Okay, not a friend. I don’t know who he is. But he accidentally appeared alongside me in a photo taken in Tokyo this week. He’s down here, above my left shoulder. Let’s call him “他人” (stranger).

I was tired, and drinking hot coffee from a can. Why on earth don’t we have this in North America?

I didn’t even notice him. But Facebook did. When I uploaded the picture, Facebook asked me his name:

When he decided to stroll through Harajuku yesterday afternoon, 他人probably had a reasonable expectation of privacy  His expectation came from three basic assumptions:

• He did not expect to be photographed
• He did not think a stranger would know who he was
• He did not expect any sightings of himself to be stored in a shared public place

In other words, he thought that most of the people who saw him would have no context to recognize him, and that any sightings would be ephemeral and soon forgotten.

But those assumptions are rapidly crumbling. Everyone takes pictures now, and as we get prosthetic brains—in the form of Google Glass—we’ll record everything we glimpse. It doesn’t matter whether we know what’s in those glimpses, or even if we paid attention. They’re recorded, and someone, somewhere, has context.

Crowdsourcing facial recognition

If I upload many pictures of the same person (say, thirty photographs of my daughter) Facebook groups them into people it recognizes.

In the example above, Facebook has found several “people” (all of whom are in fact the same person—my daughter) and asked me to name them. By telling Facebook they are all the same person, I have taught Facebook’s algorithms how to recognize her better. This happens whether I share the pictures with anyone, or simply keep them to myself.

In an entirely separate life from my own, 他人 probably has friends who tag him on Facebook. When the facial recognition algorithm doesn’t guess who he is (because of the lighting, or the position of his head, for example) his friends helpfully supply his name. His friends are training the system to better recognize him, just as I am training it to better recognize my daughter.

I have a picture of 他人; his friends have the context to recognize him; and his image is stored forever on Facebook’s servers. The intersection of these three facts is a place of little privacy:

More and more metadata

I gave Facebook a considerable amount of additional information when I uploaded the picture: where and when I took it, my phone’s GPS, the kind of phone I had, and so on.

Smartphone picture-taking is further helped by tools like Google Goggles that can find text, logos, and clues in images. It can infer things about the weather (from the exposure levels or cloud cover) and my surroundings (from the signage on stores around me.)

Here, for example, is a screenshot of Goggles from a trip I took to Prague in the Czech Republic last year:

Without assistance, using only the edge of a building, Goggles has figured out what the landmark in front of me is.

These are free tools. This is the kind of technology that Three-Letter Agencies could only dream of a decade ago; now, such software is commonplace and often runs in the background, by default.

The result of all this is that, once the training is good enough, every time 他人 shows up in a photo, Facebook will have the necessary information to know where he is and have an idea about what he’s doing.

To be clear: Facebook doesn’t really know it today.

• There are 127M people in Japan, and only 13.5M (10.5%) of them use Facebook.
• Globally, only 1.06 Billion of the over 7 Billion people on the planet (15%) use the social network.
• Facebook has better things to do than violate privacy agreements.
• Most importantly, analyzing every upload against every picture of every one of Facebook’s users is costly and time-consuming.

Let me repeat: Facebook isn’t currently doing this. But all of the systems are in place to make it trivially easy to do so, and while it might be prohibitively costly to do it for the whole world, it’s downright easy to do it for “persons of interest”.

Can a photograph steal your soul?

Much of the concern over tools like Google Glass has centred around the ubiquity of recording. To me, it’s not the pictures that matter—it’s that we’re training the machine to find context within them in shared environments. The rapid convergence of anonymous data collection, crowd-sourced tagging, and central sharing changes the game dramatically.

A world in which every public sighting is part of a searchable record of our activity is a strange one indeed. Maybe Native Americans were right.

At first, many Native Americans were wary of having their photographs taken and often refused. They believed that the process could steal a person’s soul and disrespected the spiritual world.

Consider the implications of this for law enforcement. There’s no warrant for surveillance—there doesn’t need to be. Insurance adjustors might one day subpoena such records to see if a claimant is actually ill. Spurned spouses could use this information in divorce proceedings.

