was really taken with the possibilities and the potential with this wonderful, wonderful guy, Bruno. And I’m gonna…

Jamie Dobson (54:42)
Yeah.

James Governor (55:08)
try and get involved, help him where I can. Because if you can help one person to live a better life, one cancer patient, then that’s amazing. And he was absolutely, could not, there was no way he could have done it without AI.

Jamie Dobson (55:23)
Bruno may or may not succeed, but somebody like Bruno will. One of the huge lessons of Visionaries, Rebels and Machines in the history of tech is that the destiny of a general purpose technology does not lie in the hands of its inventors. It lies in the hands of users. That’s what Bruno is. after the hype, yeah, so after the hype, practitioners, people who don’t give a shit about social media will get busy and they will find brilliant use cases. There’s loads exist already.

James Governor (55:38)
He is exactly-

Jamie Dobson (55:49)
a generative AI stethoscope, you put it on your heart, it does what a stethoscope does, sends a bunch of data back to a cloud and then says, you might have this, this and this. Way more accurate than a human being. I’m not worried about that. think finding use cases for generative AI is already massively in progress and in 10 years time, we’ll see some great use cases. Will it be the end of work? I think some jobs will vanish. I think some basic jobs will vanish. I don’t think LLMs can do what humans can do, but I use… Claude to do all kinds of things mainly around planning I Know what I’m asking Claude to do so now I can do in five hours what I used to do in 15 hours Great. It’s a production boost so things

James Governor (56:30)
my goodness, as an author, I’ve just started using Wispr Flow Wispr Flow is really good. I don’t know if you’ve used it, Jamie. You might use it when you do your vibe coding too, because Wispr Flow, you just open the text area, you know, press the button and it’s really good at transcripts. So anyway, what do want to go too much on? I wanted to stop End on a high note. We’ve got some positives. I think for today.

Jamie Dobson (56:49)
Bye.

James Governor (56:53)
We’re gonna, we’re gonna cap it now, Jamie. But I want to say thanks so much for joining me. It’s been a super conversation. And yeah, once again, you know, I’d say to people watching, Visionaries, Rebels and Machines, absolutely an enjoyable book that helps you understand the world that we’re living in today by that time’s arrow looking back and was a cracking read. And so you should buy it, you should read it. And Jamie, thanks so much for joining us.

Jamie Dobson (57:29)
Thank you, James.

Link here!

Microsoft is a RedMonk client, but this episode is unsponsored.

Links

• LinkedIn: Alex Russell
• infrequently.org
• Bluesky: @infrequently.org
• Roderick Gadellaa, “An Abridged History of Safari Showstoppers,” Webventures, 22 September 2024.

Transcript

Kate Holterhoff (00:12)
Hello and welcome to this RedMonk conversation. My name is Kate Holterhoff, Senior Analyst at RedMonk. And with me today is Alex Russell, Partner Product Architect on the Edge team and Blink API owner. You’re also alumni of Google’s Chrome team and a member of the W3C Technical Architecture group. And you’ve also been a representative of TC39 so Alex, thanks so much for joining me on the MonkCast. wonderful. Okay, so excited to talk about the web today. Let’s begin with JavaScript.

Alex Russell (00:35)
Anytime. Glad to be here.

Kate Holterhoff (00:43)
So you have written a fair bit on JavaScript excess and performance budgets. And you’re advocating more restraint amongst developers specifically. And you’re one of the earliest folks to have been banging on that drum. So at least 2016, I saw your posting about it. So talk to me about where that situation is. the argument that a lot of these frameworks are pulling down more JavaScript that they need.

doesn’t seem to have been going anywhere, but have we moved the needle in terms of where that stands?

Alex Russell (01:15)
So I think it’s an important topic, but I think it’s important because I approached most of this from a strategic lens, which is to say, what does it take to make the web a success? Today,

PC sales are basically flat, right? They have been for most of a decade, pandemic notwithstanding, had a big spike, a lot of that was low price, a lot of that was Chromebooks, which sort of made, you know, making things that were reasonably fast on low end devices much more relevant even in the desktop world. But the most impactful…

shift in computing over the last 15 years has been the overwhelming explosion of mobile, right? Mobile started to happen and then it happened for real for rich people. And then after that, rich people stopped kind of paying attention and everybody else got a phone too. So today for the last 10 years, Apple has never sold more than I want to say 20 % of devices sort of if you average year over year. So 80 % of the market, 85 % of the market, something like that are androids.

And those devices are for wealthy people, which is to say a very small fraction of people who actually buy an Android device. They’ve been sort of like half the speed, dollar for dollar, of an iPhone. Apple devices are brilliantly engineered from a CPU perspective.

If you’re buying a Samsung, if you’re buying a Google device, if you’re buying a high-end Xiaomi device, you’re getting basically half as much CPU per dollar. So that’s a performance problem if you would like to have applications compete, especially things that are mostly interpreted like web apps. And I care about that. I care about that because that’s where most of computing is now. So if the web is going to succeed, it has to succeed on the dominant form factor, which is to say mobile today. And that’s where…

we have had some room to run. So there have been sort of intersecting trends. The one trend, the mega trend has been that,

Everyone who had a flip phone on the subcontinent and in Indonesia and everywhere else in the world has gotten a smartphone in the last decade. And that smartphone has been a cheap Android. It’s, you know, it’s been one of these. These are devices that are $100, $200 new unlocked. They are not, you know, they’re not packing the latest and the greatest. And the CPU situation, if you just look at the silicon that’s been in them for the last decade has been extremely stagnant.

That’s just starting to change in the last couple of years. You might remember how Intel got itself into trouble with that kind of plateau at 14 nanometers, where, you know, year after year after year after year, there was like a refresh and then the refresh, refresh and the refresh of the refresh. The same thing was happening at the low end of mobile, except it was at 22 nanometers. And it was not paired with process that the process stoppage, the plateau was not paired with architectural improvement.

when you would get like the, you know, what was it like, Sapphire Rapids or whatever, refresh and time, it would be paired with maybe a larger cache or better branch prediction or something that was improved in the architecture, which would eke out more performance for that same dollar review. That did not happen.

Instead, happened was the same smartphone every year. I started talking about this one in 2016. This is the Moto G4. This has eight cores and I want to say four or eight gigabytes of memory. So was pretty well endowed at the time. Those cores were all A53s.

the first generation 64-bit architecture that made it big inside of the Android ecosystem. So 2015, 2016, the 64-bit architecture shift happens. And at that same moment, everyone starts making one and they put out their production lines and they all build these chips for sort of mid-tier, right? So this is like a four or $500 phone that was brand new and then like down to $300 when I sort of picked it up.

Versions of effectively exactly this device have been being recreated at ever lower price points every year since. So what does that do? It means that the marginal dollar does buy you more computer every year, but not for the same person. It means that the person who was previously excluded from buying a smartphone can suddenly buy a smartphone. there’s this been a Cambrian explosion of

device manufacturers and models, but they’ve all basically been this phone. And that is, like you can still buy a brand new phone with an A53 core in it today. And that core is a disaster. It’s from a computing, like it’s just a technological, like, what can I eke out of this thing perspective? Assuming that you’re stuck on that 22, 28 to 22 nanometer process node, this thing basically has no cache.

No cache. So you’re spending a lot of your time if you’re not relatively efficient in terms of retiring instructions, just waiting on memory. The memory is extremely slow. The bandwidth to disk is bad. It’s flash memory, right? It’s like flash storage, but it’s slow. It’s like EMMC slow in a lot of these devices. And that has been true as it sort of progressed down the price curve. So we got more compute every year. It’s just that the Wii exploded.

So we got more of the same compute every year rather than upgrading the compute that we had. So what’s happening now is that we’re finally bottoming out. In the last year or two, we’ve started to bottom out.

in most of the densest geographies where almost everyone who previously had a flip phone now has a smartphone. And that’s pushing the value of those devices up, both in terms of like to the person. Like your first smartphone is kind of a lark. It’s a very expensive, maybe nice to have your next one. You’re like, wait a minute, this is most of my life. I need I need this to be good. So you might save up a little bit more and get it. But also the process node has started to finally improve, right? Like the newest cheap devices like these. This is like a Samsung A

23 I think. This will be on an 11 nanometer node, right? Like the phone you probably just picked up is probably on a three or four. So it’s like still two or three generations behind, but it’s only two or three generations behind. So we’ve had sort of a precipitous improvement in the core technology there. I’m afraid to say that the ARM chips are still real bad. They’re just real, real bad, but we’re getting some in-process improvement. And at the very high end in the last year,

If you were to get a one of these very top-end Samsung or not Google, their devices are still not very good. Or Xiaomi devices, you’re getting a top-end MediaTek or Qualcomm part, and at the very high end, they finally started to put more cache into those parts, so they’re spending less time kind of just waiting around on main memory whenever they fall. they have gotten significantly faster. So we’re going to start to see that also trickle down the market in a few years when that sort of, that, you know, architecture improvement also hits the mid and low tier. So that’s the upside. But for at least a decade, we’ve basically

been frozen and the amount of JavaScript that we’re sending to these devices has been going up. Right? We have not been exercising any restraint. And so I sort of look back and think, okay, what does it mean for the web to succeed? I want the web to succeed. How did it succeed last time? Well, last time we had what I think it was like a reasonable amount of overhang, right? Moore’s law was in, you know, full flourish and we got

Intel giving us faster and faster and faster, more, more megahertz every year. And then was more gigahertz every year, through the late nineties and early two thousands. And that meant that there was kind of like extra CPU laying around. So we can start to throw that at interpreted languages and start to build more safely, but also move applications that previously were the domain of only compiled. Applications into the browser. And so we did that, we did that very aggressively and you know, that was the Ajax revolution and know, web 2.0 and all that stuff that was all fundamentally enabled by the physics underneath us.

enabling us to kind of have some some squish in the budget, some some headroom to play with. And that headroom turned into applications moving to abstracted, interoperable, open platforms away from the kind of hard scrabble life that you have to live in as a C++ engineer. So that’s the story. And we haven’t had that overhang on mobile. And so if we don’t, what do we what do we do to get there? Well, the good news is that browser engines have been, at least in the Android ecosystem, improving substantially over the

years. And that’s been paired with real progress in the amount of CPU that folks have had at the high end in the Apple ecosystem. But those are oil and water, right? The ecotone between the Android and iOS ecosystem is like, it’s at the $400, maybe $350 mark.

Below $350 for a device new unlocked, it’s all Androids. Above that, effectively, it’s all iOS, right? It’s like pure market segmentation. It’s a duopoly in the strongest possible sense. And what that has meant for web performance has been that if you don’t devise your website to work very well in that lower tier, it’s going to feel like junk for most people.

So the web is not going to succeed. And that’s where we’ve been. So that’s the long story of like kind of how I got to the place where sort of working on a browser team that was working on Android browsers, we sort of saw this, you know, front and center as we tried to make, make it possible for people to build PWAs. We saw them pulling in these huge piles of JavaScript that might have kind of worked on the desktop, kind of at the limit. But we’re just totally inappropriate and, and effectively unfunctional on the mobile devices that were most prevalent.

Kate Holterhoff (11:11)
I like that you have historically approached this in terms of, over-weighting developer experience at the expense of user experience, which it sounds like you’re kind of articulating here, right? The users of these devices are the ones who are suffering. And, you know, at RedMonk, we’re certainly very interested in how developers are interacting with dev tools. And so, we tend to think of developer experience as being an unmitigated good.

But when I hear folks like Miško Hevery talk about his framework Qwik, he said something similar where it’s that we have over-weighted developer experience, especially in the realm of mobile. And so am I right that that’s sort of how you’re approaching this, that developer experience is, there’s a way to approach this so that it doesn’t maybe sacrifice the UX?

Alex Russell (12:00)
One way think about this is that users get a vote, right? They don’t have to use your crappy website. They can go get a native app, right?

Like nobody wants to go to an app store and like spend like minutes downloading 90 megabytes of what is effectively a browser, but they’re going to do it if your website sucks. so as someone who’s invested in the web ecosystem, your choice of whether or not to make something that’s appropriate for your user base is, is I think of as kind of a vote for your own kind of future employment. But even if you don’t think of it that way, you can kind of view it as like, a market limited question, like how big is your total addressable market?

