This way of working enables me to write out and organize my content in markdown-formatted text first, where I can solely focus on the quality of the content in its purest form, easily organizing paragraphs and bullets using simple text-based tools. I can use any number of markdown editors, with or without AI assistance, to ensure the core learning experience is just right for the project.

I can then save that markdown source file as content.md in a folder on on my desktop, open that folder in Zed, and tell Gemini to turn it into a motion graphic video presentation.

This HTML5 motion graphic presentation was vibecoded for zero dollars in under an hour

The Presentation Tech Stack

The look and feel of the presentation was developed to my specifications in a collaborative process with Google Gemini AI. I first described that I wanted to create a responsive HTML5 presentation “stack” where I could feed markdown in and get full-screen browser-based slide presentations back. I designed a mockup layout in Pencil.dev and gave it to the AI to use as a template. Gemini recommended the most popular Javascript libraries for developing animated text experiences on the web, and built a directory of HTML, CSS, and Javascript on my machine that can ingest my markdown-formatted content and generate an HTML page with this visual look and feel.

I selected the looping video background from Pexels and told Gemini to use it with a semitransparent CSS filter layer over it, ensuring text legibility meets WCAG accessibility standards. In fact, I had the AI evaluate the entire stack to ensure that it’s fully accessible. Unlike video motion graphics, this presentation deck can be read with any standard web screen reader for users with visual impairments.

Go Straight from Text to Video

By default, the output of this presentation stack is a website, not a video. The video you see here is a screenshot recording of it. But I have also discovered a complete text-to-video solution called ReMotion where you can write markdown and then compile your content into a visual video editing timeline interface for further editing. Again you can apply any CSS theme you can dream of to your work without manually editing the colors and fonts of every little element on the screen.

Here’s a beginner friendly demonstration of how to get started with Remotion:

I’ve been experimenting with this technology in my time between jobs and am envisioning a hybrid approach where I still work in my beloved DaVinci Resolve Studio but where the time-intensive sections of any given timeline (like animated graphs or text explainers) are generated as video and integrated into my existing timelines. Going forward this way of working is sure to become ubiquitous, its AI look and feel will be easily identified as “slop”, but skillfully integrating it into something that looks bespoke but takes half the time feels like a great way to stand out in a crowded market.

Before the AI Backlash Starts…

Many people are rightly uncomfortable with the ways that AI is moving into our creative processes. Yes this process could dramatically speed up the whole creative process for instructional designers, and if we’re not careful we could certainly use this kind of technology to do all the hard work FOR US. The right way forward, I think, is to use this kind of tech so that we spend less time wrestling with pro-level multimedia software and more time designing rich and effective learning experiences.

It’s no secret that we IDs are often treated like the “turn this presentation into a video” department. This is the part of our job that can be completely automated, and the way we stay relevant is to continue to keep measurable learning outcomes, retention and transference, and learner experience core to the learning content development process since these are the parts of the work that our stakeholders routinely fail to consider. I generally skip the wailing and gnashing of teeth when a new disruptive technology comes along because I’m still always the lone voice in the meeting who remembers that we’re teaching human beings here, and the way we’re successful is not just about pumping out pretty videos but about improving learners’ performance in their roles. I say “give the bosses their pretty pictures, but use all the time you save to design experiences people will remember and grow from.”

At Autodesk, our learning design group uses a well-defined process for kicking off a training development project. Our Subject Matter Experts don’t just throw a PowerPoint over the fence to us and expect a canned eLearning product in return. Instead, we take an agile, consultative approach to deeply understand the desired outcomes of training, propose solutions, prototype working designs, solicit feedback, and deliver learning experiences that achieve outcomes and get “wows”. Here’s how….

In the first meeting, we….

Understand the Learning Objectives First

Often SMEs will come to us already thinking about the types of activities they want to offer for an upcoming training course. They know they don’t want a boring lecture and they want suggestions about interactive, engaging activities that will make their training more exciting.

This is when I put the brakes on and start asking tough questions…

Wherever possible, I like to step back from designing **activities **until we’ve articulated our learning objectives for the experience.

I coach the SMEs to think about what we want learners to know and be able to do as a result of the training. This is the question at the heart of Understanding by Design theory by Wiggins and McTighe, and I find it helps non-educators start to think about training not as a boatload of information to cover but rather a set of changes we’re trying to make in learners. The question then becomes “what are those changes” and “how will we know when we’ve been successful?”

Learning objectives can be designed like SMART criteria for to-do tasks; they’re best when they’re…

• Specific
• Measurable
• Achievable
• Relevant, and
• Time-bound So rather than just saying “we need a training about the new bribery policy!”, we would create a SMART learning objective like:
• Learners will correctly identify the three most common types of bribery.
• Learners will be able to match each type of bribery with the recommended course of action.

These outcomes tell us (as trainers) what we can measure to see if the training was effective. It also helps us edit the amount of information they’re responsible for and design activities that help them reinforce the new learning.

Measuring and Reporting Progress

When we are designing training, we always need to know about the reporting requirements —

Who needs to know what about learners’ performance?

The projects we work on vary in the level of accountability and detail that’s expected. With some trainings, it’s good enough that learners have just watched a video and answered some questions. With others, we want to get detailed data about which learning modules were accessed, which questions were answered right vs. wrong, etc.

I’ll often ask

• who wants to see a report of learner performance, and
• what data points do they want to see on that report?

This helps me work backwards from these reporting requirements to design learning activities that will give the stakeholder the information they need. We don’t generate the reporting ourselves — that happens in Sales Automation — but we design learning activities that will generate data to produce the business insights our stakeholders need.

Authentic Assessment and Activities

Once we know what learners will know and be able to do, we can start to create assessments that will measure whether they’ve successfully met those outcomes.

This can be as simple as a set of multiple choice questions or as in-depth as an online simulation of a real-world situation the learner will be expected to perform in. The solution we design for you will likely depend on your…..

• learning outcomes
• timeline
• budget
• learners’ time constraints

We have a wide variety of options and expertise that we can recommend to meet your unique requirements.

In addition to our in-house designers, we also work with a network of vendors who can provide professional services for

• narration
• videography
• motion graphics
• interactive/game design
• and more…

Localization Requirements

We always ask if this content will be localized into different languages, as it affects the budget and (sometimes) the types of media we use to create the learning experience. As a global company, our content is regularly translated into up to 11 languages, so we design the content with an eye to handing off our materials to the localization team after the English version is complete.

This means preserving the original project files when we create custom graphics or videos so the loc team can create alternate versions in all 11 languages. It also means remaining conscious of language idioms and cultural expressions that would cause confusion (0r offense!) when translated into another language. While this adds complexity and expense to our development process, it’s a necessary step in serving our global audience with high quality, accurate content.

Budget

Simple projects can often be completed inexpensively in-house, but there are costs to producing course experiences that meet Autodesk’s standards. Using professional vendor services can also increase the cost of a course development project. Be prepared to discuss a project budget for each training.