Maybe you aren’t a criminal. Maybe you have nothing to fear. But as Monty Python reminds us, who hasn’t broken the law?

What kinds of legal protections need to be put in place for this? What protections could be put in place? Can a social platform be compelled to release evidence that could absolve or convict someone? And at what point can a government apply a sort of digital “eminent domain” and access private information to find someone?

I was discussing this prank over lunch this week, and folks I was with pointed out that the company has a “movie genome” model similar to “music genomes” of Spotify or Pandora. Netflix categorizes its movies by specific “genes” that can be used to define a movie—not just genres like “action” and “horror” but also things like “a strong female lead.”

Clearly, Netflix knows a lot about what you like, which is good for finding more you like. This might also be useful elsewhere: If you only enjoy films with happy endings, maybe you’re not well suited for a telemarketing job that has a lot of rejection. If you stop movies half-way through watching them all the time, maybe you have trouble focusing.

Can we detect bad mental genomes?

There’s a more serious aspect to this. I’d been thinking about it for a while, and was going to write something last year, and then the horrible events of Sandy Hook happened. Not wanting to jump on the bandwagon, I put the thoughts aside and chose to hug my daughter a little tighter instead.

Just to be clear: I am not advocating any of this. I’m just proposing a thought experiment because I think it needs to be discussed.

Yesterday’s conversation made me think about it again. Imagine that Newton gunman Adam Lanza had been a Netflix user. Imagine that, as a result, Netflix has his movie viewing history, and that this was a relatively distinctive mental genome. And imagine—no mean stretch of the imagination, this—that the company can find other viewers with similar “genes.”

Suppose Netflix’s query returns a few thousand people who are a close match. What would we do with the results? There are 525,000 police officers in the US. There are a decent number of social workers, too. It wouldn’t even need to come to the pre-emptive arrests of a Minority Report. Just a note to a student counsellor, or casual surveillance, or maybe even enough to warrant a surveillance of their purchases, emails, and phone calls.

Unthinkable, and a clear violation of our 9th amendment right to privacy? Maybe. Certainly the weak link in my thought experiment is that Lanza’s movie-viewing habits are somehow a fingerprint of his mental state. There’s good evidence that, given enough data, we’re easily predictable—but data can’t fulfil its prophecies.

Here’s the real, awkward, I-have-a-bad-feeling question: when does our desire to save a schoolful of definitely-innocents from possibly-death override the privacy of a maybe-guilty?

I worded it that way precisely because this is a game of probability. Lawyers have a name for coincidences that lack causal proof. They call it circumstantial evidence. But the bar for a jury is “reasonable doubt.” Reasonableness is a probability too. Juries weigh probabilities.

Guilty until proven innocent

Let’s take another example. Imagine that, twenty years from now, everyone has a mobile device that records their location. This isn’t a big leap of faith: Soon, it will be nearly impossible to function in society without one.

And then someone is robbed.

There’s no data on a person having been at the scene of the crime, but one person in the whole country happened to disable their device during the hours of the theft.

Are they guilty? That’s certainly a higher probability—one in 300 million—than the kind of evidence by which many juries convict people. It’s definitely enough to haul them in for questioning. Is lack of data grounds for suspicion?

Both prediction-of-crime and absence-of-alibi challenge the fundamental notion of “innocent until proven guilty,” a basic right we’ve had for a thousand years, since the Magna Carta.

Only disconnect

One approach to avoiding all of this analysis is to simply “disconnect.”  But that, too, might backfire. Following shootings in Norway and Colorado, reporters speculated that the shooters’ lack of presence on social platforms was an indicator of their intentions.

The German magazine Der Taggspiegel went so far as to point out that accused theater shooter James Holmes and Norwegian mass murder Anders Behring Breivik have common ground in their lack of Facebook profiles.

The simple act of disconnecting becomes another gene in the mental genome, another piece of gristle for the mill of data.

Unless, that is, we do it in large numbers—letting us remain somewhat anonymous.