You can actually kind of chart out based on the device price curve, how much JavaScript is going to be the limiting factor in delivering a good experience for some fraction of the market as sold. So if you just sort of look back over the population of devices that were sold, the good news is that I put out these researchy kind of pieces every couple of years now. I guess that’s the cadence I’ve been doing it on.

Kind of looking at what are the price points? What are the volumes trying to understand the market from that kind of market analyst perspective to see like, okay, has something shifted in the kind of balance of device shipment volumes at various prices or the properties of the devices at the various price points? And the answer has been, we’ve been in a pretty stable situation, right? The median device has been about a 300 to $350 device for something. My camera’s never going to focus.

has been $300, $350 median for something like 10 years, right? And what that means is that like we got to saturation with rich people and they kept buying ever more expensive devices. And so as they did that and their devices got much more expensive to keep the median at $350, you kind of have to sell a lot more devices underneath you, right? And that’s been the trend. And that continues even through the pandemic and through…

inflation and all the rest, right? It’s just been a fantastically stable trend. so what that has meant has been that, know, Miško made a turn. Like he took that kind of desktop-centric thing with Angular and, you know, we’ve talked off and on over the years and I’ve seen him kind of make that same pivot that I made, you know, something of a delay, but like I can understand how you get there from here. I did it too, right? I thought, you know, it’s all fine. And then I tried running the websites that I was trying to build on the devices that we were shipping.

wasn’t fine. Everyone comes to it from their own perspective, so you kind of have to give people some grace. my goal with all of this has been to try to precipitate the conversation around what is it going to take for the web to really succeed on mobile. So I think Miško is on the same wavelength that I am now, and I’m happy to hear it. I think they’re doing great work over there.

Kate Holterhoff (14:51)
Yeah. Awesome. OK. And so one of the things that you are most well known for is co-naming, the PWAs, Progressive Web Apps. And that was in like a decade ago now.

Alex Russell (15:05)
It was a decade ago. I got email. Someone reminded me that it was a decade ago. Yeah.

Kate Holterhoff (15:10)
Wow. So pretty exciting. So we’re 10 years in. We’re kind of dancing around it, but what would you say is the state of PWAs? When I’ve written about them, people feel very firmly that native apps are the way to go. They’re more performant, whatever. You make very good arguments that that shouldn’t be the case and that these are really just small little browsers anyways. are things changing? Do you see PWAs as having

Alex Russell (15:24)
Mm-hmm.

Kate Holterhoff (15:39)
I don’t know, being in the Ascendant, are they going somewhere where folks are adopting it? Where do you stand on that?

Alex Russell (15:46)
So again, sort of putting a strategic lens on what would it take to make the web a success. I think we can both acknowledge that they are not the dominant mode, right? No one’s gonna say that PWAs are how people get their software today. And so I spent a lot of time thinking about what would it take to unlock the same kind of transition for mobile that we went through on desktop, where PWAs are actually quite successful, right? We see growth continuing year over year.

Kate Holterhoff (15:51)
Yeah.

Alex Russell (16:12)
largely on desktop, but also on mobile. You know, it’s been 30 to 50 % every year growth in installation and use. You know, back when I was at Google and now when I’m at Microsoft, the trend keeps going up. But again, remember that that’s inside of an ecosystem that has been exploding, right? Like the mobile ecosystem has continued to grow. On desktop, that’s a meaningful trend as a fraction of time spent. But it’s also part of the larger story of the web kind of eating most of computing on desktop, right? The web is

the dominant ecosystem on all desktop operating systems. Doesn’t matter what you would like it to be, it is. It’s the essential tool. If you didn’t have a browser, you couldn’t get things done on desktop computer today. And it doesn’t really matter which OS you’re talking about.

That’s been heartening to see, but it’s not the same thing as kind of matching the rate or the scale of what we’ve got on mobile today. So I think that’s maybe where the question’s coming from. Again, on desktop, you can go take your PWA to PWA builder, put it in the Microsoft Store. You can get it in the Play Store on your Chromebook. You can install it through Edge or Chrome or any number of other browsers on a Mac, and it works just great. And so we see lots of people doing that.

Being, again, being a relatively stagnant in terms of shipment volume ecosystem, I think that’s a form of success. But I think a lot of web developers have of reclined into that, right? Like maybe trying mobile for a bit and then thinking, well, it didn’t work. And so I’m going to go back to my priors. But coming back to the previous conversation around performance,

web developers were the only cohort that was told that they should bring effectively their same tools, like a static set of tools and technologies with them to mobile. A lot of the framework authors just sort of said, yeah, you can definitely use this on mobile. Should you? Absolutely not. didn’t write like the companies that they were working for would not ship that stuff to mobile because it didn’t work great. Like Facebook famously, you know, had to use Inferno instead of React for their mobile rewrite when it launched because React just wasn’t fast enough.

These are the kinds of things where when you look at what it takes to do the hard yards to really get there, the tools that we’ve had, again, present that of market limiting, that total addressable market contraction moment, because at a baseline, their cost out of the total budget of what you could afford to do a good job is just too high. You basically end up being pre-limited in your addressable market. Part of the reason I want people to make good…

good websites that are fast for most people is that like that’s part of the equation. The other part of the equation is you have to be able to discover them. So this is a long running problem. So getting PWAs launched in Chrome in 2015, long story, probably for some time when I’m retired, I don’t know. But it wasn’t straightforward, it wasn’t easy and it did involve overcoming a lot of internal opposition at Google.

You know, the Android team saw the web coming a mile away. They didn’t want the web as competition. You they saw what happened to desktop. And Apple had basically pulled back almost entirely in terms of team growth and ambition from the web, not long after the iPhone launched. So I would say about, you know, the iPhone 3GS was about the time when the WebKit team, their marching orders basically changed, right? You know, the original pitch was you can make all these great web apps.

job is up there saying you can do this and then iOS 2.0 comes out and then 3.0, 3.0 gives you push notifications and the App Store gives in 2.0 and then suddenly the web doesn’t get those things, right? So all the wealthy people, all the people who make decisions in technology organizations end up in the iOS ecosystem. They all carry iPhones and the web basically stops being a credible thing.

You can go to a place, can discover apps, those don’t include web apps. You can go to a browser, and in theory you can add something to the home screen, but iOS will never tell you that something is installable, it’ll never prompt you. That’s still true today. Now, it’ll give you what they invented, they called smart banners, and those smart banners are still a thing you can encounter. So if you go to a website that’s got this meta tag in it that pairs with the native app, it’ll go and it’ll prompt you to install the native app. That never arrived for the web.

So when we did PWAs on Android to start with in Chrome and Opera, those ended up being that sort of prompt moment, that sort of thing that sort of gave users a hint that they could go get the site. But again, because of that kind of oil and water relationship between the wealthy cohort and the rest of the world, the 80 % of the rest of us,

That just meant that if you were a decision maker, you couldn’t see progress in your pocket. And this compounded our problem, right? Because if you wanted to make a go of it, you would have to do it in the places where Android was dominant. Those are also the places where the phones are slowest, right? So the toxic combination of terrible sort of market positioning for the web, I mean, it should have been an easy sell, In India and Indonesia, folks were getting their first devices.

except that the websites all suck because they’re all built for the desktop or they’re built with tools which are

tuned or built to the limits of what the desktop computers of the day could take, turned into, I think, a series of problematic outcomes. again, we’re sort of rounding the corner on those sort of network and device characteristic questions. But we haven’t rounded the corner yet in terms of capabilities, of like browsers sort of generally being able to do all the kinds of stuff that a phone can do. And we haven’t rounded the corner in terms of prompts, right?

acquisition flows. And that’s specifically mostly a problem on iOS today. So there are big fights happening and I would encourage you to go reach out and talk to the folks at Open Web Advocacy to learn all about kind of how that’s going. I’ve written a lot about it and I can answer questions but they’re kind of the authoritative folks here. But I think we’re stuck at the moment, right? The web is not going to break out until such time as the conditions are… appropriate for that, and as long as Apple keeps its boot on the neck of the web, we’re not going to get to that place.

Kate Holterhoff (22:36)
Yeah, and I want to hear more about that. You have mentioned Apple and Google sort of denying, holding back this, and there’s huge competition, and a lot of this is happening behind the scenes. And I think a lot of folks just don’t even really understand the difference, right? Most day-to-day users, they can’t understand that distinction. But I would be interested to hear, what has Apple done lately that is making this so much worse?

Alex Russell (22:44)
Yeah. Yeah. They’ve been making a serial set of arguments in every capital around the world about how it’s super unsafe to have other browsers and browser engines on iOS. It seems like they are truly terrified of an idea where you might be able to get a browser from Mozilla that is not just a lick of paint on top of Safari, right? Like that scares them to their bones, apparently. And they’ve done kind of everything

you can imagine to prevent it in the same way that they did everything you can imagine to prevent right to repair or sort of USB-C for phones, right? you know. Apple will make a big show of app tracking transparency, but then they will allow Facebook to use in-app browsers to track you. They will allow that to undermine your choice of browser, where your browser could actually defend you and block some of that tracking. Apple provides Facebook with a latent advantage in that kind of tracking, simply because you don’t get the advantage of having your browser standing up for you when you tap on a link inside of one of Facebook’s apps.

Those are the kinds of places where Apple continues to self-preference its native app ecosystem in ways that are harmful to consumers and also raise their costs because they force developers into the native app store. And they don’t allow web apps into that store. And even if they did, they would try to take a cut.

the idea of having safe apps that come from the web, which are sandboxed by default in a much stronger sandbox than they provide for native apps, which have many fewer capabilities handed out by default. Like you have to actually go grant these things one by one. Like I was one of the tech leads for this project, project Fugu, where we added things like web Bluetooth and web serial and web hid and all these things that are thought of as like exotic capabilities in the web. But I gotta tell you, the use is extremely low because the grants are really in your face. Like you, you can’t go get

it easily. It’s not just like you you ask for an entitlement and you pinky swear to Apple that you’re not going to do something bad with it. And then it has to be like some press article that comes out later that tells you, my God, everyone’s like looking at like scanning all the IoT devices around you that suddenly causes them to turn it off slightly sometimes. Right. Like that’s the kind of of

ebb and flow that you see on the native side. And it was the fault of the native app ecosystems that sort of turned this into a privacy nightmare in the first place in many, in many ways. So the web has been much more conservative about that, much more secure and has put in place these sort of interposing layers where you could get a browser that has extensions, right? And those extensions could do things that you thought were meaningful for you about those concerns that you might’ve had.

And sort of instead of enabling those kind of pro user outcomes, Apple has sort of basically said, no, no, we’ve got it. beware of a dog sign really does work, nevermind all of the problems that we’ve documented over the years. pairing that with preventing other browsers and then dramatically underfunding the development of Safari relative both to the revenue that Apple makes, which is now I think reported to be something like $22 billion a year from the web and keeping staffing

effectively capped since the last decade has meant that there is no credible way for web developers to target and make great web apps on iOS. And this is just a downstream consequence of choices that Apple makes every year. Many of choices aren’t new, but you ask what’s new, and what’s new is that Apple keeps doing it, and it still sucks. They should stop doing it.

Kate Holterhoff (26:43)
the way that I have encountered that most recently is I wrote a piece on WASM and where we sit with WASM. And there were several folks online who were complaining that WASM just doesn’t work on Safari. And I was so confused because I went on Can I Use? and it said that it was supported. And then it turned out it was something about JSPI, so JavaScript Promise.

You really have to have that technical knowledge to understand exactly what it is that Apple is controlling when it comes to these app developers, because it’s not always clear. You can’t just look on the surface there. There’s levels to this, the pain that a lot of developers are encountering, and ergo users, right?

Alex Russell (27:26)
Yeah, yeah. There’s a chap, Roderick, I’m just going to mangle his last name, Dutch chap, Roderick Gadellaa, I think is how you pronounce it. And he’s got this blog post, is an abridged list of Safari showstoppers, which is basically like, if you were trying to make a web app at any point in the last decade,

You know, and remember that every bug in the WebKit engine that affects Safari affects every iOS browser. So it’s not Safari showstoppers, it’s iOS web showstoppers, because everything’s broken. There’s no alternative. You can’t get anything else.

Kate Holterhoff (28:00)
Yeah. Wow.

Alex Russell (28:04)
which is a choice that Apple makes every year. And they made it today again. So in addition to not having features, stuff keeps breaking in ways that are fundamental and there’s no alternative. And stuff like just being able to like scroll without everything scrolling underneath you, like you have a dialogue up and like the stuff underneath scrolls, which is not what you want. Like this is just basic app construction stuff. The most fundamental stuff that you need to make a UI that feels good on a mobile device.