Interactivity, Engagement, and “Wows”

Our team of instructional designers strives to create learning experiences that promote active, engaged learning. This is why we build comprehension self-quizzes and other memory retention activities into each learning module, so learners can review, check their understanding, and make sure they’ve got it right.

We do this with a focus on attractive, visually appealing content that our stakeholders can be proud to roll out.

Our toolset for creating these interactive learning activities is always growing, so ask us about how we can build interactivity an engagement into your course.

I have long believed that the standard toolset of instructional designers — Articulate, Captivate, Camtasia — is an inferior alternative to the powerful tools that web developers and designers use to create experiences on the open web. Instructional Designers’ visual, consumer-oriented tools have become the standard because of two main reasons:

• The fact that most IDs are educators first, designers second, and web developers last, so their tools protect them from the underlying HTML, CSS, and Javascript that powers the web experiences they build rather than supporting them to engage with it, and…
• The requirement that eLearning authoring platforms create SCORM-compliant exports creates powerful vendor lock-in, discouraging the kind of innovation we’ve seen in the rest of the web stack.
  

AI is changing this equation dramatically. It’s now possible for non-technical users to mock up detailed website layouts, handling much of the visual design that tools like Articulate handle for designers. Even site navigation, multi-page layouts, and responsive design can be built from a combination of text and image-based prompts.

Interactive Prototype Music Theory eLearning Course made in Figma Make

This course design mockup was generated in maybe 30 minutes of work in Figma Make. I am using the free version since I’m between jobs right now, but the pro version would allow me to export this layout as valid production code and start to populate it with original content.

Turning a Mockup into a Production Site – VS Code + AI Agents

I have also been approaching AI development from the other direction — not starting as a visual mockup, but starting the process in AI-integrated code editors like Cursor and Google Antigravity, built on VS Code. These desktop code editors allow you to work with a chatbot within a directory on your desktop machine, generating files and entire website projects in code first. I generated this working website mockup with working, clickable quizzes and interactives, with just a prompt.

I had previously created a more complex version on my work machine before being suddenly laid off and losing access to those files, but not before testing the SCORM functionality I had built into the course using SCORM Cloud, where the course validated and functioned as expected!
This tells us that SCORM compliance is no longer the binding force holding instructional designers within Articulate’s ecosystem that it once was. In fact, we are now free to break out of the limitations SCORM put on our data reporting capabilities and freely move towards xAPI and CMI5‘s richer data streams for more detailed analytics on learner performance.

Markdown is the Secret Sauce for Building Content with AIs

I have long been trumpeting the benefits of learning to write markdown for instructional designers as a web-native way to organize content before we commit it to a proprietary authoring platform. I have adopted it as the main way I write any content — first in markdown, then choose the container it will ultimately live in when we present it to its audience. Doing this gives us maximum flexibility and lets us work faster. Even when my stakeholders send me their content in PowerPoint or PDF formats, I always strip it down to markdown to remove any formatting so it’s nice and clean to add into my preferred authoring tools of choice.

Apparently, our new AI overlords agree — AI loves speaking in markdown. VS Code, Cursor, and Antigravity all work great as markdown editors where you can write and generate content in an integrated view with your AI assistant, and they let you turn a folder full of markdown-formatted content files into a production website with just a prompt.

I for one welcome our new robot overlords

To paraphrase Kent Brockman, this newfound power is a huge step forward for instructional designers and the power we have to create rich, interactive online learning experiences without confining eLearning authoring platforms like Articulate. I am excited to continue my work moving towards this web-native workflow with the assistance of integrated AI agents.

Traditional Education vs. Corporate Reality

The culture around traditional education is pretty cavalier about making demands on learners’ time. Students are expected to commit to a set number of in-class hours plus a significant chunk of out-of-class time for study, homework, and projects. It’s fair to say that most classroom instructors design learning activities without fully weighing the “cost” of the time they are requesting from learners.

Contrast this with the sales executives we support in corporate sales enablement, who are under explicit time pressures. Time spent learning for them is time taken away from their main focus: making sales and hitting quotas. All of the metrics for evaluating their success in their position are based on their sales performance, not their performance on learning and training tasks, even though those tasks directly support that performance. And it’s not just sellers — their managers and their entire division’s effectiveness is judged by the amount of revenue they bring in, so there is an organization-level tension between sales and training who each jealously covet as many minutes of sellers’ time as they can get.
In this paradigm, every second spent learning must be maximally impactful on the seller’s job performance. It’s our job as instructional designers to use their time with maximum efficiency to expose them to information and measure their comprehension, and to make sure that training activities are directly aligned to the measures of performance they will. ultimately be evaluated by.

The Modern “Attention Economy”

In truth, all learners today have to make hard choices about their time. More and more people are balancing work with family responsibilities and various adulting tasks. This is compounded by a growing attention economy—social media and streaming entertainment—that seeks to capture and monetize every free moment of our attention. We only have so many waking hours of attention to divide between all these various objectives, and everyone is acutely feeling the stretch of balancing their time amidst competing priorities.

Design Decisions Matter

This reality becomes critical when we sit down to design a learning experience for the modern learner. Small decisions have a huge effect on motivation and attention, such as encoding learning content as a 30-minute unedited Zoom meeting versus 5 minutes of tightly edited, concentrated video content limited to a specific topic. We instructional designers need to judiciously decide between offering content as a conventional clickable eLearning courseware package vs. a job aid that learners can refer to at the time of need. Now in the age of AI it’s a real option to consider which content should be bundled into an AI chatbot to provide quick answers vs. which content requires us to walk learners through a sequenced learning experience.
At every step in the design process, I am always asking whether learners need this content to be committed into their memory or if it’s sufficient to give it to them in a format they can quickly refer back to when they need it. My Design for Forgetting philosophy is based upon the idea that we should be surrounding learner with quick, searchable tools that will give them the right answer at the time they need it rather than expecting them to retain it in perfect fidelity for use at some uncertain future date.

Design for Forgetting is highly compatible with the idea of microlearning, though there’s nuance needed in how it’s implemented. While we should constantly strive to save our most valuable learning resources in small, topic-focused chunks that can easily be consumed in a single sitting, we must also think of them as an interconnected whole that can be searched for in a central location so the busy learner can quickly access exactly what they need with a minimum of time or cognitive load searching for what they need.

Efficient Training Requires Supportive Platform Decisions

The need to support learners’ valuable attention efficiently affects the platform decisions we make about how content is stored and organized. I deliberately designed my onboarding program and knowledge base around a Confluence wiki site rather than an eLearning LMS because I wanted all content contained within the whole wiki to all be accessible from one central search bar. Encoding all course content as HTML text within a wiki exposes all of that content to a centralized search bar, while saving key content as PDFs, eLearning assets, podcasts, and videos can complicate a learners’ ability to quickly search and find it at their moment of need.

We also haven’t forgotten about mastodon.online. We want to do something wildly different there and push the limits of what the software can do. We have some thoughts, but we don’t really know what this looks like yet! If you have ideas about neat or useful things we could do with mastodon.online, send a message to @staff@mastodon.online.

