I haven't written about the One Laptop Per Child project before, since I've picked it as a failure in the making from the get-go. Now that it's wandering around gut shot and waiting to fall down, it may be worth some examination in quest of lessons learned. Unfortunately, many of those lessons are old ones, but some reinforcement at others' expense never hurts.
The project has its own five point mission statement. I'd take that statement, blend with actual behavior, and distill out three de facto design points:
- Provide networkable computing hardware at a rock bottom price, such that developing countries and their citizens can afford it.
- Create a visual user interface and storage metaphor that is an alternative to the consensus files & apps form handed down from PARC and the Mac. As a matter of cost reduction and principle, build this entirely on open source.
- Support a particular type of constructivist learning for children using the machines.
And, by the way, accomplish this with an organization of academics who had never shipped a product, nor tackled any project at scale. Not that I'd have funded a squad of successful Dell and Apple veterans to try it in their place. The problem is reality gaps and internal contradictions in the design statement itself. Those are the entrails worth examining, and I'll do it in more or less the same order as the design points.
Cheap Means Vanilla
Low unit costs come from scale in production. They come from letting others drive the risk out of components and overall design and integration. If you're Apple, with its brand differentiation and affluent customers, you can push out the design edge. If you want dirt cheap, clone everything you can, from chips to mechanicals to software load. And address the widest possible base of users, to keep up your volume. OLPC broke this one as soon as they added any design point beyond 'make it cheap'.
The Trailing Edge Moves As Fast As The Leading Edge
Buying, or donating, off the back end of the technology curve does no one a favor. Anyone who's tried to maintain a nonprofit office filled with yesterday's cast-offs provided by well-meaning donors has learned this one the hard way. As the adoption wave passes by, the ecology around a generation of technology collapses. Eventually they even shut down the fabs (which has NASA stockpiling 8088 chips to keep its systems running). With an anemic processor and software combination already unable to run today's multimedia, according to this review, the OLPC was off the trailing edge the day it shipped.
Never Build An Almost-PC
This is almost an immediate consequence of the Vanilla rule, but it's worth reinforcing. Building a box that looks like a standard product and arouses those expectations, but without fulfilling them, is a loser. So is walking away from the product and support ecology surrounding a de facto standard. There are no exemptions in the Third World. Remember that when the Marines took Baghdad, they found the kids there already playing LAN games. A good fraction of the photos sent in from the latest global disaster have a Toshiba or Lenovo or some such advert in the background. Being born in some benighted place may make someone poor, but not stupid, or even ignorant of what the outside market offers.
Easy For The User Means Harder For the Maker
Call this one the Apple Lesson. You haven't been properly humiliated until you've watched a video of a user trying to start up and learn a product you helped design or package. Getting something to be reasonably usable means repetitive testing with real people, all the way from paper prototypes to the golden master, of all new features no matter how seemingly trivial. I certainly agree that the files & folders version of the user interface needs fundamental overhaul. I've done some research in the area myself in the past, so it was nice to see some new ideas in the Sugar Ui design. But what's completely unconscionable is to ship someone's brain wave for a design without substantial user testing. Adding any design points beyond 'cheap' was already tempting the gods, but conflating what was properly a research project with a low-end, scale product is rank hubris. Even a first-time-out product manager knows better.
'Free' Doesn't Mean There's No Support Cost
Anyone who's "given" a computer to a relative knows this one. The poorer you are, you'd think the more TCO is going to matter. And apparently, the OLPC organization didn't even have a distribution channel worked out, let alone a coherent plan for supporting the machines, either teaching their use or maintaining them if they failed. Couple with a version 1.0 software load devoid of substantial user testing, and the outcome is rather predictable. A lot of the OLPCs that make it out there are going to end up as small counterparts to grandiose 'aid' projects that now lie in ruins because the recipients lacked the funds or skills to maintain them, or found them irrelevant to their actual needs.
Constructivist Education Is An Unproven Theory
Those of us in technology who are aware of this pedagogical hypothesis' role as the requirements statement that drove innovations at the early Xerox PARC, from object orientation to the windows style interface, tend to give it a free pass because of this origins story. We shouldn't. There haven't been clear cut results from even well funded trials, and the one prior attempt in a developing nation was apparently a mess. Putting this agenda at the core of a machine's design, then throwing it over the wall without training and hoping for big results is substituting prayer for knowledge. (And I say that as someone whose first West Coast gig was working on a Logo interpreter.) Not to mention that it's another conflation of a research project with a productization attempt.
The OLPC project seems to be realizing some of these problems, rather belatedly, and has shifted to offer Windows XP on its machine. Now all it has to do is execute better than the Taiwanese and Chinese notebook packagers. The best outcome at this point would be if the Microsofted OLPC demonstrated the existence of a viable market niche heretofore unknown, whereupon it can declare victory and shut down.
The saddest thing is that most of these mistakes and lessons aren't even close to original. And the whole project may have been based on an uncritical assumption that the form factor that won in the West is the one that will create the most benefits elsewhere. When even war torn Mogadishu is flooded with mobiles, you've got to wonder. Even when you're standing last in line, technology adoption is subject to path dependence.
All in all, the funds invested in OLPC would have been better spent on microloans, so poor parents could get their families connected via mobiles. Or on more basic needs like public health. This is another case where Bill Gates' head is screwed on better than the academics.