Kate Holterhoff (28:09)
oh no.

Alex Russell (28:34)
totally broken half the time, even when it’s available. So there is this big pile of stuff which is missing capabilities.

which are much safer and easier to bring to the web, both in general and for Apple because they have a lower, they face lower costs to implement a lot of these things because they just lean on the fact that they’re only supporting a small number of operating systems. But when they do introduce stuff, like it’s often broken for many releases at a time. And the fact that they’ve got a very slow kind of iteration rate for deploying browsers up until very recently has meant that stuff can be broken for years. Push notifications still don’t work right.

Right? Like, they didn’t have push notifications for a decade. I know that multiple Fortune 5 companies went to Apple and said, hey, what’s going on? Where are the push notifications? On the one hand, Apple gaslit them and said, this is the first we’ve heard of it. And about the second or third time you hear that, that’s interesting. Maybe they just were talking to different people. Who knows?

But then, finally getting some pressure to implement this stuff because regulations coming. They implemented it only for home screen web apps, PWAs, which is a fine choice. I think it’s a fine choice that any browser vendor should be allowed to make. We would make a different choice in Edge, and we would have made a different choice in Chrome. And I think that’s a thing that browsers should be able to sort out for themselves based on their own engines. But neither here nor there. It just didn’t work right. It’s just like,

doesn’t work right in a bunch of cases. There’s whole APIs that are not just nice to have, but totally necessary for a lot of the UIs that you want to build. Where there would be an API there, there would be a property, there would be a method you could call. It just didn’t work. So it passed the test. It passed the public test. And so they got credit for it in CanIUse or even webpagetest.org. It just…

didn’t work. And this is so common with Apple’s approach to web browser development that web developers are rightly skeptical now that any corners have been turned. And I would encourage that skepticism where it’s going to take at least another decade of actually doing the job that they said they were going to do to kind of make up for the decade in which they didn’t.

Kate Holterhoff (30:47)
Yeah, that’s so heavy. Yeah. OK, so we’ve talked about a lot of the problems. I’m interested in maybe pivoting to talk about solutions and what developers can do, what folks in management can do. And you’ve written a little bit about this. So for instance, you say that poor web performance is not a technical problem per se, but a management issue. I want to hear more about that.

Alex Russell (31:11)
So I for my sins. I found myself working closely with a lot of teams who are trying to build, I would think of them as sort of ambitious experiences on the web. And they find themselves kind of reliably in the muck about performance. And so I have become a performance consultant by the by. It’s not, it’s a living, guess, is how I would put it. And the patterns that keep repeating are…

often very sad because people have been told a lot of things by the job, specifically by the JavaScript community. The web development community as a whole doesn’t necessarily believe the JavaScript community’s nonsense, but the JavaScript community intensely believes the JavaScript community’s nonsense. And that nonsense is, is extremely strong Kool-Aid these days. so a lot of tools are sold into this stuff like, you need Apollo. you need to react. And, and if you sit back and you ask, well, why do I need them? And you actually sort of walk back through the chain of logic. What you’ll, what you’ll get to is, well, you need these things.

because you have to do state management on the client. And starting from just a product management perspective, if you sit down and you look at an e-commerce website, the first question is, do I? Like, what state am I managing? I’m going from a landing page with a search box to a search results page that maybe has some filtering on the side. And then I’m going to a product detail page. And then I’m going to a cart.

and then they’re going to a checkout flow, there is nothing in common aside from the logo at the top and maybe some stuff on the left rail and the footer that’s got your copyright information that is common between those pages.

Like what state am I managing that is not effectively server state? Like what are we talking about here? And so you could have an architecture that pushes all of that logic onto the client. And then the question is, well, should you? And the answer is, just from first principles, obviously not. Like that doesn’t make any sense whatsoever. And so.

A lot of managers are told by their engineering teams now that they need to adopt these technologies because they will make their experiences fluid and fast, which may be true in a very narrow set of domains, right? Where you’ve got very long sessions operating effectively with small changes on the same data.

So think about a spreadsheet or dragging a map around or a Figma. These are sort of productivity tools, effectively. And those productivity tools are distinguished by very long sessions and lots of operations that are optimistically committed to the same local data set for responsiveness reasons.

But all the other kind of experiences that you might sort of build have very different flavors. And so if you’re not doing that thing where you’ve got a long session coupled to many updates to the same data.

the premise of these tools falls away. And even if the premise didn’t, we’re now in 2025. And so if you sort of then look at the tools individually and go, well, why is the tool shaped like that? In many cases, the answer is IE6, IE7, IE8. And I got to tell you, that’s not a thing now. It hasn’t been a thing ever on mobile. So if most of your users are on mobile, then you should be thinking about those as like a down-level thing to support to start with, unless you’re overwhelmingly a legacy product.

But even if you are, know, Microsoft doesn’t support anything other than the latest browsers on our top level productivity products. Google doesn’t support anything but the latest versions of those browsers in its productivity suite. I know because I made the argument to both sets of management and one both times. We are now in a place where I can tell you firsthand, that’s not a thing.

Right? Like you don’t have to think about IE and the fact that, you know, React is dragging a hundred K of JavaScript behind it versus Preact, which is like three or four K specifically because of the legacy that it has decided to support. The fact that it could still run an IE eight, nine, 10. If you just provide enough polyfills, that’s why it costs that much. And it doesn’t have to cost that much. So if you just like a leg up, you can just sort of reject the

the tools that come from legacy and think about your problem fresh. And that sets you on a better path. Because a lot of what’s happening in these organizations is people are thinking about technology and not the problems that their users have or the problems that the business has about the experiences they’re trying to deliver. And that is a management problem, right? That is foundationally a question about like, are you pairing choices about your product to choices about the experience you’re trying to deliver, right? So.

You know, managers are not idiots. They’re trying hard to do the right thing. And, um, you know, when we talk about kind of exploring these trade-offs and sort of using tools like small bake-offs to build confidence inside the team that they can go a different way, those are, those are repeatable tools that folks can use to try to understand the situation around them. That gives them, you know, something other than just cargo culting what everyone said to do. Um, and so that’s, you know, that’s, um, uh, an effective way to kind of help teams.

get there. And we just did the same thing in Edge. We just rebuilt almost all of our UI services, an increasing fraction of them, more than 50 now, the settings page, a lot of the dropdowns that you click on in the top level UI. A lot of stuff was built in React, and now it’s being built in Web Components because we did that same bake-off thing. We said, how fast can we go? And okay, we can go this fast because we don’t have to support the legacy. And so now we’re getting 40 to 50 percent wins across the board at P75.

And so that’s what we were hoping to see and that’s what we’re delivering and that kind of change is possible. But it’s enabled by finding that small screen, that one page, that one thing you can prototype on. Give the team a couple of days, a week, two, three to go rebuild it a couple of different times in a couple of different styles. Try something fully server side. Try something with a different client side framework.

Try wasm. don’t know. Just try everything. Like try all the things that could be reasonable and modern that fit, know, the user base you actually have. And then, you know, having, you know, when you set up the bake off, set up the criteria that you’re going to use and test things based on those criteria and then go with what works. Cause everything can work, but you just have to choose things that are going to put you in the mind of your product’s problems.

Kate Holterhoff (37:38)
so the management sounds like, is one avenue for trying to fix some of these, tremendous overwhelming issues, but another one might have something to do with, specifications and how we are, I don’t know, fostering the sort of web that we want to see in the future. And so I’m interested in what is happening, currently around web standards. So we got the… W3C and TC39, you’ve been involved with both. Is there some hope

Alex Russell (38:07)
So for full disclosure, I also used to be the tech lead for web standards on the Chrome team. So it was my responsibility, even if it wasn’t my fault in standards land for a long time. Someone better has a job now, thankfully for them, and someone that better has a job on the Edge team now for us, but I still have my hands in the till a little bit, in the dirt, not the till. Those are very different.

So, my mental model for this is that standards are downstream of implementer interest. So a lot of web developers kind of see standards as this event horizon.

Right? Sometimes the Hawking radiation escapes and you get a new spec in the world. Sometimes, you know, a bunch of enthusiasm pours into this thing. can’t see, you can’t peer over the edge, but then something happens. and, and you’re like, Hey, it happened. And it seems deeply inscrutable and no one really knows why. and I’ve been writing recently about kind of how those things actually happen in practice. and I think it, if you.

take the intent, maybe not the form, but the intent of the IETF’s motto of of rough consensus and running code, and you kind of play that forward.

That is how all new feature work on the web works. So standards are documents that are effectively dead by the time they become a standard. Because what is a standard? A standard is a group of companies who have a lot of intellectual property getting together and then saying, we disclaim any rights in the patents that are imputed to this document. That’s a standard. That’s what a standard is. That’s what standards development organizations do. They write.

IP clearinghouse checks, They receipts for IP to the ecosystem at large under specific licenses. And web standards have the property of those things being franned, not just ran. They’re free. They’re free to implement. Everyone can do it. You don’t have to pay anybody licensing costs for those patents. So.

That’s like the very legalistic version of like, is the standard and why does it matter? Well, because once something has that sort of imprimatur, it has no cost and it’s got a, you know, a big answer to you and what army. So if you go and you see the little guy, all the big guys who were in the room are going to go, excuse me, right? cause those are our patents and you should see our portfolio. And so that dynamic is what makes, standards an accelerant for adoption for designs.

But standards are not designs. So my mental model for all of this work is to try to separate the idea of what is a standard from how does design work happen. Because a lot of people are confused about this. A lot of people who spend a lot of time in standards work are confused about this, which turns into its own problem. Mostly in terms of disappointment for them. It’s actually quite sad when it happens. Because what you.

Want is an implementer, someone who has a big piece of software that can move, to be invested in changing their software such that they can solve a problem that someone who’s got the problem actually has.

And so how do we get there? Well, that’s usually through conversation. So the other important thing about standards bodies is that they enjoy protection from collusion, effectively charges of collusion in competition law. so standards bodies are one of the few places where competitors and customers can get together and be like, Hey, what should this thing look like next year? If I was designing a vacuum cleaner and I went to a vacuum cleaner,

conference and we talked about pieces of things and maybe interoperability and we didn’t do it inside of a standards body, we could be charged for colluding.

because maybe we would incrementally change the price of something, or we would maybe be pre-agreeing to change the quality properties of these things to remove competition from the market, right? That would be bad for the market. And standards are a specific carve-out in competition law that allows everyone to go, okay, we’re gonna give this away, more or less, and there is a societal benefit to that which overrides the concern about collusion. So.

a lot of folks take those properties together, like the, the accelerant property and the ability to have open discussion and say, well, we have to design things in standards bodies. And they’re right technically that you have to do the design work inside the, the IP umbrella of the standards body. But that doesn’t mean that the working group that sort of eventually writes the specification that gets the sign off and gets the receipt written for the IP is the right place to do the early design or any of the first 10 iterations of the early design. And so what you see are.

these kind of, we call them incubation processes, but they’re kind of like feeders, they’re farm teams, right? They’re like places where people can go and they can have conversations around what the problem actually is, because that’s often very fuzzy, and they can have conversations around what the potential scope of all the solutions looks like, and they can have conversations around, okay, how do I judge which one’s best? How do we try them out? How do we play with them? How do we iterate on them?

And those processes are things that have become much more explicit over the last 10 years in web standards land. We’ve worked hard to do that. The W3C probably has the best one outside of the IETF. It’s called the community group process. And we’ve got a big one called web incubation community group where you’ll see dozens of proposals and some of them fail. And that’s great. Sometimes people will like look at this and go, look at all those failures over there. And we’re like, that’s not a graveyard. That’s a set of ideas that we tried. And that’s how you make progress.

when you find out and if you found out that it wasn’t the thing to ship, you shouldn’t have shipped it in the first place. It’d be super expensive to take it back if you tried. So you don’t want to do that. You don’t get any mulligans on an old stable platforms. You just have extra barnacles on the hull. So you absolutely have to get it right. And that iterative process is necessary to do that. So one of the failure modes of standards bodies is that people will go into one of these big rooms.