I have some ideas for how to do something different with mastodon.online. They’re all about removing friction from the onboarding experience of Mastodon, targeting less technical users who are stuck on facebook and other legacy instances. I’m old enough to remember when America Online was a scrappy startup who rose to dominance by smoothing over the rough edges of the mid ’90s Internet and creating a safe, comfortable space for a nation of non-internet users to start to figure out what to even do on the internet. I believe that old story possesses some valuable insights that can help us envision an entirely new chapter beyond the current paradigm of today’s social platforms.

Mastodon.Online, meet America.Online

In 1995 when “the internet” mostly consisted of BBSes, IRC, and other geeky text-based protocols, AOL filled the gap with rich content programming and dedicated networking tools within their walled garden that met the needs of mainstream users. It made it dead simple for non-technical users to find news, weather, movie showtimes, people you know, and topics you want to explore at a time when finding that information outside of AOL was intimidating to unfamiliar computer users.

Within the span of a few short years, AOL really did get America online, getting the everyday mainstream netizen comfortable with then-new concepts of real time chat, email, message boards, news feeds, and several other features that made the Internet a compelling destination for a new generation of tech users.

Time-Warp to Now: Post-Social Amnesia

Since then, Facebook and Instagram have made a Meta-sized hole in all our lives that few know how to get ourselves out of. Millennials and younger generations don’t remember a time before the modern social web. Meta’s apps have become the main way that most users keep touch with old classmates, distant family and friends, buy and sell things locally, host a small business presence online, follow brands and celebrities, get news, instant message any of your connections, and more… Having a Facebook account means you have at least one identity that can easily connect with your school, your work, your IRL connections, and anonymous strangers.

It has become the de facto “connective tissue” of the Internet by systematically eroding users’ privacy, pushing us to expose more of our personal information online, overshare with third parties and strangers, and fuel outrage and political division and misinformation to further its business objectives. Others have tried and failed to make a non-evil Facebook, but the evil is part of why Facebook is so successful.

Facebook, Instagram, and especially Twitter/X have become a shared common space where public policy is debated, news is made, reported, and consumed in real time. They have become a “digital commons” where large political movements like the #ArabSpring, #MeToo, and #BlackLivesMatter have led to large scale mobilization and change. Having these important conversations on a for-profit commercial platform has become a threat to democracy and safety everywhere.

Mastodon.Online as Fediverse 101

Mastodon was conceived by Eugen Rochko as a quieter alternative to the excesses of the algorithmic commercial social space. It respects privacy by default and doesn’t push us to find our friends and connect with commercial accounts. Users have control over the way algorithms serve their information to produce a saner social experience for readers.

But precisely because Mastodon’s aims have been so limited in scope, it has failed to become a new internet nerve center, a new digital commons, and a new social layer that offers mainstream users a complete suite of online experiences based on Fediverse values rather than on surveillance capitalist business objectives.

My vision for Mastodon.Online is that it offers diverse content types with opinionated user experiences optimized for the kinds of core activities that people now do on Meta’s social networks. Not just mimicking the UX of “Twitter but open”, it could offer long-form blogging and business/brand homepages (similar to WordPress or Tumblr), free classified ads (like Facebook Marketplace or Craigslist), a coherent Pixelfed experience for photo sharing, a central news reader feed (like RSS or Facebook) for customizing one’s incoming content streams, and a topic discussion forum like Lemmy/PieFed/Reddit. Finally, it should be a coherent and simple place to manage one’s online identities – similar to a cross between a password manager and OpenID, that helps mainstream users keep their internet accounts under their own private control.

The experience should be opinionated to reduce the complexity of joining the Fediverse, and should innovate on simplifying complex technical tasks. It should serve as a default trusted option where non-technical users can follow internet safety best practices without having to consciously understand what’s happening under the hood. Where Meta helped us connect by eroding our privacy, there is an opportunity to help us connect at a similar baseline level just by offering a coherent suite of interoperable tools that fill the most common social internet use cases with minimal friction and complexity.

There are current projects that are well on their way to offering a cohesive suite of products that also protect user privacy, such as Proton Apps, Vivaldi.net, and Cozy Cloud. Most of these are centered in EU countries who are (rightly) growing nervous at the increasingly erratic and predatory behavior of America’s tech giants and political leadership, and are looking for a saner way forward. These can either serve as allies in the mission of moving towards an open, connected, Fediverse-friendly future tech ecosystem, or at least a forward-looking vision for what’s possible.

Uniting Disparate Fediverse Projects

As is always the problem with open projects, they tend to squabble and fuss about joining forces and working in the same direction. There is a vocal faction of Mastodon users who like Mastodon just like it is and don’t want to see it any different. This is the same with many of the open projects mentioned. Uniting these disparate tools into a coherent experience may involve forking existing tools so they play nice with one another and behave consistently across the suite of experiences. While Mastodon.Social may continue to lead the way as a pure example of what Mastodon is and should be, Mastodon.Online can flex and stretch to become a maximally accommodating home site for mainstream social users who want a simple, non-evil, digital home.

While some Mastodon faithful might fear that changing the platform is the first steps on a slippery slope towards diluting the Fediverse (similar to the controversy around federating with Threads and other commercial Fediverse nodes), the opportunity is to give the average internet user a real viable option for deleting Facebook, Instagram, X, and other predatory commercial social sites in favor of a pro-user social connective layer of the Internet.

What is the Authoring Innovation Meeting?

The Authoring Innovation Meeting is a biweekly gathering where instructional designers from across the organization come together to share expertise. It serves as a forum for:

• Show-and-Tell: Demonstrating new authoring tools, multimedia techniques, and design hacks.
• Collaborative Problem Solving: Thinking together about complex design challenges.
• Toolbox Management: Making collective decisions about the team’s shared set of authoring tools and workflows.

How It Started: From Stand-up to Scale

The AIM didn’t start with a grand mandate. It began as a practical necessity when our team was transitioning from one eLearning authoring platform to another. We needed a space to learn the new system together, raise issues, and share early wins.

As the platform transition matured, the scope of the meeting naturally expanded. What was once a narrow discussion on tool features grew to include the broader artistry of learning design—multimedia, game development, and experimental pedagogy. Recognizing the value of these discussions, we began recording the meetings and sharing them globally, turning a team meeting into a company-wide resource.

Anatomy of a Typical Meeting Agenda

While the tone is democratic and informal, the AIM follows a structure that ensures value for every participant:

1. Platform Updates: A quick review of recent change notes from our primary authoring tools to keep everyone abreast of technical shifts.
2. The Show-and-Tell: The meat of the meeting, where designers demonstrate a specific technique or a finished course.
3. Open Forum: A dedicated time for “I’m stuck on X” questions, where the collective wisdom of the group is applied to a specific project.
4. Recording & Archiving: Every meeting is recorded via Zoom. These archives are indexed and published on our internal wiki, allowing for “just-in-time” learning long after the meeting ends.