It’s a kind of “herd immunity” for online privacy.

Herd immunity (or community immunity) describes a form of immunity that occurs when the vaccination of a significant portion of a population (or herd) provides a measure of protection for individuals who have not developed immunity.

Maybe going off the grid is a kind of privacy vaccination; those of us who don’t need to, do so to protect those who need anonymity. It only works when enough people do it to avoid the nightmare scenario of guilty-by-lack-of-information.

At the same time, maybe it gives the bad guys somewhere to hide. As another mental experiment, imagine a town where every action by every citizen—including the act of accessing others’ logs—is logged, and open to the world. Would that be a better or worse world? Would it be free of crimes, since everyone would know they’re being watched? Or would it soon turn into a eugenic nightmare, a digital Lebensunwertes, where increasingly small deviations from societal norms are frowned upon, even excised?

There’s no right answer. But what’s clear is that we’re quickly becoming a new species, one half of which has a hive mind. The legal system of a hive mind would be vastly different from ours, governing things like encryption, public shaming, tolerance, prediction, freedom to speculate, and amnesty.

In discussing whether the US constitution should include an exhaustive list of rights afforded to citizens, Dr. Benjamin Rush said, ”our Rights are not yet all known, why should we attempt to enumerate them?” I’m no constitutional scholar—I’m not even an American—but it seems to me that as we start to sequence our mental genomes, it’s time to enumerate some of their rights.

Some credits

Hat tip to Chris Taylor for the title of this post; he’s written about data privacy eloquently before and is good at picking the right titles, despite what he claims.

And a quick thanks to my trust T-Mobile Mifi giving me better bandwidth in a taxicab than I enjoy at home in Canada.

(If you wonder why I care about such things, you should read this post about onerous roaming fees for Canadians.)

Nilofer Merchant calls sitting “the smoking of our generation,” and that’s not hyperbole. Lured into chairs by our online lives, we’ve become sedentary. Our children are growing, horizontally, at an alarming rate. And when we do get up, it’s often to sit elsewhere—over lunch, in a coffee shop, and so on.

In a series of conversations over the last few weeks, Nilofer and I have been discussing all manner of things, from the power of networks to how to change behavior. Her admonishment to get out and walk got me looking for other simple hacks that might help me be healthier.

For starters, there’s ambient logging. I’ve been wearing a fitbit for the last few months, and I noticed some spikes in activity when I thought I wasn’t exercising.

Checking my calendar, I realized that these coincided with conference calls. Turns out I pace. A lot. I’d rather be doing that outside, but it’s unfair to inflict background noise on others. I’ve seen ads (and a pretty compelling demonstration) for a phone headset called The Boom, which apparently cancels out all background noise. So hack #1: go on a walk for conference calls. At least when the weather has improved a bit. (You can read more about Nilofer’s TED talk here.)

This seems to be the inherent problem with life-logging. It doesn’t, on its own, change behavior. It makes you more aware (and sometimes changes what you do—putting calories next to menu items reduces calories consumed) but the hard thing is to tie metrics to action. Short of a fork that beeps when you eat too fast (and yes, such things exist) we need to tie information to action by correlating it with behavioral patterns.

I’ve also been trying to get better at working on tasks in chunks. From what I’ve read in various places, humans are good at 90-minute blocks of focus—maybe that’s why movies tend to be that long. According to the New York Times:

Working in 90-minute intervals turns out to be a prescription for maximizing productivity. Professor K. Anders Ericsson and his colleagues at Florida State University have studied elite performers, including musicians, athletes, actors and chess players. In each of these fields, Dr. Ericsson found that the best performers typically practice in uninterrupted sessions that last no more than 90 minutes. They begin in the morning, take a break between sessions, and rarely work for more than four and a half hours in any given day.

If that’s the case, hack #2 is breaking the day into four 90-minute chunks, and planning some kind of physical activity in between them. This also helps me realize that, no matter how well-intentioned my list of activities for the day might be, I can really only do three big things a day.

