I’ve had the pleasure of leading the design of a number of projects that have had some impact. These include a mobile app a company could point to. Also a game that helped real kids. Even a context-sensitive performance support system that was worth a patent. Then, of course, are the projects that didn’t, for whatever reason, see the light of day.  So here are some reflections on a few projects that didn’t fly.

Back in the mid-90s, I was part of a government-sponsored initiative in online learning, and we were looking for a meaningful project. We made a connection to two folks with a small company that taught about communicating to the press. They could’ve come out with a book, but they wanted to do something more interesting. We collaborated on an online course on speaking to the media. I partnered with an experienced digital producer, and backstopped with a university-based media team. We had a comic skit writer, and cartoonists, to augment our resources. The result was technically sophisticated, educationally sound, and engaging both visually and in prose. It never flew, however, as we didn’t partner it with a viable business model. Which was reflective of the times.

Then, at the end of the 90’s, I was asked to lead a team developing an adaptive learning system. The charge was to help learners understand themselves as learners. I had a stellar team: software engineer, AI expert, psychometrician, learning science guru, visual designer, and an interface designer. The model was to do an initial profile, then present you with learning elements (concepts, examples, practice, etc) and update your model based on your performance. There was even a machine learning component to improve the models as we went along. We actually got a first draft up and running (10 elements in the student model), before ego and greed undermined and killed it. The lessons learned, of course, have continued to inform me, including, for instance, my calls for content systems.

Then, around the mid-2000s, I was given the task to devise a content model for a publisher.  They wanted to develop once and populate a variety of business products. Drawing on previous experience, I developed a robust model, which started from individual elements and supplemented and aggregated them in a systematic way. This also ended sadly. In this case, the software side never reached fruition.

There are lots of reasons good intentions can go awry.  In my case, it wasn’t going to be on a lack in the learning design ;). What I’ve learned, however, is that learning design isn’t the only element that matters. There’s vision, and execution, and partners, and more. All are ways in which things can go wrong. Yet, that doesn’t mean we shouldn’t try. It just means that we should, to the extent of our abilities, also try to ensure the success of the other comments. It’s worth exploring projects that didn’t fly so as to see how future ones might.

A couple of weeks ago, The Computer History Museum hosted a panel session on Smalltalk, which I watched via video. Alan Kay (who’s vision for the Dynabook drove Smalltalk) came in via recorded video. Dan Ingall (the technical guru) joined by live video link. Adele Goldberg (who documented and tested it), showed up live. John Markoff, well known Silicon Valley documenter, hosted. All to talk about Smalltalk. It prompted some small thoughts about Smalltalk.

I was a regular Byte magazine reader, back in the day. I had created my own major in Computer-Based Education, and was designing and programming educational computer games. I’d done academic research as part of the degree requirements, so I was aware of the work at Xerox PARC. (In fact, I flunked a job interview there because I didn’t know what ‘protocol analysis’ meant, though it turns out that’s what I’d been doing!) So, when the Byte issue in Aug 1981 on Smalltalk came out (I checked the date), I was enchanted.

Smalltalk is an object-oriented language that is dynamic, in that you can edit and immediately run it again; it’s not compiled. It was also reflective, in that make itself visible and operate on itself, like Lisp. In Smalltalk, you model your world in objects and they communicate by messages. It has windows, icons, and interactions comes from the mouse as much or more than by the keyboard. You can edit the objects while running and they change. While it wasn’t available to me, I was a fan of the concept.  (Machines running Smalltalk were what Steve Jobs saw on his PARC visit that led to the Lisa and then the Macintosh.)

It’s ironic that between then and when I ended up teaching in a school of computer science, I somehow lost that focus.  I’d gone to grad school to get a grounding in cognitive science in just such a place. After a post-doc looking at learner models, I ended up teaching interface design (and researching educational technology). Along the way I got involved in other issues, though I did get involved in HyperCard, which in many ways was Smalltalk Lite(tm ;).

In the talk, besides the enlightenment of the thinking behind it, there was also the practical aspects. While relatively lean, the language did take up memory and as a dynamic machine wasn’t blindingly fast. There apparently were also decisions about pricing and markets that were classic Xerox. Thus, while it was and is a fabulous modeling environment (still in use in a variety of markets), it didn’t take over the world.  When Steve Jobs built the NeXT computer, he took on the object-oriented model of Smalltalk, but used C as the core language for a variety of pragmatic reasons.

In the session, they talked about the vision of Seymour Papert and Logo, and how they wanted more. Alan Kay walked around with a cardboard model of what a Dynabook would look like, and people begged to buy one. Doug Englebart’s work also was an inspiration. It was a glorious flashback to the days when we dreamt bigger than our tech would support. These days, it seems, we’ve reversed that. I’ve heard that computing isn’t living up to the potential we have for digital technology to be an optimal augment for cognition, and I agree. We can do better, and should. So these are some small thoughts about Smalltalk. And get off my lawn!

I was asked about the latest techno-hype, bionic reading. At the same time, there’s a discussion happening about learning affordances of the metaverse. I realize my strategy is the same, which I learned many years ago (wish I could remember from whom!). The short version is, wait until the dust settles. Why? Let’s evaluate the late adopter strategy.

So, for anything new, there all-too-frequently seems to be a lot of flash. In my experience, a lot more than substance! That is, many things rise, and most fall. When things calm down after the initial exuberance, most simply disappear. There are myriad factors: acquisition and shut down by competitors, other elements fail despite a good premise, or even unexpected factors outside of control (e.g. a pandemic!). Of course, the usual suspect is that there’s no real there there!

I remember the hype over Second Life, and recognizing that the core elements were 3D and social. Yet, what we saw were slide presentations in a virtual world. Which was nonsensical. I’ve suggested before that you can infer the properties of new technologies, in many cases, by considering their cognitive affordances. I’ll await the meta-verse manifestation, but it seems to me to be the same, just more immersion. Still, lots of technical and cognitive overhead to make it worthwhile.

Similarly with bionic reading. There’s now  lots  of anecdotal suggestions that it’s better. That’s not the same, however, as a true experimental study. Individual experiences don’t always correlate with actual impact. There’re myriad reasons for this too, e.g. self-fulfilling prophecy, perception vs reality, etc. Still, I really want to have some more convergent evidence. Here it’s harder to do the affordances. Yes, it might support people who have difficulty reading, but might it interfere with others? How will we know?

On the basis of the above, however, I suggest waiting until something’s been around, and then if it persists, start investigating what the affordances might be. Many things have come and gone, and I’m glad I didn’t bite. I might then be late to a platform, but that’s OK. I still tend to get opportunities to innovate around ideas of application  after they’re established, because, well, that’s what I do ;). Affordances help, as does lateral thinking and having on tap  lots of mental models to spark ideas.

We’re too easily enchanted with the latest shiny object. No argument it’s worth experimenting with them, but don’t swallow the hype until you’ve either had your own data, or someone else’s. I reckon rushing in has a greater opportunity for loss than gain. Let those with needs, resources, and opportunity take the first cuts. There’s no need to bleed prematurely, there’ll be plenty of opportunities to need to tune and test again even once principles emerge. So that’s my take on the value of a ‘late adopter’ strategy. What’s yours?