Reactions from Teammates and Management

The internal reception has been overwhelmingly positive. One of our senior designers, Chris Melton, once remarked: “This is my favorite meeting of the week.”

Management has also seen the value, particularly in how it accelerates the onboarding of new talent. During feedback sessions, new hires have described our onboarding program—with AIM as a central pillar—as “the best onboarding experience of any job they’ve ever had.” Managers appreciate the “time-to-productivity” gains achieved when new designers can learn directly from veterans in a low-stakes, high-support environment.

Organizational Effects: Performance and Job Satisfaction

The impact of the Authoring Innovation Meeting goes beyond simple knowledge sharing; it has fundamentally altered our organizational culture:

Unsiloing Expertise

By sharing knowledge freely, we’ve broken down the “disparate islands” of designers who historically didn’t interact. Expertise is no longer trapped in individual brains but is part of our collective intelligence.

Creating “Digital Exhaust”

The practice of recording and indexing these meetings creates what we call “digital exhaust”—repurposing the work we’re doing anyway into an enduring resource. These archives have helped raise our team’s profile within the company as a center of design excellence.

Improved Job Satisfaction

The AIM provides a “joyous social event” every two weeks. It alleviates the isolation often felt in corporate roles and fosters a sense of belonging and pride in our craft. This social connection, combined with the continuous growth in skills, has significant positive effects on long-term job satisfaction and retention.

The Authoring Innovation Meeting stands as proof that when you give creative professionals the space to think together, the benefits ripple across the entire organization.

Long Tail Learning — Not Back On It, Still On It

In the late 2000s this philosophy led me to have all my students set up their own personal blog site where all their assignments would be posted. I used an RSS reader to subscribe to all their blogs so I could review their work (essentially forming a DIY LMS system). Most important about this configuration, though, was that the students were in control of the data they produced in my class. I made it clear to them that they were creating their work artifacts not for me their teacher, but for themselves as lifelong learners. I encouraged them to look at their work as something they can continue to refine and improve in the years after my class was over.

I was (and am) a fan of Gardner Campbell’s idea that students should cultivate their own personal cyberinfrastructure — a set of tech tools under their own personal control (not the control of their school, their employer, or a commercial 3rd party) that they can continue to grow in across the span of their academic and work careers.

I was (and am) a disciple of Jim Groom’s Domain of One’s Own project, in which learners get their own web domain and start to build their own personal cyberinfrastructure using free and open tools that let them see how their tools and data interact on a privately controlled web server with their own domain name.

I was (and am) a fan of Stephen Downes’ and George Siemens’ learning theory of connectivism which emphasizes that learners in the age of the internet should not only stuff their brains with information but also connect with valuable digital sources of information they can continue to refer back to over time.

I was (and am) a follower of Mike Caulfield’s model of The Garden and the Stream and Ward Cunningham’s federated wiki movement, which emphasizes the continuous cultivation of a personal knowledge “garden” in digital form — an external “second brain” that both contains our best learnings but also reveals it back to ourselves and encourages us to update it as we learn.

I have been (and still am) a practitioner and evangelist of “second brain” personal knowledge management tools. Though my tools and techniques have evolved over time from Evernote to now UpNote and Obsidian, I fastidiously maintain both a messy “catch all” of notes and documents I want to have with me all the time, as well as a more organized notebook of my writings, learnings, and ideas that I am actively developing around my many core areas of interest.

My blog Ted Curran.net serves as my public facing document of my best thinking on teaching and learning with technology, and I periodically refer back to it and update it as new things occur to me. It contains content dating back to around 2008 when I started my first teaching blog, and it reflects my growth as a teacher and instructional designer over all those years.

I also won a Brandon Hall Award in 2021 for designing a corporate onboarding experience based on a shared knowledge base that learners are guided through in their first weeks on the job, but that grows into a shared repository of best practices that even career veterans can benefit from, documenting and disseminating their best practices while de-siloing the best thinking happening across the team.

Suffice it to say, I think there are very valuable benefits that come from treating learning as an ongoing process of discovering and saving valuable information so that we can not only remember it but continue to refer back to it and develop ideas over the years and decades that follow the moment we learned it. Too often teaching and learning is assumed to have a discrete start and end, and long tail learning posits that that initial moment is just the beginning of a much longer process of remembering, thinking, and building our knowledge.

Long Tail Learning in the Age of AI

So what does having a personal second brain learning garden look like in the age of artificial intelligence? Especially now, in the time that we’ve seen big tech’s surveillance capitalism model erode the very quality of the information we can find online, it’s even more important to have a private repository of knowledge you trust under your own personal control. Where commercial search engines and AI chatbots might show us enshittified results to further their corporate masters’ business objectives, we can take matters into our own hands and maintain a personal knowledge garden that serves as a personal assistant based only on information we feed it ourselves.

Google’s NotebookLM presents a model of what a personal AI notebook can be. Rather than being trained on a vast dataset of random sites across the web, you can train a model based only on your own writing and personal documents that you’ve saved in a specific repository. This enables you to reap the benefits of AI chatbots while limiting the focus to only the documents you’re interested in.

Now of course, not everyone trusts Google and other big tech companies to protect their privacy — but you don’t have to. The same type of toolset can be cobbled together with a local LLM model and a set of markdown-formatted text files on your local machine. this means that you can experience the same power available in something like NotebookLM with free, open source tools.

Of course, what’s really important is keeping a future-proof record of your learning that you can refer back to even years after you found it, from whichever device you’re on. The individual tools are just a way to get there.

That can be hard for teachers and learning designers to internalize, because many of us were the kind of students who did well in school. We liked learning, tolerated the process, and were willing to put in extra effort to master concepts. When you’ve had success inside the system, it’s easy to design instruction that assumes learners will behave the way you did.

But plenty of learners don’t.

My own monkey mind shows up as what I think of as my “inner punk kid.” As a student, I was impatient and distractible. I often tried to guess answers instead of doing all the reading. I completed work half-heartedly when I couldn’t see the point. I’ve always had enough punk-rock instinct to ignore orthodoxy and find my own messy path to understanding—even when that meant resisting the “correct” way my teachers expected me to learn.

The funny thing is: that same DIY attitude has been a strength for my learning as an adult. It’s helped my metacognition—my ability to notice what I don’t know, seek what I need, and fill gaps without being spoon-fed. I’m stubborn enough to resist being told exactly how I’m “supposed” to learn, which means I’m constantly building my own approach. My current manager told me I’m one of the best lifelong learners she’s ever met—because of that punk attitude, not in spite of it.

On educator teams, though, I often work alongside people who really were that “good student” archetype: the Hermione Granger types — organized, compliant, thorough. The problem is when we build systems that quietly require Hermione-like behavior from everyone, even though our learner population includes plenty of Harrys, Rons, and Nevilles: capable learners who will take a more chaotic route to competence.

That’s where our measurements can go wrong.

There’s a real disconnect between the steps we (as teachers) imagine learners should follow and what they actually do. The map is not the territory. UX research tells us that many people don’t patiently read every word on a page. They skim. They search. They look for the key detail that lets them move forward with minimal wasted time. This is not laziness—it’s time management, and it should not be punished. 