They’ll go into like the CSS working group or TC39. They’d like, I’ve got a brilliant idea and you should do this tomorrow. And everyone will go, excuse me, that’s not how anything works. Worst case they’ll go, yes, we should. And then they’ll have a bunch of big fights because those rooms are also not designed for doing design. They’re designed for sort of dotting eyes, crossing T’s, doing integration work and making sure that things actually fit together rather than like that messy process of early iteration. So is there hope from standards? Yes, but it’s downstream of two things.

The first thing is developers believing that they can get their answers, get answers to problems. And so I would encourage developers who are interested in seeing the web be different, hopefully better, but at least different, to go and find these incubation groups. Go find the YCG, go find, you know, IETF, it’s the bar BoFs. WTBG doesn’t have one of these venues, but TC39 has the early stages process. And those are places where you can go and talk to other…

people who may have the same problem under the protection of that IP umbrella at low cost, low friction, low cost. You don’t, your company doesn’t have to start paying dues immediately. You can just sort of join as an individual and relative to, you know, your own employment agreements, you know, aside from that, you know, you, you should be able to, you know, participate in the design, especially if you’re an independent. And then you can find people who are interested in helping you solve those problems. We’re in an unfortunate position today with regards to the web ecosystem because

Overwhelmingly, just in terms of team size, most of those people tend to be Chromium contributors, right? Like there just are a lot more of us than there are Firefox engineer, Gecko engineers and web kittens. There’s just more blink people. So, you know, come find us. We’ll talk about it. And, you know, we have a pretty good process now for doing responsible iteration and evolution early on.

We got this whole system called origin trials where we can try things out early and then like test them out in the field and then find out whether or they actually work. You can try them on your real website and then we can still take them back. Right. We do. don’t get like burn in. Like we talk about like incidental standards where things get tried out and the early drafts sort of get burned in and no one really likes them and it’s hard to take them back. So we’ve got processes to even address those risks. And so that stuff is good and it works and we’ve gone a long, long way. The missing quantity now.

though, frankly, again, it’s iOS, right? Like, it’s no good if leaders can lead everywhere except the one operating system where it matters, right? Where you can’t take the thing that you built that uses the shiny new API and show it to your boss and be like, hey, it’s awesome. And then sort of spark a competition between browser vendors to go do that stuff. iOS is holding us back definitively because we don’t have any true browser competition on iOS. And so.

I would encourage web developers who are interested, and companies who want to make web products that are good to go discover OWA. And if you can, monetarily support them. They have been extraordinarily effective. I know that they need funds now for some of the hard work that they’re doing. So if you have time, enthusiasm, and or money to pitch in, they are fighting to make it possible for us to kind of have a web that is vibrant in the pockets of wealthy folks, not just everyone else. So, and that’ll change the game.

Kate Holterhoff (47:51)
OK, so I am going to start to wrap us up here, but I have one final question. I can’t, or you rather, can’t get out of this conversation with me without talking about AI, just the littlest bit. So I can imagine two questions to ask you about this. The first would be on the AI browser phenomena. mean, there’s a lot of them. There seems to be new ones all the time. Everyone’s acquiring them. So anyways, that just seems like.

Alex Russell (47:53)
Mm-hmm.

Okay.

Kate Holterhoff (48:16)
an interesting touch point that maybe you have thoughts on. The other one would be on vibe coding and how every time you vibe code something, it insists on using a JavaScript framework rather than just like vanilla JavaScript and like CSS. It forces you to use like Tailwind and very frustrating for me. Anyways, do either of those pique your interest? I’d be curious about your thoughts.

Alex Russell (48:38)
Sure, I mean, my mental model for this stuff is very similar to kind of, I don’t know if you know who Casey Muratori is but like I kind of take his perspective on a lot of these things specifically around the AI thing where there was a true technical achievement, a brilliant accomplishment at the bottom of this, which is that for many years, like search engines were just short word distance.

vector summarizers, right? You could sort of compute vector distance over a small number of words because you made the words much, the context much larger, you couldn’t do it, right? We just didn’t have the hardware to do it. And now with, you know, graded descent learning and transformers and the techniques that have come along, we can get very close to both understanding English and other languages, you know, without having to kind of special case all the stuff, which is incredible, right? This is just

This is a true revolution. And on the backside of that, you know, we get systems that can, can have those questions posed and do better than word, you know, small word distances. that’s wonderful. That’s, that’s new. It’s brilliant. those systems also are token predictors, right? Which, without more context than we can give them today and without,

you know, better guardrails than we have for them today are just tiny little liars. and that shows up everywhere, right? So the S can just sort of take the same model, take a smaller quantization of it, or take the same model and take like a smaller version of it and it’ll lie to you more often. Right. So that’s like, that’s just the thing. That’s just the thing that happens. and you can dress it up in, whatever terms you want, hallucination, you know, it’s got a lot of terms for it, but this is now the fundamental challenge of these systems is trying to get them to be.

A, more transparent. I’m looking for someone to kind of build a model that’s got zero temperature, right? Like that. I’m fascinated by someone, you know, spending the compute to do that. But also, you know, seeing these things put into places where we’re going to have to reckon with the idea that reading anything through this system now becomes that system considering it and then doing something about it, potentially.

autonomously. So the threat model for this, you my friend Simon Willison sort of has this, he calls it the lethal trifecta. And I was, I was fascinated to see it because it looks exactly like what we call the rule of two in Chromium. Like our idea is you shouldn’t have, there’s like three properties and you should never have more than two of them. Untrusted inputs, memory unsafe languages, and

Extent capabilities, right? So like you should never have more than two of those things in the same process. And so we put sandboxing around some of them so that we, you know, and it makes the communication channel very narrow. This is now a challenge for AI because the threat model is the copy equals execute. Copy means do. So anytime you hand anything to an LLM, you’re basically telling it, Hey, you could do anything you’re allowed to do. And this is a sort of a product level question that we don’t have an answer to. So that’s a big problem.

In terms of the ecosystem, I think we are only beginning to understand the way in which some, but not all of these firms have abused the kind of, you know, want to call it gentlemen’s agreement, whatever you want to call it, that kind of made the web work up till now. And that seems like a paired challenge. I don’t, I don’t have more thoughts about the AI browser thing other than that. Like they’re all kind of, you know, small products at the moment.

They’re not, I mean, they’re not like, they are not the primary way that people do anything yet. We can see it coming. And so there’s something to do there and it’ll stress all of those pieces. And so that will stress the ecosystem side. It will stress, you know, our questions around trustworthiness and truth. And, you know, I would like to see a lot of these companies just as a personal point, I would like to see companies stand up for truth.

and believe that it’s important that we be able to distinguish and differentiate the truth from the false. Edge is one browser. We’re not that large yet. But I’m concerned that we’re not focused enough on the question of truth.

Kate Holterhoff (53:13)
Well, that helped me tremendously. Okay, so before we do wrap up, how can folks continue to follow your thoughts on the web, your philosophical musings, your political thoughts, you know, what is your, yeah.

Alex Russell (53:26)
I don’t recommend it. but, but if you’re so inclined after the warning, my website is infrequently.org. I promise it’ll go back to being more infrequent than it’s been over the last month. It’s just been, it’s just been a lot of gaslighting from Apple, that needed some response. there was an RSS feed there. also, I’m on Mastodon and Bluesky and you can find me, from links there. So thanks.

Kate Holterhoff (53:48)
Fantastic.

All right. Well, I’ve really enjoyed speaking with you today, Alex.

Alex Russell (53:52)
And you. thank you.

Kate Holterhoff (53:53)
Again, my name is Kate Holterhoff, senior analyst at RedMonk. If you enjoyed this conversation, please like, subscribe, and review the MonkCast on your podcast platform of choice. If you’re watching us on RedMonk’s YouTube channel, please like, subscribe, and engage with us in the comments.

The conversation also explores how hardware limits and efficiency are reshaping cloud-native design, from power and cooling innovations seen at Supercomputing to squeezing more capacity into every data center. Holterhoff shares how Kubernetes, AI conformance efforts and projects like OpenTelemetry (OTel) are coming together to support AI agents and complex workflows with strong community backing and observability. Looking ahead, Macleod points to a future of millions of accelerators on Kubernetes clusters and better “graceful degradation” as systems hit scale ceilings, while Singh envisions true platform agents that can auto-size and reshape pods so developers simply deploy and let the platform optimize.

This RedMonk video is sponsored by Oracle.

Links

• LinkedIn: Sharat (Shar) Chander
• dev.java
• learn.java
• oracle.com/javaone

Transcript

Kate Holterhoff (00:12)
Hello and welcome to this RedMonk conversation. My name is Kate Holterhoff, Senior Analyst at RedMonk, and with me today is Sharat Chander, Senior Director of Java Product Management and Developer Engagement at Oracle. Sharat, thanks so much for joining me on the MonkCast.

Sharat (Shar) Chander (00:23)
you

I appreciate being here. I love the MonkCast. It’s one of the things that I regularly listen to. So you already have a listener in front of you.

Kate Holterhoff (00:34)
That’s amazing. I love to hear that because sometimes it does feel like I’m speaking into the void. But it’s the kind of conversations that I like to listen to. So I suspect I’m not the only one.

Sharat (Shar) Chander (00:44)
You betcha.

Kate Holterhoff (00:44)
So today we’re going to be talking about a very exciting anniversary, and that is Java at 30. And so I can’t think of a better person to talk to about this than Shar. So let’s begin with your background here. Talk to me about your long tenure around the Java community and why it’s so special. Yeah.

Sharat (Shar) Chander (01:06)
It’s amazing that we were even saying 30 years of Java. You don’t think you sometimes reach these milestones until it happens. And here we are in 2025. It’s an important and significant milestone. I oftentimes pinch myself to think that we’ve reached this plateau, but we still have many more years to go and to keep rising. And it’s

It’s 30 years young, as I call it. It’s not 30 years old. And we can get into some of that. I’m standing on the shoulders of giants. I may not have been there at the onset back in 1995. My journey with Java didn’t start until I actually formally accepted to join Sun. And I still remember the date, was December 8th of 2000. So

Sun was already many years into investing in Java and just building it out as something that would touch every part of our globe and our industry. And to see how much it had already come in those four and a half years before I came on board was sort of like, it’s like looking at a car price, it’s sticker shock, but in a good way because of what it was already capable of doing.

We talk about Java with so many prominent names. James Gosling, the father of Java, folks that worked on the original green team like Ed Frank and Chris Rath, and I think it’s Patrick Naughton and Mike Sheridan. Those were some of the members of the original green team that made what we call Java Java. But there’s so many other people that history needs to know.

I’ve always said that as great as Java has been as a technology in terms of a programming language, a platform, something that solves so many interesting use cases across different verticals, it’s a very people first technology. I’ve often said Java is a people first, technology second initiative. Half of it is technology, of course, and the other half is the people not just building it, but the community behind it.

I always try to reflect on the people that have brought us here and history needs to remember these names. mean, folks like Diana Ystoffer and Jeffrey Aboud and Bao Fan and Jennifer Arale, they built a lot of the programs that gave Java visibility or Wendy Yamaguma and Alvino O’Neill, who literally were the heartbeat of what was one of the most famous tech conferences on the planet, which was JavaOne, which you’ve now

for a few years now brought back and hope to bring back that same level of vibrancy. But other people like Matt Thompson and Reggie Hutcherson and Ashley McReynolds, these people built out foundational programs that we often take to market and talk about with developers that are basically the heartbeat and ethos of what Java represents, which is people.

Kate Holterhoff (04:13)
Yeah, the fact that you are able to name all of these individuals who’ve been so pivotal to the Java community. I mean, I think that really demonstrates that, yeah, your mantra is quite accurate, right? People first, technology second. That, yeah.

Sharat (Shar) Chander (04:28)
Yeah. And trust me,

they were, they are, they’re still my peers. Many of them were my leaders. So I imprinted on, their teachings, on their perspectives, which was, you know, fundamentally, when we look at Java, why has it become so ubiquitous? Why has it touched every part of our globe? Why do we say that it runs half the world’s GDP? it’s, it’s because we’ve made a connection with end users where they find purpose.

in the actual technology that they’re using. And I personally think that’s a very strong signal to send to enterprises because what ultimately we have done, or many of my peers in the past have done, which I’m hoping I can continue to do in their spirit moving forward and hopefully the next generation who sort of takes the baton up when I sort of like, hopefully not age out, but like move on, is this concept of virtuous cycle.