But why stop there? I should be picking smaller activities for the batches of short, more reactive, interruptive tasks like responding to Twitter or catching up on correspondence. Finished reading unread mails? Time for some push-ups. Written a short note? Time to stretch. So hack #3 is associate proportionate exercise with each chunk of work. Not sure if this works, but it sounds like an experiment worth trying, albeit one that requires a bit of office reconfiguration.

I’m also someone who writes and speaks a lot. I need to remember that old parable about having two ears and one mouth and using them accordingly. But when I finally do get a chance to consume something (like Jonathan Haidt’s The Righteous Mind) it pays off a hundredfold, giving me new angles and examples.

On the other hand, I had Lasik a couple of years ago, and it’s made it harder to look at screens for a long time, so maybe I need to push 5-10 blogs or PDFs a day into an e-reader, and consume them. And it has to be something without an Internet connection that interrupts me. So hack #4 is to invest in the modern equivalent of a newspaper, using something like Prismatic to select content and push it to a device with big fonts that’s easy on the eyes, like a Kindle.

If I’m breaking things into chunks, why stop at reading? For most of us, documentaries are the mental equivalent of exercise or healthy eating—we know we should be doing it, but we default to junk viewing and escapism instead. There’s so much good content on Netflix and other sources that I’d like one of my 90-minute breaks to be streaming a documentary.

The problem here is finding stuff I wouldn’t normally want to watch. I want Netflix’s suggestions not for me. If the point of this hack is to broaden my mind, I want to avoid the filter bubbles that reinforce my worldview.

I could get through 200 documentaries a year with that model. Hack #5 is to find a list of must-watch documentaries, and consume one a day.

Back to health: I spend a bunch of time in transit. When I go to a city for more than a few days, I’d prefer to ride a bike. It’s cheaper than cabs, and better for me. It’s better for the planet. And if I leave a bike behind in each city, donating it to a Boys & Girls Club or Goodwill, I’m making the planet a better place. So hack #6 is to find a way to use bikes instead of cabs.

(A smart librarian in Austin pointed out that David Byrne has been doing this for a while, and I tried it at SXSW, and learned a lot. I’ll write that up at Lean Analytics, as it’s chock-full of lessons for entrepreneurs, too.)

There are other hacks worth trying: I’m using Flux to dim my screen in the evening, which helps combat the chronic insomnia we get from staring at bright white lights. And I should probably use public transit in strange cities, instead of hiding in taxis.

Provided I can remember this advice:

(Hat tip to redditor nathanwpyle, who posted a series of animated GIFs based on what he’s learned about living in NYC. You can see the comments, and the link to the rest of them, in his reddit post.)

Any suggestions for other things I should incorporate in the coming year?

I bought a T-Mobile MIFI device for $125, and a 5GB SIM for $50. Right now, though, you can buy the MIFI online for $25, as they’re having a promotion.

I decided to do the math. I’ll have used this 5GB data plan for nearly a month, and according to T-Mobile’s (good, transparent) diagnostics, in that time I’ve used 1.9 GB of data. I’m getting roughly 4Mbps of data down—enough to watch a movie if I feel like it.

I then compared this to Rogers’ current US data roaming rates. They were a bit hard to find by searching on Google—the page that’s indexed by the search engine (http://www.rogers.com/web/content/wireless-network/wn_usdataroaming) is blank.

I eventually found the page by going to Rogers first, and grabbed the pricing plans listed there. Based on this math, I’d have spent over $1,900 without a roaming plan (using incredibly badly named “preferred” rate.)

Even with the most preferential plan, I’d have been able to buy two MIFI devices with data plans. And now that I have a MIFI, my data costs me around 12% of the Rogers rates. Oh, and I’ve still got a few days at SXSW and in NYC. So don’t quote me on that 12%—it’ll get even better before I’m done.

If this incenses you, you’re not alone. I wrote a bunch more about it in a previous post, but more than anything, I really want everyone to do this instead of consenting to the highway robbery that Canadian carriers get away with. It hurts Canadian business and holds back our technology growth, and it has to stop.