Learners do the same thing in training. They make rational choices about how to spend limited time. They cut corners—sometimes judiciously—in order to balance learning with their other responsibilities.

As a designer, I’ve learned to listen to that monkey mind—my inner “punk kid” who wants the answer without jumping through unnecessary hoops. That voice is valuable data. It points to friction, unclear value, and needless compliance disguised as rigor. When I feel that resistance, it’s often a sign that I’m designing for teacher-pleasing behavior rather than actual learning.

Civil Disobedience

I’m going through a mandatory training at work right now, and I can feel the tension between being a “good little student” and simply getting the information I actually need. The course design makes that tension worse: it treats compliance behaviors as the goal, not learning.

I’m required to read every word of documentation about a new LMS system and answer questions chosen at random from the full text. The text is extremely redundant—spelling out every detail of every feature, even when the same functionality is repeated in different modules, there’s another nearly-identical passage of documentation about it that I’m required to read as well. Much of this should ideally be “just in time” learning that you look up when you need, but I’m being required to learn it all “just in case” before I even get access to work in the system. 

Everything is tracked and enforced: how long I watch each video, how long I spend in each PDF, how many attempts I take on assessments. In a setup like this, the LMS becomes less a learning tool and more a mechanism for surveillance and compliance enforcement — especially when the training wasn’t designed considering real constraints, real work, and real learner needs.

Measure Learning, Not “Student-ing”

I’ve seen versions of this “teacher fallacy” since my classroom days: using grades and rules to force the learning behaviors teachers wishstudents would perform. Thankfully, I had mentor teachers who helped me separate “teacher-pleasing behaviors” from actual learning. Some learners get it right away; others benefit from repetition, structure, and scaffolding. The danger is using one narrow set of metrics to evaluate everyone—rewarding certain behaviors and punishing others, whether or not those behaviors correlate with understanding.

If we want learning systems to support real learning, our measures—and our courseware—need to leave room for diverse paths to meaning-making. Otherwise, we’ll keep designing for Hermione and calling everyone else “unmotivated,” when they’re often just navigating the system the way humans naturally do.

Design Recommendations: Building for Monkey Mind

Here are some principles I try to follow when designing learning experiences that respect learners’ autonomy, time, and varied approaches:

1. Assess outcomes, not process.

If a learner can demonstrate competence without watching every video or reading every page, that’s evidence of prior knowledge or efficient learning—not cheating. Design assessments that measure understanding and application, not compliance with a prescribed path.

2. Make content skimmable and searchable.

Use headings, bullet points, clear labels, and a logical information architecture. If your content requires linear reading to make sense, that’s a design problem. Learners will skim anyway—make it easy for them to find what they need.

3. Distinguish “just in time” from “just in case” learning.

Not everything needs to be memorized up front. Reference materials, tool documentation, and procedural details are often better delivered at the point of need. Reserve structured learning for concepts, frameworks, and skills that require practice and integration.

4. Avoid redundancy and filler.

If you’re repeating the same information across modules “for reinforcement,” ask whether learners actually need that repetition or whether you’re padding the course. Redundancy frustrates efficient learners and wastes everyone’s time.

5. Don’t weaponize analytics.

Time-on-task, page views, and video completion rates tell you about behavior, not learning. Using these metrics to enforce compliance creates a surveillance culture and punishes learners who already know the material or learn quickly. If you track engagement data, use it for design iteration—not gatekeeping.

6. Offer multiple pathways to mastery.

Some learners benefit from worked examples; others prefer exploratory problem-solving. Some want detailed explanations; others want the TL;DR and a chance to try it themselves. When possible, let learners choose their route and demonstrate competence in ways that align with their strengths.

7. Test with your own monkey mind.

Before you publish, put yourself in the shoes of a learner who is busy, skeptical, and unwilling to tolerate busywork. If you would be tempted to skip sections or game the system, that’s valuable feedback. Redesign until the learning experience respects the learner’s intelligence and time.

Listening to monkey mind isn’t about lowering standards—it’s about designing systems that measure what matters and don’t mistake compliance for competence. When we make space for diverse learning paths, we’re not being permissive. We’re being realistic about how humans actually learn.

I started reading up about roguelites and roguelikes and learned that punishing difficulty is a common feature in this genre of games. Fans of these two closely-related genres love the high level of challenge paired with the “permadeath” mechanic where every time you die in the game, you start over at the beginning with all your progress wiped out. This seems like it would be a huge de-motivator for players, but instead roguelike games are also famous for being infinitely replayable, for actually motivating players to persist and try again, with players logging hundreds or thousands of hours into the same game, happily playing the same level over and over. Why? 

Producing motivation in learners is the holy grail of gamification in education and instructional design, and getting learners to engage with truly challenging tasks and persist through failure to try again is similarly critical to supporting struggling learners. These are also things that elearning course modules do worse than human teachers in real-time interactions with learners. So how does the programming of these games succeed where most eLearning fails? 

I wanted to understand the learning psychology behind roguelike games to see if we can harness some of what makes them so compelling in our learning design work. This blog post is an attempt to synthesize the best thinking about motivation, persistence, and playability in gaming crossed with the psychology of learning. 

Good design, not gamification

For the record, I remain proudly skeptical of gamification in education. Though I am an enthusiastic gamer myself, I don’t believe that games are ideally suited for teaching skills and content within the constraints of formal education institutions or workplace training programs. 

Why? Good games are often measured in time spent playing — the best thing you can say about a game is that it’s “addictive”, meaning that the player becomes highly motivated to play the game over and over again. 

Teaching and learning in schools and workplaces must be time-efficient. We have more content to teach than we have time to do it in, so efficiency is critical in these contexts. Getting learners “addicted” to playing a repetitive game is not an efficient use of their time, especially if there’s little relationship between “time on task” and learning outcomes achieved. This is why gamified courses look mostly like ordinary courses with some gimmicky game mechanics sprinkled on top – the goal of teaching is to share knowledge (hopefully efficiently for everyone’s time), not just to enthrall and entertain for endless hours the way games can.

So this post is not about “what if we made roguelites educational?” It’s more about “what design inspiration can we take from roguelites to make eLearning experiences feel as urgent, as relevant, as self-directed as roguelites?” This is a fruitful path that can inform the work we create. Just don’t call it gamification.

What characterizes a roguelike/ roguelite game?

A note on terminology: There are real technical differences between roguelites and roguelikes but they’re more alike than different. What’s more, the wide variety of games in these two genres actively blur the distinctions between them, so it’s hard to talk about them as two distinct genres. We’ll go into the differences below, but this article may freely use the terms “roguelike” and “roguelite” interchangeably. I’ll also use the term “roguish” for when we’re talking about these two closely-related genres and the characteristics they both share. 

Roguelikes are games that share a specific set of characteristics with the seminal 1980 game Rogue, while Roguelites follow only a subset of those rules while embracing a wider variety of genres and gameplay mechanics. 