Business leaders want business value. We all get that. So fundamentally, they want a technology that’s

them solve a variety of use cases or challenges. And Java has done that brilliantly because of just the thoughtful innovation that we’ve put into it over the last 30 years. But they also want to make sure that they have a constant pool of extremely skilled developers that can be productive in the types of tasks and solutions that you’re trying to build.

That’s why I say it’s so important to focus on people because if you focus on people, you’re trying to understand what are their pain points? What are they scared of? What are they trying to achieve? How can they better learn? How can we accelerate the learning? How can we have more shared common knowledge? Because there’s only so many screencasts you can watch. There’s only so many books you can read. Oftentimes, some of the best learning comes from just being immersed with your peers. And we here on the Java team,

Historically, even now, I want to ensure we have programs where we can galvanize developers, just online, but in person to share those best practices, share those use cases, share those experiences because we learn better together. And you gave a really long title description of what my role is. It might be the formal title, but I kind of look at my role in a very simple way. I’m a sherpa.

What’s the role of a Sherpa? Sherpa’s job or task is to try and help someone reach their summit. We’re not there to set the path. We’re there to give them guidance and advice as far as what path to take. And then ultimately it’s up to that person to decide what to do with that advice that they get. And if they reach their summit, that’s success for me. So if developers become more empowered in what they do, it creates longevity in terms of future proofing their job.

It helps their enterprise ensuring that they know that they’ve hired someone that can be there for the long term to solve the problems of today, but also the problems of tomorrow. And then more importantly, this sort of shared experience around Java has been participation. Like we care about developers. We want their input. We want their feedback. We want to know what we’re doing right. That’s great. Let’s celebrate every once in a while. Like we are here with 30 years of Java.

But we also know what we what aren’t we doing? What should we be doing? And you know, we do so much of our development transparently. I think a lot of of the industry still doesn’t realize this after after so many years that at the in the OpenJDK community, that’s where all of the engineering around Java is done in the open. mean, developers can come and see what decisions are being made, what decisions aren’t being made, when the code check ins are happening, what features are being

invested in, what’s the horizon in terms of those investments, how features have evolved. And when you, I get it, in your day job, you have so much you can do and you have to put blinders on and just stay focused on the task at hand. But if you just raise your eyes up, just even for a microsecond to see how Java is continuing to evolve, you become more empowered in the job that you do. And, know,

The entire organization at RedMonk has done such a great job to highlight and demonstrate the influencing role developers have. I still have Stephen O’Grady’s book, The Kingmakers. It’s something I go back to and look at very often. And he’ll probably be like, you do? I’m like, I do because developers play such an instrumental role in the direction that companies take and…

When you allow developers to participate in the very technology that they’re using, they find more of a purpose in it. And that’s what we want is people to have a longevity with the technology that we continue to evolve. So we are ensuring that their career stays future-proofed, that their business understands that they’re providing value. And at the end of the day, Java is going to solve a million different business problems that are inherent.

across the industry.

Kate Holterhoff (09:44)
Well, Steve will be happy to hear that, I’m sure.

Sharat (Shar) Chander (09:47)
It’s a little ragged now. mean, I’ve booked or I’ve talked to many of the readers and I’ve scribbled so many notes I might have to buy another one. Yeah.

Kate Holterhoff (09:50)
Yeah, dog-eared.

How funny. No, that’s awesome.

Okay, well, you know, and I’m always interested in the personal stories of how folks get involved in that Java community. And so you’ve spoken a little bit about an early Java users group in New York. I would love to hear more about that experience, including some of the, like, frustrations that led you there.

Sharat (Shar) Chander (10:19)
Yeah. So, you know, my background is in Telco back in the early and mid nineties, just like every other enterprise on the planet. There were so many different programming languages trying to solve so many things and you would have different development teams all utilizing something quite different, targeting different hardware architectures. And so just from that nature, there was a, just from a cost perspective, there was a lot of overhead to try and keep all of those applications.

relevant, investing in the teams around it, and learning about Java in terms of its ubiquity to be able to solve many of those challenges across use cases, but also target a variety of hardware environments really simplified enterprise issues that we had faced from a technical perspective. And so I happened to be working in New York at the time.

Some of you might know Verizon, happened to be there. JavaOne, the inaugural one had happened in 1996. I happened to be lucky enough to go and attend it and it wasn’t like your typical conference where you sit in sessions and hear a lot of keynotes and you walk away with a bunch of notes. It was literally an experience of how you could meet other people facing the same issues you were.

and talk about it and discuss it and share it and the people behind it cared and wanted to hear from you. That really resonated with me. And, you know, coming back to New York, I’m like, well, so where are the other like-minded people? And it just so happened that in 97, a gentleman named Frank Greco, who history should absolutely remember his name, started the New York Java Special Interest Group or New York

Java SIG for short, happened to see a post on a message board that there was this Java SIG going on. I happened to go to one of the meetings and lo and behold, it wasn’t just me. There were dozens of people from different industries, finance, telco, insurance, retail, coming and showing up, just trying to understand how do we better learn about this relatively new language called Java? And that

that user group actually manifested into a formal program at Sun, exploded into the Java user group program. we now, last count we did, there’s close to 340 Java user groups on every continent except Antarctica. I think that’s a stretch goal. Perhaps almost every country, every major city within probably a train ride or driving distance for most developers.

And these organizations are all run by volunteers just like Frank, who find purpose in Java and want to create an experience where they can accelerate their learning through peer knowledge transfer. And so here on the Java Developer Relations team, we try to focus in on user groups and try to give them the knowledge in terms of what’s happening in terms of innovation.

How do we empower them with information? Like say a new feature is in the works that’s already visible in the OpenJDK project because we’re transparent. We want to make sure that’s populated and federated out to user group leaders. When is there a specific initiative coming up? How do we connect people and different user group leaders or members with others so they feel part of something larger than themselves? so, most of my day is spent literally doing relationship management, but…

when you nurture relationships, you’re creating something that’s durable and you build trust. so when, say there, say if there’s ever an issue, which rarely there is, but like if there’s a difference of opinion in terms of specific direction, we can have an open and honest dialogue with our ecosystem as far as these are the reasons why we did what we did. And, you know, you probably see it all too often. Developers love to say, why can’t you just do X, Y, and Z?

Oftentimes that question comes from a position of not understanding what it takes to do X, Y, and Z, or what impact it will have with existing investments. And so we love the input and feedback, but we want to be honest and transparent in terms of what we’re doing and why we’re doing it, but also take that input inside. And so we always are asking developers, please come and join the OpenJDK community. Please come and join some of our projects. Please read the mailing lists.

or become an author or a committer. And I know those are lofty things and that takes time, but there are other ways developers can participate. Go give a talk at a conference, record a podcast. Perhaps they want to stand up a user group. And we’ve now seen a couple of new user groups just sprout up this calendar year. The newest one was in Gujarat in India, which…

I’m kind of like, yay, because I got a lot of relatives back home. And I’m like, wow. To see 14 user groups in my parents’ home country is kind of like sense of pride. But in Latin America, there’s, I think, 15 user groups in Latin America, at least a dozen or more so in North America. In Japan alone, there’s three. mean, they’re spread out. Germany, I think there’s 20. I think Germany has the most number of Java user groups on the planet, just from a number standpoint.

To see so many people finding purpose and wanting to share, have this collaborative spirit is a reflection of that 1997 experience that I had with Frank. So Frank, thank you for putting me on this path.

Kate Holterhoff (16:06)
That is such an amazing story, and I appreciate the statistics that you can bring here, talking about the newest ones and how many there are and, the geographic locations. my goodness. Yeah, I mean, I’m obviously the most familiar with the AJUG group because it is in my hometown, and that’s right. And of course, that one hosts the Devnexus.

Sharat (Shar) Chander (16:22)
H-R-G-L-A-T-L.

Kate Holterhoff (16:27)
conference, which I also enjoy and have spoken at. and where I typically see you, Shar.

Sharat (Shar) Chander (16:32)
Yeah, well, I mean, you bring up a good point. know, JavaOne really is a a I know we’ve we’ve probably heard this term overused in our industry for these last 10 years, but by developers for developers. And that was literally the mantra of JavaOne. And that didn’t come from us. So there was a gentleman named John Yeary who ran the Greenville Java user group in South Carolina.

And I happened to be in Atlanta at an Atlanta Java user group and John Yeary made the drive down and he’s like, you know, JavaOne has always been a by developer for developer conference. And we’re like, what a great tagline. I mean, should have, we should have trademarked it at that point back in 2006 or seven, but he’s right. So JavaOne literally is meant to be an opportunity for developers to not just meet the experts from the Java team.

Kate Holterhoff (17:15)
you

Sharat (Shar) Chander (17:28)
and learn from them, but meet all of the global luminaries and peers that come and have this moment of galvanizing together. And what did that then do? Java One that helped launch a dozen other Java conferences around the world. You talked about Devnexus in Atlanta. There’s JavaZone in Oslo, Norway. There’s Jfokus in Stockholm, Sweden. There is the great Indian Developer Conference in

Bangalore, India, we have things like J-Fall run by the Netherlands Java user group as another Java conference. And there’s GeeCON and Devoxx and I mean, the list goes on and on. And that’s just a reflection of just how popular Java is. we actually think it’s very complimentary. So I always like to celebrate when I see other conferences.

that are focusing on Java because it’s just an extension of what we’re doing. And what a great way to bring something at a local level to these different geographic locations.

Kate Holterhoff (18:33)
All right, so I do want to talk a little bit about some of the history and how things have developed because you have this phenomenal insight. You have an insider perspective and you’ve just been there the whole time. I know, I know, you’re laughing.

Sharat (Shar) Chander (18:46)
Yeah.

I’m like, so it’s really, I can’t give enough credit or pay homage to the leaders at Sun. I mentioned a few of their names and that all came from James himself. I typically call him the non-executive executive. All too often sometimes that glass ceiling seems impenetrable.

to feel as though like you have a place and a say and a voice in the things that you are working on. And James’s philosophy was literally an open door approach. know, everyone is able to say what they want to say. You had a format, you know, a place and a location to do it appropriately. And that spirit and ethos has lived on in Java. And, you know,

My first intersection with James was back at Sun, we used to do these things called Donut Wednesdays, know, aggregate all the employees in the office, have some coffee and some donuts, you know, break the ice, have a little bit of camaraderie. And here comes James walking down the hallway and, you know, you know who your executives are. You never think you’re ever gonna meet them. And to see him just come up, pour a cup of coffee, start chatting.

and I’m sitting there in my head going, I’m allowed to talk to you? And it was just so casual how he approached it that it taught me that there’s no one person or one voice behind anything. It’s everyone rallying together as a team. And some people might get the accolades appropriately so, and some people might become the figureheads. That’s great. But we all play a role. And that’s why I always try to reflect and say thank you to those.

that have done so much or who are doing things because this is a team effort. we’ve had our sort of funny situations, like JavaOne, I remember there was an experience where we were showing off Java on a BMW and we wanted to show it on stage and it was showing how the infotainment was running Java behind the scenes.

And the car carrier that had this car showed up to Moscone Center in San Francisco on a Sunday before doing the build out for Java One. And there was no one there to sign for the car.

There was no one there to sign for the car. I happened to be in the bellows of JavaOne with the tech team and I thought it would be okay to drive off the carrier into Moscone. And apparently that was a no-no, but I thought I was being productive and proactive, but let’s just say that didn’t go over well in the short term, but the long-term release, we got the car inside and inside, but you know.

Kate Holterhoff (21:32)
You

Sharat (Shar) Chander (21:42)
You look at those moments and going, hey, listen, I’m trying to feel empowered. Maybe I overstep my bounds. I was just trying to do what was right. But I learned from it. You learn where are the guardrails. You learn where the guardrails are.

Kate Holterhoff (21:56)
Yeah. Sometimes

it’s better to ask for forgiveness, you know? This is…

Sharat (Shar) Chander (22:01)
Yeah, mean,

I was, or, you know, one year with a Java release for the Tiger release, we had a white baby tiger on stage on the keynote stage. And I’m sitting there going, there’s a tiger on stage at a tech conference. What is going on here? Like, it’s sort of like an out of body experience, but like,

Kate Holterhoff (22:22)
You

Sharat (Shar) Chander (22:26)
So many people have attended JavaOne. They probably have similar memories or experiences or at other Java conferences and something unique happening. I try to collect them. I hoard memories, as I say, and I try to hoard as many of those memories as I can because it tells a story.