(Oh, and if you’re curious—no, I have nothing to do with T-Mobile. Use anyone’s MIFI. You can buy an unlocked one on Amazon pretty easily. T-Mobile just has the least locked-in contract of the US carriers.)

This is stupid.

It’s not stupid to phase out the penny. That makes perfect sense, given inflation and the rising cost of natural resources, and even the growth of credit and near-field payment systems. No, what’s stupid is to stop there.

Canada should have killed the nickel, too.

This has been bugging me since I heard of it. So I’m sorry if this seems like a rant, but it is. Here’s why.

Rounding nickels is bad. We’re a metric country. We like orders of magnitude, and the decimal place. We know how to round up or down from an even number—it’s something we’be all been taught in math class. Rounding to a nickel is counterintuitive. There are even signs on cash registers explaining how to do it now.

How to round a nickel, according to the Canadian Broadcasting Corporation

The nickel is big and expensive. I don’t have the Canadian costs, but in the US it’s $0.05 to make a penny and $0.10 to make a nickel. A dime is around $0.057 to make—the only one of the three that’s cheaper to produce than it’s worth, also known as seignorage. Of the three, only the nickel generates government revenue when introduced into circulation. The other two “cost” money.

Decimals are useful for data. Consider spreadsheets. If you only had to express a dollar as $0.0, you’d save a character per column. There’s even a button right in Excel that does it for you.

So what’s the downside? Well, at most, about $0.05 per retailer. Once. Because on average, things will even out. I mean, sure, you can game the system a bit by buying jellybeans at $0.14 over and over. But there are ways to address that—such as selling a $0.20 bag.

The reason this grates on me so much is it’s a classic example of incrementalism, and not looking at the big picture. We’ll inevitably eliminate the nickel one day, and doing both at once costs far less. Sadly, we don’t think holistically about such things.

Anyway, end of rant.

My first take on the event was that it felt like crashing someone else’s high school reunion, only to find that all the cool kids had already snuck out back for a smoke. But on closer inspection, it’s much more like Burning Man: there is simply too much to take in, and you need to abandon your FOMO and just go with it.

Once I found out I was coming, I started to find other things to do:

• The Lean Startup event at the Hilton on Saturday, March 9, where Eric Ries assembled a daunting lineup of smart people.
• A slot at the #ideaDrop library house, 4pm on Monday, March 11 (at 2902 French Place, if you’re here!) to talk about Big Data and a billion bad librarians.
• The main reason I’m here, to talk about social media monitoring with Blake Robinson, Susan Etlinger, and Margaret Francis on Tuesday, March 12. Apparently it’s sold out; not sure I can help with that if you wanted to get in.

With Lean Analytics done, I’ve got dozens of things I want to write about: crapweeding, bicycle experiments, the elimination of the nickel, and how RSVPs bite back. With all these ideas banging around in my brain, I most desperately need some time to write; unfortunately, that won’t happen soon.

• I’m headed to NYC to talk with iNovia’s team and portfolio about the problems of TMI on March 14.
• There an IBM tweetchat on March 18.
• There’s a day-long workshop at MaRS on March 19
• Later that day, there’s the inaugural Startup Metrics meetup. I think that’s sold out too.
• Then there’s Cloud Connect in early April.
• Then Tokyo, for an April 8 event at Open Network Lab; the Nikkei BP Big Data Conference on the 9th; some mentoring of Onlab Startups on the 10th, and a workshop at the Open Network Lab.
• Then off to Frankfurt and Bonn at the end of April.

I’m writing this list for two reasons. First: it would be great to hang out, if you read this or subscribe to the mailing list. Bug me and let’s grab a coffee or better yet take a hike. And second, making lists temporarily relieves the sense of not accomplishing stuff.

As often happens, one student asked me where I thought tomorrow’s hot tech startups would come from. This is the second-most-common question I get at universities, after “can you help fund/advise my company?”

My current answer is that we’ve grossly underestimated the impact that manufacturing at the edge will have on society. 3D printing is the visible tip of this iceberg, but it’s really about replacing supply chains with edge engineering, whether that’s through machine-optimized plant growth, or making new toys from the ground-up remnants of old ones, or printing spare car parts.