Roguish games share a few key features: 

• high difficulty – roguelikes set the challenge bar high, expect players to die frequently, learn from their mistakes, and progress through the game only after having mastered key gameplay actions
• procedural generation and replayability – game levels are generated dynamically by an algorithm, not designed by a human, so the game changes in important ways every time the player replays the game. This builds a sense of unpredictability into the game, as even when you’re playing a level you’ve been to before, it’s never quite the same as it was before. You can’t memorize specific things to look for, so you’re encouraged to understand the game mechanics deeply and use that knowledge improvisationally to overcome unpredictable challenges you encounter.
• “permadeath” – In roguelikes, when the character dies, all progress is wiped out and the player must start again from the beginning without any of the enhancements they’ve collected along the way. Roguelites soften the difficulty somewhat by letting players keep certain enhancements that persist through death and help them on future runs. This makes roguelites more forgiving, allowing less skilled players to progress by trying multiple times and accumulating advantages that will help them in later runs, while roguelike players can only progress by developing their own skillset to overcome the game’s high difficulty.
• exploration and resource management – the character finds items along the way to enhance their character and adjust key conditions of the gameplay, such as health, damage output, and more. These items can confer advantages or even disadvantages to change the game in interesting ways moving forward. The player has an unusually high degree of freedom to choose, discard, and combine these resources creatively to change key conditions of the game. It’s even possible to “break” the game difficulty by combining upgrades in such a way that the player becomes overpowered and can win with ease.

Before we go too far — are these games I might have heard of?

I have played and loved video games for most of my (nearly 50 year) life — I’ve followed the progression from Atari through Nintendo, PCs to consoles, from the 80s to today. In all that time I had heard of but never investigated the genre of games called rogue*likes* (and the closely related) rogue*lites*. 

There are over 8000 games on the Steam Store tagged as roguelikes/ roguelites but the ones mainstream gamers might’ve heard of include Hades, Returnal, Balatro, Dead Cells, or Slay the Spire. There are also less well-known “classics” of the genre like  The Binding of Isaac, Enter the Gungeon, Rogue Legacy, Spelunky, and  Vampire Survivors, that are spoken of with respect and awe in roguelike forums but might not be familiar to mainstream audiences. 

Roguelikes are having something of a moment in 2025, and mega-budget game franchises like Elden Ring, God of War, and  Hitman have released spin-off roguelike games set in the universe of their blockbuster game worlds. Still the vast majority of roguelikes/roguelites are from smaller independent developers with smaller budgets and simpler graphics who manage to keep players enthralled with their compelling variations on the core roguish gameplay features so beloved by this genre’s fans. Poking around in the spaces roguelike lovers dwell online opens up a small and passionate world of enthusiasts who love these games more than any other. 

Roguelites and Motivation

It seems paradoxical that roguelike games, expressly designed to be extremely difficult and kill you off right away, would actually attract a loyal following of players willing to play them over and over again for hundreds of hours, to play the game multiple different ways in an attempt to progress. Do these games just attract masochists? Or do they subtly condition the player to embrace the cycle of trying, failing, and trying again to “come back stronger” ?

I found myself being “shaped” or “coached” by Returnal to get over my natural aversion to dying in video games, to see the deaths as a part of the process, and to start to think forward beyond each individual death to the long-game strategy that would get me to my goal. Each run becomes an experiment in changing little things about your gameplay, making different choices each time, and seeing how those affect your fortunes. If your experiment fails, it gives you a clearer idea of how you can overcome it next time. 

This is consistent with my belief as an educator that learners thrive when they have the freedom to try, fail, reflect, and try again. Too often in learning/training, learners are given one chance within a narrow time window to prove their understanding of new material. It is inherently motivating for a learner/player to know that they can experiment in ways that make sense to them, monitoring feedback, and adjusting performance to get the outcomes they want. All it takes is time on task, sustained attention, and compelling activities that reward multiple attempts. 

Building motivating experiences, then, requires us to create safe spaces where learners are challenged to master new skills and knowledge above their currently ability, given timely feedback of the success of their attempts, and multiple tries to achieve the outcomes they want. This is the opposite of “dumbing down” curriculum — as in roguelikes, the level of challenge stays high, and the learner progresses through skill and perseverence. The courseware must be optimized to support this virtuous cycle. 

Roguelites and Player Progression

The thing that separates roguelites from roguelikes is that they grant the player lasting advantages the more times they try and fail, making initial runs harder but later runs successively easier as they progress. True roguelikes don’t do this — they start the difficulty high and the only way a player can advance is by increasing their skill. 

The second roguelite I played, Hades, also structures its experience so that advantages are meted out gradually after each player death brings them back “home” to the House of Hades where they can regroup, re-tool, and set out again. Interacting with various Gods of Olympus so they’ll bless your run creates vastly different patterns of play that change the experience every time. These boons from the Olympian gods grow more powerful over repeated attempts, and the player can leverage these to advance more effectively through the game. 

What do roguelite players love about the genre?

I asked in a roguelite forum what seasoned players like about games in this genre, and their responses rang true with my experience. The sections below are a synthesis of those answers and my interpretation. 

## Roguelites and repetition

These games inspired me to go back and watch Groundhog Day, the classic comedy about a man cursed to re-live the same day over and over. The comedy turns existential after Bill Murray’s character has lived the same day hundreds, maybe thousands of times, learning everything there is to know about the tiny Pennsylvania town he’s trapped in and all of its inhabitants: 

A similar scene in Tom Cruise’s Edge of Tomorrow depicts a conversation he’s had hundreds of times, trying and failing to convince a tough army general to help him until finally he succeeds:

In these stories, the hero prevails simply because they have tried every possible path forward, made every mistake, and come back again to finally prevail, finding the only way to truly progress. 

The Curse of Infinity

Of course, human learners do not have infinite patience for infinite tries, infinite failures, with no guarantee that they will succeed. The characters in these stories feel cursed with the power to relive the same moment over again. If they had their way, they would move forward, or even give up — anything to escape the endless cycle of trying and failing. And yet, it is precisely because they persevere that they eventually succeed. 

This infinite try-fail-try-succeed loop is how AI machine learning works. Researchers use computers’ ability to perform the same action millions of times, rewarding it for achieving desirable outcomes. AIs learn to recognize the difference between an apple and a cat by looking at millions of pictures of apples and cats until they’ve seen all the variations and can differentiate them accurately. This is why CAPTCHA tests ask us humans to tell the difference between a bus and a crosswalk — because those are the ambiguous edge-case images that the computer needs our help telling apart. 

Of course we humans have far less patience for repetitive tasks that machines or even enthusiastic gamers do — we like to do things once and be done. Often if something isn’t done right the first time, teachers often will refuse to grade a second try (that’s “cheating”, isn’t it?) and learners will refuse to try again for fear of wasting effort and suffering the disappointment and shame of failure a second time. 

I have successfully redesigned training programs into experiences that embrace the “long tail” – to be woven into learners’ and workers’ schedules over time rather than force fed once and never again. 