Kate Holterhoff (22:43)
Yeah, that’s really special. OK, so let’s talk about some noteworthy releases, because we all know some releases are more interesting than others. Java 8. I want to talk about Java 8 here. So before the release, I don’t think I’m going out on too much of limb here to say that Java faced a sort of difficult period. There was a perceived lack of innovation and some other troubles. Describe to me what that period was like.

Sharat (Shar) Chander (22:56)
Yeah.

Yeah, you know, I’m glad you brought it up and for full transparency, there’s various opinions on this and you know, I’ve always been the glass is half full. Let’s face it, know, Sun did face some business challenges. That’s no secret. But for those of us in the ranks, we can’t control business challenges. We can only do what we can to keep, you know, progress going.

You know, for Java there, there, there, there was still innovation happening. We also had to deal with, you know, the acquisition between Sun and Oracle. I think we landed in the right place for me to come to a company that has such stalwart business practices and principles was probably the right place to be. If we were going to go anywhere. Cause I think it helped put Java even in a more focused direction. And that’s why for me, you know, Java 8 was.

An important release that demonstrated Oracle’s seriousness and commitment to Java because many developers, anecdotally speaking, felt like Java may have been languishing between 7 and 8. And there’s some technical debt. You got to pay down. You got to right the ship. You have to create a sort new North Star. And the North Star for Java is basically following the principles of the past, which is focusing on

performance, security, and stability. And then, by the way, ensuring that the platform continues to evolve with net new innovation. And when you do incremental innovation, which we saw with our six-month release cadence, which we started with just after Java 9 with Java 10 is you’re reducing the friction and burden of adoption. And so when you have incremental enhancements,

it’s easier to approach that instead of like hundreds of changes, which can sometimes just create a deer in the headlight look and might take multiple years to adopt. And so I think fundamentally, you know, we righted the ship with 8, 9 put us on a good direction. The six months release cadence starting with 10 and with various projects like Amber, where we started to really focus in on reducing the ceremony and verbosity of the language, make it more readable, more writable, more maintainable.

We look at projects like ZGC, which was all about reducing latency time when it comes to garbage collection so your applications can run at scale. We’re entering this new world of AI, which I hate even saying new. AI has been there for a long time, and Java has been solving a lot of the underlying problems and challenges that AI faces. Matters like working with native libraries. We’re doing that in Project Panama.

How do we make Java more optimized for data? That’s Project Valhalla. These investments have been happening not just because of AI, but these are fundamental issues across a variety of use cases and programming types. It just so happens it works very nicely and addresses AI very well. just a few weeks ago, there was a very important developer conference in…

in Belgium in Antwerp that’s been happening yearly called Devoxx. And Paul Sandoz, of the most gifted people I know on the Java team here, did a talk or a keynote on Java and AI. And he said, Java is everywhere where AI needs to be. And I thought that was very profound because Java truly is everywhere. You have enterprises with applications that they want to apply AI solutions to. like,

How do you use, you know, accelerate code generation with code generation tools? How do you apply LLM models? How do you take advantage of some of these emerging libraries or frameworks for that, for example, like LangChain4j or even Embabel, which is one of the new ones, which was talked about at Devoxx by none other than Rod Johnson, who was the creator of Spring. How does Java play within all of that? Those are areas of investment we’ve already been doing before AI became sort of like

lingua franca in terms of the popularity of topics that we now have debates about. So I am very confident in terms of Java’s next 30 years. It’s all built on these past 30 years, which is thoughtful innovation. It’s not about how fast you bring something to market because that mantra of move fast and break things in principle, yeah, I get what it means.

For Java, I oftentimes say we want to move thoughtfully and build things. So introduce features thoughtfully so developers and enterprises can build things that are long lasting and bring enterprise value. And so that’s what we do. We have sort of last leader advantage where we can see and evaluate the market and see what are the things we need to enhance and evolve around the language and the platform and ensuring investments of the past then don’t break because…

Enterprises want value from things that have been running in production for the last 10 years or beyond. And so that’s an important sort of like balancing act we have to play is how much newness do we bring there versus how much stableness do we maintain? Oftentimes it’s called the tip and tail model, know, innovation on the tip and stability on the tail. And we try to keep that as much in balance as we can.

Kate Holterhoff (28:43)
Okay, so you have convinced me that the naysayers, the folks who are bringing up the old chestnut that Java’s dead, they’re clearly misguided here. This is obviously just not the case.

Sharat (Shar) Chander (28:54)
I love whenever I read that headline. It’s not as popular as it used to be. I remember in 2010 that the entire industry and every tech journalist had written off Java and said Java is dead. And then in 2011, the same headline showed up again, Java is dead. I’m like, how can it die twice?

Kate Holterhoff (28:58)
You

Sharat (Shar) Chander (29:18)
And then, you know, started sort of waning by 2012, 2013, that sort of headline started diminished because the best way can put it is the way you change perception is by using that mantra of cutting wood and carrying water. Keep doing what you’re doing, doing it consistently, doing it transparently, doing it authentically will then prove

that you are true to what you’re trying to accomplish, especially around Java. And I think we’ve turned the tide a few years ago. Every once in a while you might jump on to Reddit and there might be a crazy sort of, you know, challenging confrontation, meant to be confrontational, that Java is dead. like, I try to view it as everyone has the right to express their opinion, but for sure Java is alive and vibrant and.

With an ecosystem of over 10 million developers, some say it’s 14 million developers. I’m sorry, but when you’re eight digits, that’s huge. When you talk about tens of thousands of companies that are running applications across every vertical at scale every day, utilizing Java, that’s important. The fact that we have so many user groups with so many members, think close to, as I said, 340 user groups, that’s almost 700,000 developers that are part of that program.

The heartbeat of Java is alive. I love this past JavaOne that we just did in 2025 where we had Netflix on stage. They literally demonstrated how important Java is to run Netflix at scale with so many instances and the pace of streams that are happening. The fact that Java is enabling that sends a signal. And I think it’s important sometimes to…

to give an anchor a name so it becomes more real. Because listen, everyday consumers don’t see Java, but it powers our banking system and our trading systems. It powers flight control systems. It powers e-commerce systems. It powers social media platforms. Things like Netflix, as I mentioned, things like Uber, things like LinkedIn. These are all important organizations that are using Java.

to help them solve their business challenges. And I love back in the mid 2000s where we even talked about how Java helped land the Mars Rover. That’s an important sort of like milestone experience where not only do consumers understand the importance of Java, but developers themselves see it and feel it and understand Java can do some really big things.

Kate Holterhoff (31:56)
Okay, you have 100 % convinced me. We talked about the community. We’ve talked about the technical innovations. What I’m interested in hearing now is, what is the lasting legacy going to be? You’ve mentioned AI briefly. You’ve mentioned Valhalla. You know, we can keep this as hand-wavy as you choose, but I guess I’m just interested in the long view. I think you did say, you know, Java for the next 30 years. Yeah, what is that going to look like?

Sharat (Shar) Chander (32:21)
Yeah, so from a from a technical perspective, from just from the broadness of it, it’s it’s adapt and evolve. And that’s what Java has done so brilliantly for 30 years is to adapt and evolve to a ever accelerating, changing technology landscape. And we do that thoughtfully. And I think that’s really going to be the next 30 years of Java is adapting and evolving thoughtfully. From the other side of the coin, it’s

people first. I don’t think that’s ever going to go away when you when you create such an important thread into the ethos of something like Java that requires and needs the input and participation of your community. It’s very it will be very challenging to erode that. Now that’s for us to decide. We can either choose to keep investing in it, which I we I believe we are and We will continue to do that.

But that’s gotta be an inherent investment area for the future is continue to invest in the ecosystem, in the community, give people purpose, create that virtuous cycle so businesses know there will be very skilled Java developers to help them solve the business problems of today, but the business problems of tomorrow.

Kate Holterhoff (33:36)
I think that’s a great place for us to wrap up. for folks who would like to hear more about Java at 30, perhaps your own social media musings, Shar, where would you direct them?

Sharat (Shar) Chander (33:48)
You know, the best place to go is to just turn their attention to dev.java. That is our destination for all things Java in terms of learning, in terms of community programs. I would like to give a plug to learn.java. So part of inspiring your existing ecosystem is to also bring in the next generation. And we understand there’s a whole new generation of developers that are out there that can absolutely…

find value in Java. So we’ve now also launched Learn.Java. So plug out to my peers, Heather Stephens and Crystal Sheldon, who are leading that endeavor. So we also want the developers who are just cutting their teeth on Java to come and learn. So please report your attention there. And then, of course, on LinkedIn, we have an official presence of Java on LinkedIn. Please follow us there. But Kate, 10 million developers.

Everyone’s going to have a different modality. And so we have dev.java, we have learn.java, we have Java on X, we have Java on LinkedIn, we have Java on YouTube. We have Java in places that try to hit as many of popular ways that developers try to learn and stay engaged on the technology.

Kate Holterhoff (35:01)
I have noticed that as well. It is sprawling, it is robust. You can do it in person at the JUGs or you can find it all online. So amazing. Okay, well, that certainly helped me to get a better understanding of this.

Sharat (Shar) Chander (35:09)
Do it online? Yep.

Kate Holterhoff (35:18)
I have really enjoyed speaking with you today. Again, my name is Kate Holterhoff, Senior Analyst at RedMonk. If you enjoyed this conversation, please like, subscribe, and review the MonkCast on your podcast platform of choice. If you’re watching us on RedMonk’s YouTube channel, please like, subscribe, and engage with us in the comments.

Sharat (Shar) Chander (35:32)
Thank you, Kate.

This RedMonk conversation is sponsored by Arcjet.

Links

• LinkedIn: David Mytton
• X:@davidmytton
• Bluesky:@davidmytton.social

Transcript

James Governor (00:12)
Hey, this is James Governor from RedMonk and I’m here with David Mytton of Arcjet and we’re here to talk today about one interesting one. Not quite sure what to call it, but we’re gonna call it security as a feature for now. I think the main thing, David, you’ve got a real focus on developer experience. So tell me a bit about what’s the problem you’re trying to solve? Why is developer experience so important? and whether we’re ever gonna finally get developers to care a bit about security and building it into their applications.

David Mytton (00:49)
Developers have been used to some pretty amazing innovations over the last five or six years where developer experience has really been the focus. We’ve got new platforms being released, cloud platforms, Render, Railway, Vercel, Netlify, Fly, that really do focus on the developer experience when deploying code. And the IDE is getting better every day as well, whether it’s VS Code is the leader for a while and then it got forked and we’ve got all the really interesting AI IDEs.

And just the focus on developer experience when it comes to building and building your application and writing your code is something that developers are used to. When it comes to security,

That’s just another story and we just haven’t seen the innovation that suddenly when you go into production, developers have to use a completely separate system. It’s often point and click or it’s entirely disconnected from their code and it just gives developers a lot of headaches and the result of that is the reputation that they just don’t care about security because it’s completely outside of their workflow. It’s just not in how they’re used to doing things. And this idea of getting developers to think of security as a feature, think actually…

aligns with how developers think about everything. They’re fixing bugs or they’re building features for users, what the customer is asking for. And security should just be another feature that comes in, whether it’s asked for by a compliance team or a big customer or because it’s something you need to make sure you’re handling certain things in your application, like just filtering, validating user input. It should be thought of just like we’re integrating a database. We’re thinking about performance. We’re setting up our design. These are all just features

that developers have to think about today and security should just be another one. Unfortunately, that’s not really how most developers do it. The incentives are very misaligned and it’s often forced on them by security teams. And it feels like just toil that they have to deal with. Has that been your experience? You’ve been talking to developers? It’s just a lot of hassle.

James Governor (02:50)
Yeah, think that’s

exactly right. mean, you know, let’s face it, we’ve been through this. I mean, we’ve been talking about shifting security left for quite a while now. But it’s sort of interesting, you know, even though the sort of the players that we would associate with that idea and, know, Snyk is obviously a good example of that. It ended up being to a point, a different workflow, having to go sort of manually fill in a bunch of information.

alongside the task that you were doing. And I think, you know, we, if we’re about developer experience, we’re thinking about developer flow and yeah, so developers have definitely been faced with being given more responsibility for security or being told they had more responsibility for it. Whilst at the same time, yeah, the tool chains and the processes didn’t really map to that. And I mean, honestly,

David, and we’ve still got organizations where you’ve got a separate security organization that really is not very well integrated with what the engineering team is doing. So I think that the tools to your point have really, yes, when I talk to developers, they definitely find that security is something that’s a hassle for them.