This student, however, wanted more. Which got me thinking—what is it that makes a truly great startup? My initial answer was that there are two key factors.

First: an idea that’s obvious in hindsight

• Do you like streaming movies into your house?
• Do you like buying music and books online?
• Do you like hailing a cab with your smartphone?
• Do you prefer to search the web by typing in what you’re looking for?
• Do you prefer to have software automatically stay current, rather than upgrading it all the time?

Of course you do.

A successful business has to be so obvious, so blindingly stupid, that once it’s passed through the early adopter stage, the whole world will embrace it. This is the key to crossing the chasm into the mainstream.

The problem is that we have a kind of technology Stockholm Syndrome. We’re trapped by the way things are done today, and it’s really hard to see what could be, if we could only throw off the shackles of today. That’s why Blockbuster—which had customers, film licenses, and industry clout—collapsed under the weight of stores and employees while Netflix flourished, unfettered.

Here’s another example: it turns out cameras are for sharing pictures. Which is why the iPhone is the most popular camera in the world, according to Flickr; and why Kodak is going out of business. It wasn’t obvious ten years ago that the killer feature on a camera was an Internet connection; today, it’s so obvious it’s hardly worth pointing out.

Great founders are delusional enough to see the world as it could be for others. Don’t mistake this for seeing the world as you’d like it to be: You’re a lousy market. But if you can clearly see the world of cars, or smartphones, or social networks, or mass air travel, you have a much better chance of success.

Second: the baby steps to get there

I used to spend far too much time worrying about a company’s technology. Maybe that was right at the dawn of the Internet industry, but today, I spend much more time on their go-to-market strategy. I firmly believe that they can do what they say; I’m skeptical that anyone will care.

The winning startups don’t just anticipate the future. They also figure out the small steps needed to get the world from their current state to that future state without alienating them or weirding them out. Here are some examples:

• Netflix knew that videos should be streamed, but broadband infrastructure wasn’t ready yet, and set-top-boxes didn’t exist. Today, we have ample network capacity and most console games can play Netflix movies. So what did they do? Simple: they mailed DVDs to people. This let them build the model, gain customer traction, and learn about user behaviors, pricing, and preferences. Mailing DVDs was a baby step.
• When Uber launched in Seattle, it knew it needed a critical mass of drivers so that riders would get a ride quickly. So it hired many drivers to sit around and wait. It paid them $30 an hour to idle, and watch their smartphones for a ride request. Within a few months, the demand was big enough to switch those drivers to a commission structure. Buying the supply of limos was a baby step.
• When Twitter first launched, it knew that the app worked well in a mobile model; but smartphones weren’t as common and the app wasn’t installed everywhere. So it let people send updates by SMS. That’s one of the reasons for the 140-character limit, after all. Today, few people use Twitter via SMS—but at the time it was a great baby step.
• reddit started as a link sharing and voting site. Then it added comments. Then self-posts. And eventually a “reddit Gold” reward model. This wasn’t planned—the founders stumbled across these enhancements, gradually shifting the company from a link-sharing site to a thriving community that makes a significant amount of its money through donations. But voting on pictures is a fun and easy first step.

It might seem like these are consumer-focused; to some extent they are. But there are plenty of enterprise examples. Blackberry made it easy for sales executives to get a two-way pager, then backed into the enterprise IT world. Keynote Systems and Google Analytics offered basic services that anyone could deploy. Salesforce.com started with rogue salespeople and small businesses, but eventually became a strategic platform for big companies.

Two big questions

These are the two really big thoughts running through my head when someone pitches me an idea. First, is this something that will seem absolutely obvious in hindsight? If so, I believe in the problem. Second, have you found the baby steps it will take to get mere mortals there? If so, I believe you’re the one to solve it.

My basic thesis was that just as we didn’t know the killer feature of a camera was an Internet connection, so we don’t know the killer feature of a book is an Application Programming Interface.

I wrote up my speakers’ notes beforehand, and they’re included below. Here are the slides from the event.