I asked in a rogueliite Reddit forum: 

How would you explain the extreme replayability of roguelite games? What motivates you to put hundreds of hours into them? What specific mechanics made the difference between the ones you stick with and the ones you drop?

I got some very thoughtful and generous responses from members of this passionate subculture of gamers, telling me what about roguelites compels them to spend hundreds of hours replaying a single game.

Themes include:

Trusting the player with greater autonomy and decision making

As opposed to conventional games, roguelikes give the player an unusually high degree of control over their own experience. The game exposes its core mechanics to the player through frequent decision points, and allows them to change key conditions of the way the game works so they can experiment with different “builds” to better match their play style and increase their chances of success on their run. 

For example, in Returnal your character encounters alien parasites that give you both an enhancement (like automatically regaining some health after every enemy defeated) AND a penalty (like taking damage every time you pick up an item). The player makes a risk/reward judgement to see whether the benefit is worth the drawback, and they have to alter their play style going forward to avoid negative consequences. There are multiple overlapping systems like this in the game, resulting in each run feeling very different as your customizations interact in varied ways. 

A firm-but-gentle place to try-fail-try-succeed

As devastating as it can feel to die over and over again in a roguish game, players appreciate the feeling of transparency about the reasons why they failed — it’s valuable feedback so they can adjust future attempts and overcome past mistakes. The experience of playing these games feels like the game is training you, coaching you, in the skills you need to be successful in it. The level of difficulty remains high, but the criteria for success are also highly visible to the player. 

Asking the player “how far will you get” through a challenging, endless task?

Many roguish games are truly endless due to their algorithmic level generation, and yet players continue playing to see how far they can progress through the endless levels of the game. Early roguelikes used the marketing message “how far will you get” to the game’s end goal. There is a thrill in knowing clearly what your end goal looks like and being shown clearly how far along that path your current skill level can bring you. 

In Hades, your protagonist is the prince of the underworld, son of Hades, in this Greek-mythology inspired game. His goal is battle his way out of the land of the dead and up to the surface to escape his father’s influence (mostly out of a sense of relatable teenage rebellion). There are five levels of the underworld to battle through, and actually making it successfully is nearly impossible to do on your first try. Again, there are lots of game criteria you can experiment with to see which will give you the edge you need to successfully defeat your grumpy dad and reach the land of the living. As you play through, you realize that the journey is as important and enjoyable as the destination.

Creating a “sandbox” or laboratory full of toys and tools for the user to combine creatively, rather than a linear one-size-fits-all experience

1. Sufficient number of meaningful mechanics and systems
2. Randomization across a sufficient number of those mechanics/systems
3. Sufficiently distinct starting states
4. Sufficiently meaningful control over the game state
5. Sufficiently enjoyable core gameplay loop
6. Sufficient depth (this could be in meta progression, character mechanics, game scaling, etc) to facilitate game runs feeling purposeful
7. Sufficient brevity in runs to feel accessible and not exhausting

• Janube

A rigid set of rules perceptible to the player, while giving them freedom and tools to follow (or even break) those rules

The upgrade mechanics in roguelike games are often highly detailed, spelling out what specific upgrades or penalties will result from using each in gameplay. Roguelike players love deep upgrade systems full of discoverable synergies — where upgrades can stack on each other and create multiplier effects that dramatically affect gameplay. The most beloved roguelikes put an outsized effort into making these game mechanics discoverable and comprehensible to the player. Similarly, great eLearning should make the intended outcomes of instruction transparent to the learner, and all elements of the course design should reinforce and support that intended learning outcome. 

Progress is continuous (endless) but game sessions can be short

Roguelike games are called “runs”, and they are expected to be short enough that a player can finish one in a single half-hour sitting. This is great for busy players who want to get a sense of accomplishment amidst other responsibilities, since you can play, make progress, and die within a short time window. Roguelite players like being able to see the results of their forward progression as they unlock key game upgrades that will improve their fortunes in future runs. Having progress be visible and transparent to the end user is great design practice for eLearning as well, so learners can budget their time and feel a sense of accomplishment after working for a reasonable stretch of time. 

Zone of Proximal Development — letting the player feel a strong degree of challenge while giving them the support they need (or at least the transparency they need) to reach a very ambitious goal.

“It’s the fact they respect u as a gamer. If you want easy go for a walk or play ubisoft games or story games. If you want to feel challenged and rewarded in success the best games give u that” – JimBoonie69

Permadeath is a motivator, not a de-motivator

Yes, that feeling of “how far can you get” was constantly with me in Returnal and it really raises the emotional stakes. Like you say – you realize how much work it was to get there and how fragile your progress is at that point. At that point, though, I still had options. I realized I could progress in a variety of different ways – that I didn’t just have to follow the linear progression that the devs had laid out, as is so common in other AAA games I’ve played. That feeling of agency, of being in control is extremely motivating. I went in thinking “permadeath” would be just plain frustrating, but instead it challenged me to form a strategy that included dying on my way to my goal. Dying was not the simple showstopper it is in most games – It was part of the journey and just another challenge to think through. That’s incredibly empowering. –Me

Roguelite game mechanics in teacher-speak

Permadeath

{.aligncenter}

Even though it sounds a lot less “metal”, we educators might call this a “try-fail-try again-succeed loop“. It’s good teaching practice to create a safe space where learners are not penalized harshly for trying, even if they fail at first. The ideal approach is for learners to be able to try, possibly fail, receive timely feedback about their performance, and incorporate that feedback into subsequent attempts until they achieve the desired learning outcome. 

At first look, permadeath sounds like it would hurt learner motivation, except for one thing — you get to try again. Think about all the times in teaching and learning where you only get one chance to get a good grade, and never get another opportunity to show what you know. The weekly pop quizzes we endured where we weren’t allowed to know ahead of time what to study, and our permanent educational records were tarnished by every missed question on a high stakes exam. A game where you play through, see what’s required, die, and start over until you succeed is positively humane by comparison. 

By structuring games into short runs where players can freely experiment with different play styles, see the effects of the changes they make in the form of instant game feedback, and adjust their approach on future tries means they eventually practice their way to better performance organically. 

The takeaway to educators is to design experiences where learners can plunge right ahead, see what’s required of them (maybe with a low-stakes, ungraded assessment), receive instant feedback and reteaching about the questions they missed, and get the chance to try again until their graded performance actually meets or exceeds the desired learning outcomes. 

This may require you to invest more of your development time and effort into creating great interactive assessment tools with specific feedback rather than spending all your time making a beautiful slide presentation that learners will see once and never again. 

High Difficulty, Transparent Success Criteria, and Player Progression

{.aligncenter}

These games start hard, stay hard, but they give you multiple tries and they also gradually introduce “power ups” designed to help the player to eventually achieve success. In education this sounds very similar to keeping the learner in the zone of proximal development – a perfect “goldilocks” state of challenge calibrated to the learner’s starting abilities. Tasks in the ZPD start off difficult (because they are above the learner’s beginning level of skill) but are within the range that the learner could reach within the time allotted with appropriate effort and instructional guidance/support towards that goal.