That’s one of the reasons they’re constantly, I think, rebuilding something. And when it’s done well, you do see possibly explosive adoption effects. I if you look at Supabase making auth something that’s a little bit easier, that really drives adoption for them. So I guess the answer is yeah. When I talk to developers, the tool chains are just not… They’re just not buttery. They’re not…

They’re not the sorts of experiences, I think, as you’ve said, that they’ve been led to expect in some of the areas of their daily work.

David Mytton (04:50)
Supabase is a great example because it’s basically made the database a feature. You don’t really think about Postgres being the underlying database until you start getting into…

James Governor (04:58)
Not at all.

David Mytton (04:59)
all the advanced details later on. And Supabase has built this amazing product on top of Supabase and developers, they just suddenly realize later, oh, they’re using a database and they’re using Postgres. And that’s what we’re trying to do with security as well, is that you’re trying to solve a problem, you reach for a particular tool that solves that specific problem, but then when you look at it in detail, what you’ve actually done is you bought a security product. And.

They are very different disciplines. And if we go back to the database example, in the olden days, you’d have DBAs who would manage the database for you, deal with the schema, the performance, the upgrades. And at the extreme scales, you still do have specialists like that, but that’s not the majority of applications these days. Startups, medium-sized organizations, they’re not hiring DBAs. And security is still in that early era where it’s a completely different discipline. Often, security specialists are not

engineers or at least they’re not product engineers, they’re not usually writing code in the same way that an engineer is. And that’s just how it used to be with databases as well. And this is where I think security should go and hopefully is going, is that developers get to integrate it as part of their code just as another feature rather than it being on a completely different layer of the stack looking at different signals that have no connection to the application. Because ultimately security is all about context, it’s understanding what the business logic

James Governor (05:55)
Mm-hmm.

David Mytton (06:23)
is trying to achieve and you can’t do that if the system is entirely separate.

James Governor (06:28)
Okay, tell me what sorts of use cases that you’re actually, what are we talking about in terms of these features? Like what are the things that you’re making easier to integrate? What are the security sort of those annoying problems or annoying things to implement that you’re doing for the developer? What do you do?

David Mytton (06:49)
So we think about it as pre-prod and once you go into production. so there’s actually been some interesting innovations on the preprod side of things. As you’re writing code, this is able to tie into AI, is able to do static analysis. So interesting tools like Semgrep, for instance, or Socket for dependency analysis, that are able to give you useful insights before your code goes into production. Those are much more connected into your code. They run locally, they run in CI, and developers are able to use them. And there’s been some really interesting focus on DX there. Once we go into production, which is where I’m focused at,

it’s another story because often that is added later. It’s not connected to your application in any way and you usually can’t run it locally. It’s only deployed in production and these kind of network level tools, they analyze packets, they don’t understand users, they don’t understand sessions, they don’t understand authentication.

And what we’re trying to do when it comes to these production security issues is to build a product that can tackle those in the same way that developers are used to. So specific examples of bot detection, that’s very popular right now, particularly with AI kind of muddying the water around what is a bot and which one is a good bot versus a bad bot. But even traditional problems like sign up form spam, form spam, you just have a simple lead form on your website, then I’m pretty sure everybody has seen

the spam submissions that get sent through those. E-commerce is another example where we’ve been doing a lot of work where fraud has basically existed since the beginning of the internet and the beginning of e-commerce. And going through a sensitive checkout flow, that really needs to be connected to your business logic because the worst thing you can do is just as someone’s about to press checkout is just to block them on the network and serve them a generic 403 error.

In that case, you want to know that you’re in this flow. Maybe you know that the user’s logged in. Maybe they’re a repeat customer. Maybe they’re not. And you can then tailor the security rules based on what you know about the user. And then maybe flag the order for review, or even give it priority service and reduce the level of security just based on the characteristics of the user. And this is connected to what developers think about when it comes to security, is they are getting spam sign-ups or.

they’re getting bots scraping their website or they need to deal with email validation. To solve those, you buy a bot detection product or you buy a WAF. But developers don’t think about that. They think about the problem that they need to solve.

James Governor (09:22)
And so, well, again, specifically, so bot detection is one. What are the, you know, is there a sort of a, yeah, what is the list of functions that you’ve initially tackled so that developers, as they’re working, be like, aha, I need an X. What are the Xs? Or I need X, Y, and Z. What’s X, Y, and Z? Or Zed?

David Mytton (09:41)
Yeah, so developers typically come just for one thing, and then they’ll see that we do a load of other things. So bot detection is pretty common. So let’s take the sign up form spam example. You’ve got a form on your website. It’s a sign up to your application, and you’re getting loads of fake accounts, people abusing the service. So to solve that problem, you need to do bot detection first, make sure it’s actually a human signing up.

You need to do rate limiting to prevent multiple submissions and spam. You probably want a WAF component in there as well to protect the form in case you have made a mistake or there’s a vulnerability in the framework or the server that you’re using. And then you also need to do email verification and validation to make sure it’s a valid email, stop free email addresses if you want to, or certainly to stop disposable email addresses. And for all of these, you’ve either got to build it yourself or you’ve got to buy a product.

a separate product for each one of these categories and then you’ve got to integrate it. What we do with Arcjet is to offer each of these as individual components. We call them primitives and you can use each one individually or you can compose them to build up layers of protection. So for the form protection component, that is bot detection, rate limiting, email validation and WAF and you can integrate those all in just a couple of lines of code with Arcjet and the idea is that that solves a very specific developer problem.

It’s a couple of lines of code. You drop it in, it solves the problem, and you move on to something else. But because each of these are individual components that can be configured independently, as your use case becomes more sophisticated and you want to tweak the rules, change things dynamically at runtime, then you can do that because you have the full power of it all just being code.

James Governor (11:30)
So, well, I’ve got two questions immediately there. But if we think about as code, one of these is very notable, I think, about your platform. Let’s just say you didn’t start with Java. You are very much about modern languages, modern frameworks. So tell me a bit about your decision-making, what you feel you need to support, and I guess from that perspective, what your initial developer adopters look like.

David Mytton (11:57)
Yeah, so building this, was looking at where are most applications being built today, most new applications, and what kind of security challenges are they facing on these new platforms. So our target, the little tagline we have is developers building using modern frameworks, deploying to modern platforms who don’t have a security team, because those are the developers who are really having to solve the problem. And on these modern platforms, they are…

They’re new and they’re doing loads of interesting things, but that also means the security story is a lot less mature than, say, deploying on AWS. And they’re often also serverless environments, which makes using traditional security tools really difficult, because in serverless environments, the serverless function is going to spin up dynamically. It might stay warm for a period of time. It might not. It might recycle.

And you only have a very limited runtime environment. So you can’t install an agent, for instance, from one of the traditional tools. And often all of these platforms will also bundle a CDN, their edge network, where they do caching. And so putting another CDN in front of that CDN causes problems. It slows things down. It obfuscates the headers that get sent through.

It causes routing problems, and that’s why someone like Vercel, for instance, specifically says, do not use Cloudflare in front of us, because it causes all of these problems. And so that basically rules out all of the traditional security products. You can’t use a network proxy, and you can’t install an agent. And then these new applications are basically being built in JavaScript. It’s JavaScript on the front end, and it’s JavaScript on the back end. It might be TypeScript, it be Node, it may be Deno or Bun, but it’s essentially JavaScript.

James Governor (13:36)
Yep.

David Mytton (13:41)
And this is where a lot of the innovation has been happening over the last few years is JavaScript certainly has an interesting reputation for how fast it moves. There’s always a new JavaScript framework to play around with, which can be super annoying, but also really fun with all the new things that are coming out. But it’s the most popular language in terms of new applications. And so that’s where we started. Arcjet is literally MPM install Arcjet. It’s a dependency in your application. It’s in the package.json file. and you bring it into your code just like any other library.

James Governor (14:14)
Okay, and how does it work architecturally? I’m like, you know, I’ve got a couple of lines of code, but obviously there’s a bit more happening than that. So where does that, where does that actually happen?

David Mytton (14:24)
So the SDK bundles a WebAssembly module. And this is an implementation detail for our users. They don’t need to know about it. I think it’s super cool. It’s how we do things. But as far as the developer’s concerned, it makes no difference. But what it enables is to have a secure sandbox so we can pipe in the full HTTP request. Arbitrary HTTP is really dangerous to deal with. And so having it inside the WebAssembly sandbox, is a

a key feature of WebAssembly gives us a secure environment to do the analysis. And it also does the analysis at runtime performance equal to native code. And so we get this secure sandbox. can run the rules locally in your environment so that the body of the request, for instance, never leaves your environment. We can do all the analysis locally. One of the things we can do, for instance, is personal information detection.

And most of the other services that offer this require you to send that data out to a cloud service through an API, which I think kind of defeats the point of doing it, because you never want to handle that person. Yeah. So I think that should be analyzed entirely locally. So we can do that with the WebAssembly. And the WebAssembly module is written in Rust. WebAssembly also is cross-platform.

and although currently we only support JavaScript, we’re working on other SDKs for other languages like Python, Ruby on Rails, Java, and we can compile it once to WebAssembly, and then the work we have to do for the next SDK is just about hooking that WebAssembly component into the language runtime itself.

James Governor (16:05)
Okay, cool. I mean, think there’s still a little bit, I mean, for all the, the powerful idea of WebAssembly, there’s still a little bit of, there’s still a little bit like, we’re still kind of waiting for it to take off. Security seems like a good context for it. And that’s an interesting implementation detail from your perspective.

David Mytton (16:19)
Thank you

That’s right, I think we’ve seen some really interesting WebAssembly products. I think Figma is obviously the one that everyone always cites, just because it’s been so successful in using WebAssembly to implement their design tool. And we’ve seen startups and projects use WebAssembly for WebAssembly’s sake, rather than using it as a, because it’s the best technology necessarily. And I don’t think, I’d question how successful some of those platforms have been.

when you have to write your code to compile to WebAssembly and deploy that WebAssembly module to a cloud service. I think the server-side WebAssembly is where the most interesting stuff is happening, or least until Microsoft released their recent WebAssembly sandbox for AI tools, basically a sandbox that the AI agent can run inside. then you can, because WebAssembly has this capabilities-based model where it has basically the…

It can’t do anything by default, and you have to tell it, yes, you can access the network, or yes, you can access certain libraries on the host. And that means it is truly secure by default. And this is a real challenge with AI doing stuff by itself. An agent with arbitrary access to the internet, arbitrary access to personal information, and different APIs on the host is really dangerous. And it’s where we see a lot of the recent attacks.

And Microsoft released a Rust-based WebAssembly tool just a few months ago where the agent will run inside the WebAssembly sandbox and then you can choose what you want to give access to. So that only works because of WebAssembly, but it’s not selling it because it’s WebAssembly. And that’s kind of why I see the subtle differences. And I think that’s how we’re also using it is you’re not using Arcjet because it’s using WebAssembly.

you’re using Arcjet because of the features that WebAssembly allows us to do.

James Governor (18:25)
Right, okay, interesting. And tell me a bit more then about the, you sort of mentioned being able to run these rules locally. And I mean, you’ve even sort of, I mean, it’s definitely marketing by an engineer, David, you’re, what was the, you use the term local first? So, yeah.

David Mytton (18:26)
Thank

Local first. Yeah.

James Governor (18:50)
Tell me, I I think this is interesting. So we’ve got the notion of local first security. And I think from my perspective, it’s just an interesting one because it’s not necessarily a term that everyone’s using. But say a bit more about that as a sort of, I guess, a… as a principle and also I’m interested in whether you think it might start getting used more broadly.

David Mytton (19:17)
It’s probably a terrible marketing tagline. I think developers love it. And you see this with local first, like the consumer apps, like a note taking tool or some other tool where everything is synced locally. All your data is available on your computer, hopefully in like a plain text format so you can easily move it somewhere else. And that means it works offline and it maybe it syncs to a cloud service if you want to sync across devices. we’ve just not, consumers just don’t care. They’re not interested in it. It’s the challenge with privacy.