A book has always been a hack.

Years ago, when I first started blogging regularly for GigaOm, I would write long, meandering, narrative pieces. Carolyn Pritchard, their ruthlessly efficient editor, called me out after one particularly long submission and said, “Alistair, the only good thing in this is the third paragraph. I want you to delete everything else, expand that, and submit it.”

She may have been more gentle in her actual words, but it didn’t feel that way. I licked my wounds, rewrote it, and sure enough, the post thrived.

Months later, after I’d been writing for a while, I said to Carolyn, “I think I understand the difference between journalism and storytelling. In a mystery novel, the last line of the book is ‘the butler did it.’ In a blog post, that’s the first line—and the rest of the post explains why.” I’d finally learned that content and narrative were distinct.

Content, structure, format

When you peel back the onion of a book, you realize that it’s really three things: the content; the structure; and the format. The content may be whatever you like: fiction or nonfiction; objective or subjective. The structure may vary widely—even in the era of only physical books, we have chronological order, alphabetical order, and other ways of accessing the content. But for centuries, the format was ink on wood pulp.

Because the format was fixed, the content and structure were locked in at the moment of printing—what Mathew Ingram has called the “false epiphany” of publishing. This has been terrible for two big reasons.

First, content is ever-changing. There’s always an update, a comment, a thread to follow, a typo to fix. Even a time-worn Homerian epic has new criticism, new insight, new illustrations, or a new Hollywood remake that’s somehow associated with it.

And second, the structure needs to adapt to how we’re using the content. The first time I read a book, I may want to view it as a narrative, revealed in the order the author intended it. The next time, I may want to view it as a timeline. Or from the eyes of different characters. Or as a glossary of terms. Or as waypoints on a map. Form, quite literally, should follow function.

Consider the following diagram, which looks at two dimensions of published content: fiction and nonfiction, subjective and objective. Depending on these four quadrants, audiences want to consume content differently.

Looking at books in two dimensions

But we can’t possibly build all forms of interaction for all content for all readers. That way madness lies.

SaaS beats the Gold Master

Software developers used to talk about a Gold Master. This was the final release, and anything you got wrong in it cast what Ben Einstein, co-founder of Bolt, a hardware startup incubator based in Boston, calls the Long Shadow. It represents the things you have to live with if you make a mistake in the final release. For hardware, this might be things like battery life, hardware power, or component failures. For software, it’s things like security vulnerabilities, or a broken update process, or lack of support for particular platforms.

Remember when we named software by year or version? That was fun.

Look at the changes that software has undergone in the past two decades. Once, software was installed on desktops, where it grew slowly moldy until someone updated it. That ended with the rise of Software as a Service. What version is Google Docs at? Salesforce.com? Freshbooks? You don’t know, because they’re publishing now. And now. And now.

Software-as-a-Service thrives largely because it’s far easier to administer and maintain a central application than it is to manage millions of installations.

Unfortunately, economies of scale still apply. A company like Salesforce has thousands of requests for new features, which it can’t possibly satisfy. It did something smart: it created an ecosystem that let others extend the platform. Salesforce’s App Exchange has tens of thousands of third-party tools that can extend the company’s CRM. Users get extra features; Salesforce doesn’t get distracted by niche demands.

Data, data everywhere

As a species, we’ve migrated from atoms to bits. Whatever the medium, we consume it digitally. The move to e-books is only part of a much bigger sea change, as we colonize a new world online.

What’s the most popular camera in the world? I’m not sure, but I’m willing to bet it’s actually a telephone. Flickr thinks so. Meanwhile, Kodak is out of business. We didn’t know that the killer feature for cameras was an Internet connection. But of course, it is: pictures are for sharing. Hindsight is easy.

Similarly, we didn’t know that hyperlinks were a natural part of books. It should have been obvious, because we’ve been stuffing books with footnotes since, well, the second book came along. There has never been a time in human history when we’ve created as much writing as today, because everyone is a publisher. We just have the wrong lenses: as Jeff Jarvis explains, traditional publishers “see content as a scarcity we produce and control. Facebook and Google … see content as an abundant resource to learn from, value and exploit.”