Too often, it’s not clear to the learner what successful performance looks like, so they don’t know how to demonstrate success. Often teachers and trainers deliver training without specifying clear learning objectives first.

Pre- and post- testing

Offering a diagnostic exam at the beginning of training and a summative exam at the end (based on the same learning objectives) can help document the growth learners made from the beginning to the end of that experience. It can also make transparent to the learner exactly what they’ll be expected to know and be able to do as a result of training. Roguelikes function this way, since the initial experience is usually brutally, demoralizingly difficult. Players get a visceral sense of their inexperience with the game mechanics, but gradually that sense is replaced by an authentic satisfaction and confidence that they have actually grown better at the game. We use pre- and post- testing in our eLearning courses to document learner growth over the duration of a course, and making this transparent to the learner also shows them directly that they have improved as a result of instruction. 

References:

What Are Roguelites? A Deep Dive | HackerNoon

Roguelike vs. Roguelite Games: The Differences, Explained – whatNerd

The Typical Approach to Workplace Training Doesn’t Work

Whenever there is a new corporate initiative, somebody calls for a big training “rollout” or “kickoff” event — a big splashy mandatory training to signal the beginning of this new way of doing things. It usually involves hours of mandatory eLearning courses or live training events to debut at the kickoff date — even when the details of the project are still being sketchily cobbled together by the teams tasked with developing them to align with that date. Too often, we instructional designers are given a high-level training topic and a due date — and that’s it. 

“Make a training about The New Widget Production Method. It needs to happen in two one-hour sessions and it’s due in four weeks. Go.”

Then what normally happens is someone takes a stab at writing two one-hour PowerPoints on the finer details of The New Widget Production Method, they send it to the instructional design team to turn it into two one-hour Storyline courses, and everybody is forced to spend two hours of their lives clicking through it. Once it’s clicked through, the course gets marked “complete” in the LMS and is never viewed again. The finer points of the plan were not remembered the first time, so they’re not recalled later, and have very little effect on the way anybody does anything going forward. The details of this new way of doing things are not written out in an greater detail than these introductory PowerPoints, so the next time we do this again there will be no detailed record of the way we were doing it before. Rinse and repeat for the next big kickoff next year. 

Design for Forgetting

There’s a false assumption at the center of this ineffective approach — they assume that learners will remember everything they’re trained on in perfect fidelity and know instantly how to change their own work processes in light of this new information. Unfortunately human brains don’t work that way. 

“Education is the Kindling of a Flame, Not the Filling of a Vessel” — Socrates

At best, you have a very short time during a conventional training session, and you can leave a maximum of 3-5 lasting mental takeaways in your learners’ minds. Everything else is forgotten. Those 3-5 bits of information need to be something they’ll remember long after the training is over that can help them answer questions when they arise. 

What’s more — training experiences END. That great eLearning course we built for you gets wiped out of your learners’ LMS “to do list” the moment they have completed the SCORM package. At that point they can no longer refer back to that information, and have to rely on their incomplete memory of the hour they spent clicking through training. What they need is an ever-present way to summon up correct answers at the moment they need them, even if that moment comes months or years after the big splashy kickoff event. 

How can we do better?

My approach is to create an evergreen knowledge base full of complete, detailed answers to the most common questions our learners will encounter in their daily work. Then use the two hours of mandatory training to show them what’s there and how to use it when they need it. Optimize that knowledge base for searching and easy finding — tie it into your work AI chatbot if you like. Make it easy for learners to search it whenever they need a definitive answer on how the company expects them to do something. 

Main Features of the Design for Forgetting Approach

Learner-centered pedagogy

• A learner-centered approach rather than “information-centric” approach. You need to understand and acknowledge how do they do things now, how will they have to do things differently in the future, and what do they need to know NOW vs. LATER to be successful with the new procedures? Any training you give them must use their precious time with maximum efficiency to produce those desired changes in behavior. 

The Vault: An evergreen knowledge base for ongoing reference

• The Vault: Build a “single source of truth” that contains all the answers they will need to do their job effectively. Something like a wiki knowledge base where all the details of the new way of working are spelled out and organized for ongoing reference. 

The Walking Path: A structured tour of The Vault for new users

• Have a guided path through that rich knowledge source for beginners. Design a scaffolded tour of the resources available to the learner, focusing first on those they will need to familiarize themselves with right away. 

Circling Back: An ongoing series of events and communications to remind and reinforce

• Have ongoing activities that draw attention and engagement about specific parts of the whole.
• These can be spaced repetition questions to reinforce existing learning.
• These can be periodic communications to draw attention to valuable “hidden gems” of information within the Vault, especially at the times when they’re most useful.
• These can be “barnraising” events — meetings where teams’ focus is to review and update existing materials or create new resources based on new learning.
• If you have Q&A sessions, respond with direct links to the Vault. Often people forget that “yes, we already have a best practice around this, and here it is”. Get everyone accustomed to going to the Vault for answers first before reinventing the wheel.
• If changes are needed to the way we do things, connect the new changes to the existing practices to make it easier for your workforce to change their practices. 

How do I know this works?

In 2020 I used this basic design philosophy to create the Instructional Designer Onboarding program to train new instructional designers joining our team. 

Rather than bombarding them with lectures and handouts in their first days, I created a Walking Path to structure their first two weeks on the job. It was a checklist of people to meet, platforms to log into, training modules to complete. It existed within the Vault knowledge base, but it showed them only what they needed to know and do in the first few weeks. At the end of that intro period, we introduce the wider Vault — a Confluence wiki site with detailed, reader-friendly pages and articles explaining every tool we use, every procedure we follow, and our greatest work products from the last five years for inspiration. The entire wiki is searchable by keywords and tags, as well as having layout-based navigation for those who’d like to click around and see what’s there. In other words, it’s an incredibly dense resource of valuable information that’s optimized for easy access, quick searching, and timely updates from all team members. 

In 2021 I won a Brandon Hall Gold award — Best Unique or Innovative Learning and Development Program — for this onboarding program. 

Since that point, though, it has grown more valuable as a training and reference tool for our veteran instructional designers, not just new hires. It’s cited in our team’s yearly reviews as a consistently useful tool in our toolkit, and multiple of our new hires have called our onboarding experience the best they’ve ever had. 

The team has a twice-yearly wiki maintenance sprint where we divide up all the existing pages, analyze them for any needed updates, and make those updates. The light task of maintaining the wiki is a small price to pay for the value it brings to us in our daily work. 

Conclusion

The core idea at the heart of this approach is that training is not a one-and-done event. Events end. Work goes on. Instead, we need to surround our learners with a rich toolset of evergreen resources they can refer back to at key moments in their daily work, knowing they will get detailed relevant answers at the moment when they matter.

Don’t make people click through a two-hour training course they’ll never see again once they finish it. Rather, use the precious time you have with them to show them how you want them to work and give them all the tools they need to actually change their work habits in the ways required. It should be OK for your learners to forget… knowing that the answers will be ready at the moment they need them.