And the fact that consumers are very happy to use services for free in return for giving their data away, they just, on the margins people care, but really ultimately it’s not worked out as a marketing strategy. Apple is probably the biggest company that’s been able to do it, but I think it’s kind of incidental just to Apple’s philosophy that they want to run everything on the device, and so it’s local first, it’s privacy first, just in their approach. But we’ve seen that consumers generally, they’re not too interested in it, but.

And so that’s why I think it’s not a very good marketing tagline for us. And it’s not the headline on the website, but it’s in the details. And what I’ve found is that when I explain how Arcjet works, just like when I was talking about the person information detection side of things, security engineers in particular, but developers, that’s when they get it and they understand, okay, so that’s what local first means. That makes intuitive sense to me.

James Governor (20:31)
Yeah.

David Mytton (20:43)
rather than them going onto Google and searching for local first security. Pretty sure there’s zero searches for that term. But when we’re explaining it in the context of a sales discussion or going into the technical architecture, saying that we will try and run all the security analysis in your environment inside the WebAssembly sandbox first, makes sense from just the data side of things and not sending it to the cloud, but also performance. Because when you run the Arcjet SDK,

and the rules you’ve configured mean we can take a decision locally, like with the personal information detection or with the basic email validation, we can return in less than one millisecond the result. so the overhead is basically negligible. And that combination of the privacy component of not sending data anywhere and also the minimal latency makes the local first tagline work, but it doesn’t work by itself.

James Governor (21:36)
Yeah, yeah. mean, certainly, yeah, anything that reduces latency makes developers happy. I that’s, I mean, I guess all users, but developers in particular. Although we do see, I mean, it’s quite interesting living in a world where developers are back to waiting for things because, you know, waiting for agents to go off and do things.

David Mytton (21:44)
Yeah. Yeah, well that latency is like 10 seconds, 30 seconds. That’s serious latency. You can’t…

James Governor (22:03)
Yeah, but also, mean, DevTools were things that we weren’t paying for and now we are.

David Mytton (22:09)
Yeah, yeah. This is

important because Arcjet is often put in this category of runtime application security products, RASP, which existed kind of 10 years ago and basically failed as a category because the way that those products had to integrate into the code base was monkey patching code, it was causing performance problems, it was getting in the way of developers doing things, and as soon as you have a tool that’s doing something,

that may possibly slow down your application. If there’s any performance issue or any kind of problem somewhere, the developer’s gonna blame that first. And so this category of RASP just failed because of the poor developer experience, the latency and the interference with code. And whilst what Arcjet is doing is very similar in many ways, it’s completely different on the architectural side of things, which means we don’t have any of those issues.

James Governor (23:01)
Okay, yeah. So I mean, again, and funnily enough, it’s a term also not used by developers very much. One of the ways to describe it is sort of inner loop. Just being in the, well, rather than it being a separate production thing, it’s much closer to what the developer does. Okay, so I think one of the questions for me is, there are some, and look, obviously, you

Competition is what this whole business is about in a sense, solving customer needs, but there’s competition wherever you are. But increasingly, mean, there’s sorts of functions that you’re talking about providing. Whilst they may not be provided local first, they’re increasingly sort of platform concerns. expect, know, if you’re, you we mentioned Vercel, certainly Vercel, Cloudflare, they’ll be like, don’t worry, we’ll solve this problem for you.

They’re pretty good at getting people to adopt their technology. what’s your perspective on how do you build that independent story, that independent business when competing against cloud platforms is not easy?

David Mytton (24:22)
And we’ve seen this with Vercel. They added their WAF product first. They’ve added Firewall, which is their bot. They’ve got a bot protection product. And they recently just copied us with their bot ID product, which is an SDK that you install. But these are all tied to Vercel. And show me a developer that deploys to just a single place. The challenge there is you’re then tied into that platform, and you’re using that security product. Then if you decide you want to move from Vercel to fly.io,

and it’s a different architecture because you’re deploying almost a Docker container. How they do it behind the scenes is not actually Docker, but it looks like a Docker container. Let’s say you’ve got a marketing website on Vercel, you’ve got an internal tool on fly.io, then you’ve got some core infrastructure on AWS and GCP. You’ve now got four separate cloud environments with four separate paradigms on how they all do security. And whilst you could put Cloudflare in front of AWS,

you can’t put it in front of Vercel and you can’t put it in front of fly.io. We’ve partnered with Fly we’re they’re recommended application security product and the great thing about how we’ve imprinted the Arcjet SDK is it’s completely agnostic to where you’re running it. And so you can move the code from your local laptop where it runs in the exact same way when you deploy it to Vercel and then if you decide to move that same application over to Fly, you don’t have to do any changes with Arcjet.

really a truly platform independent security platform. And that’s part of the story for developers, because they’ll often start in one place and end somewhere else. Or in large organizations, there’s going to be all sorts of different platforms in use. And this is where we start to talk to security teams who need to deal with compliance across all these different platforms. They want to have visibility everywhere. They want to make sure that there’s a consistent set of rules that are deployed in every environment. And if you’ve got

four separate tools on four separate platforms, most of the security breaches that you see are due to configuration problems or holes, basic holes that you just didn’t realize existed. It’s not some amazing zero day that got through 37 layers of security and just happened to find the right floor that was sold for a million dollars on the dark web. You actually left an S3 bucket misconfigured and it was public by accident. It’s these simple problems.

And it’s because there are so many platforms, so many tools, so many configuration options. And having this cross-platform product, I think, is one of the selling points for organizations once they start to get larger.

James Governor (26:52)
Right, so let’s talk a bit about that business model. And particularly, mean, so you’re building developer experience first. You you do want developers to get excited about using the tools. And, you know, certainly if we go back, that Snyk story was supposed to be, developers will like it and they will get the organization to use it. Tell me a bit about that balance between like, when would you begin, when would you begin, you know, when does the security team

know, ping David and say, hey, my dads are saying this is, you know, about a mousetrap. We need to talk. You know, the story sounds pretty good. We’re not sending data off somewhere else. So yeah, like what’s the free tier for developers? If there is one, how do developers get started? And then when, when do the annoying security people get involved?

David Mytton (27:49)
So think about this in three separate stages. There’s companies that have no security team at all. So it’s just developers doing it when they have time. Then there’s when you make your first security hire. And that’s generally as you start to scale. So maybe series B, perhaps late series A in a startup world. And that’s where you have someone come in to start dealing with IT security, getting things sorted out from a compliance perspective. And they are probably one of the busiest.

James Governor (27:53)
Right, yep.

David Mytton (28:17)
It’s the busiest person in the organization because there’s so many things you have to deal with. And our goal at that point is that they recognize Arcjet and they know that we’ve dealt with a load of things on their checklist. And then the third stage is when there’s an actual security team. And that’s when the buying conversation shifts. The first two stages, it’s very much in the engineering side of things. It’s the developer who’s installed it. You just sign up. You click to join through Google or GitHub, whichever social login you’re using.

You can install it, npm install Arcjet, within just a couple of minutes, and then you deploy it it’s running. The free tier gives you half a million requests per month, and there’s certain restrictions on different features that you can use. And then above that, there’s basically request-based pricing, and it’s split out by the different features that we have. So email validations are built separately from the personal information detection, for instance.

So it’s request-based within quotas that we allow, but it’s a soft limit. So we basically want people to use the product, try it out, and then start having a conversation with us once they start to be successful, really. Half a million requests a month basically allows any side project to use it for free. And then you start to get into the pay tiers, which start at $25 a month for once you get into the serious usage.

When we’re talking to security teams, that is often a completely separate question. It’s product security engineering and the CISO organization, as opposed to the CTO or engineering organization. And they have very different incentives. The security organization is risk mitigation compliance. And it’s often default to no in the less developed organizations, rather than trying to enable developers.

Whereas developers are trying to solve the problems that we talked about and they’re trying to find a tool that will just allow them to move on to the next thing, deal with the customer request. And the tension comes when those teams have to interact because security is often forcing developers to do things that are not going to have customer facing impacts. They’re not going be building a new feature. They’re not going to generate revenue necessarily. And this is where we see the historic tension between those security and engineering teams.

And my hope at this point is that because the developer experience is something we focus so much time on, when developers are asked to use Arcjet or they go out and they look at different tools and they’re evaluating them, the developer experience is the thing that is the clinching factor for developers. And for security teams, it’s the fact that we can deal with all of their questions and their concerns on these new platforms. In large organizations,

there’s probably already a security organization. There’s a lot of tools that they already use and they probably have a well-defined pathway for developers deploying things. what I have seen over the last year or so is that as AI has become a mandate within organizations, developers are given almost like a separate area of the org with different permissions and the ability to buy things and use the best tools rather than necessarily using the ones that are within the normal envelope of the organization.

So that means they can pick Next.js, they can pick Vercel, and they can pick all of the really interesting cool developer tools which maybe the core organization for the core product is not using. And when they’re thinking about security, our goal is to help developers continue to ship quickly, but to do so with confidence so that they can deal with all the security challenges but still maintain the velocity that the new AI world really needs.

James Governor (31:56)
Wow, that’s a great snippet. That one’s definitely getting cut and used in social. Nicely done, So I guess one question, mean, the sorts of people that are going be interested in security as a feature, in the sort of frameworks that you support. I mean, those are largely startup people. So just tell us a little bit about your sort startup journey, where you are on that, funding round, investors.

David Mytton (32:00)
Thank you

James Governor (32:26)
Some of that, you know, I just know that the sorts of people that are going to be interested in this tooling are interested in those questions. So where are you on that journey?

David Mytton (32:34)
Yeah, so Arcjet’s raised a total of $12 million. We closed our Series A earlier this year, 2025, and the round is led by Andreessen Horowitz and Plural, who are the European fund with ex-Skype and Twilio founders there. Also participating in the round was ex-CTO and the ex-CEO of Twilio. So we’re really focusing on the developer side of things.

I’m getting experience, people who build developer first products to scale in the past. And we’ve got about 20, 25 other angel investors as well. So we’re well capitalized. The company has been around about two and a half years, founded in 2023. This is my third company. My first one, Server Density, I started in 2009, which is a cloud monitoring product. We raised a little bit of money and then sold the company in 2018 to StackPath, which was an edge security network.

I ran product engineering for them until the end of 2019 and then left and started console.dev, which is a DevTools newsletter that I’ve been writing now for five years. I know you still get it every week. Just little reviews of interesting developer tools. And through that, just playing with developer tools all day, every day, spotted this disconnect between security, which I was just doing for fun, doing hacking challenges.

And the dev tools which was having all of this innovation, all these really interesting products coming about and just the lack of anything in security, which prompted me to start playing around with different ways of implementing things, messing around with WebAssembly, and that ultimately became Arcjet.

James Governor (34:12)
By the way, everyone, yeah, just to Dave’s point, the console.dev newsletter, I mean, it’s sort of my job to keep abreast of what’s going on in developer land and keep an eye on cool new tools and communities and why they’re adopting them. And it’s a really good newsletter. you know, if you’re interested in, what I like about it is it just boils things down to…

you know, just a few things that you might consider. It doesn’t, you know, it’s not like trying to be entirely comprehensive, but it’s just, are some cool things. This is the good, the bad, and possibly the ugly. And so, yeah, no, it’s really good newsletter, and I heartily recommend that you subscribe. Okay, I don’t think we need to do the full six hour Lex Friedman style podcast today.

David Mytton (35:01)
Thank you.

James Governor (35:11)
But really interesting to hear about, I think a different approach to security, one that will hopefully appeal to the modern dev. So yeah, basically David, thanks so much for joining us today. And I don’t know if you’ve got any sort of yeah, I don’t know, final thoughts about, so the world is how it’s gonna change and why developers should come and check out Arcjet.

David Mytton (35:37)
think the motivation for developers is perhaps changing their reputation. It’s not that developers don’t care about security, it’s just they have really bad tools and they’re focused on solving problems, not buying a WAF. And so that’s what we’re trying to do with Arcjet is just to make security something that they can do in the same way as they have fun with Tailwind, making things look nice, set up a database without having to think about scaling it with Supabase, mess around with networking like with TailScale or deploy something.

quickly to 10 different regions around the world with fly.io. And those kind of really interesting taglines for each of those products, we want to do the same with security.

James Governor (36:14)
Okay, there you go. Developers do like security. They just haven’t found the right tools yet. And on that happy note, thanks so much, David. Thanks all for joining us. And yeah, look forward to you all staying on board with MonkCast. There’s gonna be plenty more great interviews to