We’ve tried to wall our books off from the surrounding world by erecting castle walls around them. But that’s not how to add to the usefulness of a publication; increasingly, there’s more stuff about the book outside the book than there is within.

I’m asking you to grant me the following assumptions:

• That a book is actually a convenient bundling of content, structure, and format.
• That the format content takes needs to adapt to how it’s being used.
• That the content of a book shifts constantly, and that the “publication date” is a false epiphany.
• That the rising tide of Big Data means books that aren’t linked to the world are hermits, outcasts, shut-ins. 

If you’ll grant me these four assumptions—and many of you won’t, and I look forward to beers and/or angry 140-character ripostes as a result—then you have to agree that the future of a book is its interfaces: Specifically, the one between a book’s structure and its content.

This interface frees the content from false epiphany of publication: it can be updated constantly. It’ll have its own lifestream, beginning when the content was just a gleam in the author’s eye and continuing on every time someone updates a Wikipedia entry or writes some fan fiction. It plugs the book into the Big Data our quantified society is constantly generating.

This interface also frees others to build new interfaces the original producer didn’t have the time or ability to create. Just as third-party software helps SaaS companies tailor to niche markets and discover unexpected new ways of using them, so new structures allow us to understand content in ways we didn’t before.

The killer feature of the book is its API. Those that ignore this are Kodak; those that pay attention to it are Android.

“Mobile first!” they cry.

Not so fast. Mobile devices aren’t consumption devices, they’re prosthetic brains. And that means they need to tell us when they have useful information (interruption); and answer our questions (contextual search).

This is the trend we see with Siri, as well as Google Now and Field Trip. Mobile design is partly smart memory (which knows, for example, that when you say “where’s a good sushi restaurant?” at 4PM it should find something nearby, perhaps that your friends have recommended, and within your price range.) And it’s partly smart interruption (which says, “you should leave now or you’ll miss your meeting” because it knows where you are, what’s on your calendar, and what traffic is like along your most likely route.”)

Techstars’ Brad Feld nailed this in a March, 2010 post entitled Email is still the best login, and Fred Wilson calls email social media’s secret weapon. Because email jumps into your lap. It interrupts you. Want to know why SMS is still popular? Because unlike many other messaging platforms, it always pushes itself onto your screen, vaulting past the gates of opt-out and commanding your attention.

But do it too often, or for unimportant reasons, and you’re the app that cried wolf. You need to be clever, knowing when to interrupt someone. After all, a spare brain that simply distracts your main one is a recipe for ADD. That’s why smart notification services like Prismatic use machine learning to mine your social graph and the things you discuss, sending you targeted reading lists.

Saying “mobile first” is wrong. Too often, that just means “make this web page work on a tablet.” Just as web designers who simply took offline documents and rendered them in HTML missed the point of the Web, so mobile designers that simply make the web work on a phone miss the point of mobility.

Mobile isn’t the point. Interruption is.

We don’t call interruption an interface, really. It’s almost an afterthought. Blackberry did it reasonably well with a small bar at the top of the screen; Android had a decent one from the start; Notification Center was slowly built into MacOS and IOS. But few companies design for “interruption first,” even though that’s how most of us engage with the services on which we’re most active.

Ask yourself: how often do you say, “I should go on Facebook”? And how often do you instead see “someone commented on your post” and start from there? The reality is that, in a mobile posture, we start by being interrupted. Services that interrupt well avoid disengagement, which is the worst thing that can happen to a startup.

Stop worrying about taps, screens, or swipes. And start worrying about how to interrupt your users well.

Sidenote: My co-author Ben Yoskovitz and I have been writing about this stuff, and how to measure it, in our forthcoming book, Lean Analytics. We’re learning a lot.

The book is part of O’Reilly’s early release program, which means you can buy it now and get updates as we finish it off. Right now we’re up to 150 pages. If you want to buy a copy, you can do so here.

I’d love to just embed it here, but for some silly reason embedding is restricted. Grrr, Internets.