skip to main bit
a man slumped on his desk, from 'The Sleep of Reason Produces



the XO abides

There’s a lot of people who have written off the OLPC: a pet project of Negroponte that lost momentum the moment the old staff got jettisoned and replaced with a CEO who said “the mission is to get the technology in the hands of as many children as possible” (in stark contrast to Negroponte’s original “It’s an education project, not a laptop project.”. I think the worst criticism has come from those on the Get-One-Give-One projects, who have regularly expressed disappointed with Sugar, the OLPC’s user interface, and the general state of the software.

What I find fascinating — and this isn’t just true of open source projects, though I think it’s more transparently noticeable — is what happens after that bump of enthusiasm fades. I’m beginning to believe that the great advantage of more open software (whether it’s open standards or open source), is the growing importance of slow-cooked software.

Firefox is a great example. The original Mozilla project, in a commercial context, should have been shuttered long before Firefox was developed: it pretty much was shuttered, by AOL, its major sponsor. But still development trundled along, fixing bugs, developing new enthusiasms, attracting young turks, accreting knowledgeable coders. And it slowly got better. Far too slowly for anyone to notice, until the Phoenix/Firefox team turned it all inside out.

I’d say the same is true of Unix in general. People say that those who do not understand Unix are condemned to repeat it badly, but in everyone else’s defence, Unix’s smug position is largely due to Unix folk making all the mistakes, and then veeery slowly backing out of them over a period of decades. When other, proprietary, systems go over a cliff, you frequently never see them again: and certainly the market gives them no time to learn from their mistakes. Who knows the lessons that PenPoint learnt? A lot of OS X’s benefits come from being a slow-cooked product: years of gently baking the Cocoa class library under the faint heat of NeXT’s limited audience.

With everyone’s attention off the OLPC, it nonetheless abides. The platform has shipped something like 400,000 laptops already. They’re getting ready to release a new update of the software, based on the latest version of Fedora, and with a whole bunch of UI and activity updates. Most G1G1 users won’t know all this, sadly, because they’re not a school, and consequently miss out on a lot of the support that the OLPC is designed to benefit from (if you do have a XO sitting on a shelf, you might want to try the latest builds).

It’s still not quite there, in my opinion, but it’s getting somewhere. They’re learning lessons, and the lessons they’re learning are school lessons, taken from educator’s experience in developing countries. The hardware is still gorgeous, especially the screen, and they’re only just beginning to exploit its potential (the only bugbear being the mousepad which turned out to be a bit a of lemon: there’s a great deal of hacky code in place just to stop it from jumping around, and I believe they’ve abandoned its graphics tablet mode entirely).

It’s true — there was a great deal about the initial rollout of the OLPC that was screwed up, and if it was a strictly commercial concern, I wonder if it wouldn’t have gone to the wall by now. But it wasn’t, and it didn’t, and I’m fascinated to see what happens next.

5 Responses to “the XO abides”

  1. Fred Blasdel Says:

    Another lemon you left unmentioned is the custom CaFE chip, its SD card controller is a POS (and isn’t open-source like they promised).

    I think there’s a significant problem with applying your “slow-cooked” thesis to the OLPC project: they didn’t start small, they didn’t try to grow organically. Instead they did all the fundamental design and development in secrecy, then had big press conferences where they could boast about how open they were — “Look, you can mess with the chrome!”. They attempted to push it fully-formed into the world while it was still an infant — and instead of devoting OLPC resources to improving it, they held another press conference to debut a new vaporware laptop!

    Their fundamental plan of making software for ‘other’ people was incredibly condescending — it’s not Unix/Smalltalk/Mozilla, it’s Vista. I have great respect for OLPC’s questioning of WIMP, but I don’t like the process they used to do so one bit.

    Their biggest problem overall was probably Negroponte, he just kept fueling new PR blitzes like clockwork, that were increasingly more disconnected from reality. It probably would’ve been a huge disaster if he had managed to live up to his expectations and sell millions in the first year. 100,000 pissy G1G1 purchasers are nothing in comparison.

    The fate of the OLPC foundation is awful depressing, hopefully a stable organization can form outside its shadow, and do the real work of slow-cooking. Walter Bender’s SugarLabs might become what I’m hoping for, though they’ll have to do a lot of work to really make Sugar usable. I hold out a lot of hope for Alan Kay’s ongoing NSF grant project, he’s managed to keep the OLPC hype away from it, while still publishing fairly often (see

  2. Danny O'Brien Says:

    Well, Mozilla didn’t start small either: I think one of the benefits of slow-cooking is that you can back yourself out of those kind of terrible mistakes. There’s an interesting quandary with hardware: the longer you work on creating a hardware platform, the more likely it is to be out of date when you launch. The XO barely hit that window, even though it wasn’t competing with cutting-edge technology — it was only a few months before systems like the Eee and Classmate began eating into its sector. Would it even be possible to create hardware without a hasty race to hit a launch date? The impression I got with the OLPC project is that many inside wanted to be more open, but simply didn’t have the time or the resources.

    I hope you’re right about Alan Kay — it’s interesting to see so much OLPC development work take place on its eToys/Squeak implementation, even though that seemed to be a bit of an afterthought to Sugar.

  3. Ben Says:

    The theme of your article reminds of a William Gibson quote: “The future is already here, it’s just not widely distributed yet.” Maybe that’s a little cliche, but I think it makes sense. Sometimes things get invented, but it’s not enough for the invention to exist for it to reach its full potential. Once the honeymoon is over and developers and users are familiar with what a certain technology (which can definitely be a piece of software – think Mosiac!) is capable of, that is the time when people can look at something and honestly say what parts/concepts from it are good or bad and worth keeping/pursuing, once the “dust settles”, and which are bunk.

    My favorite example is Javascript. IIRC, Brendan Eich “invented” it back in 96 (?) due to the need for some scripting in Netscape Navigator. Maybe JS didn’t back then have all the features it does today, but certainly the idea of something like gmail could have been implemented back then. But it took a long time for the idea of javascript as an application platform to “slow cook” here: the platform was there, but no one had really implemented applications to put javascript through its paces, because up to some point in time, there was no JS standard.

    Unfortunately on the time scale of the Internet/Blog news cycle of hours/days no such judgements can be made. I have seen criticism about Google’s android being “late to the party” but I find that to be impatient consumerist BS. Only time will tell if Android will be a success. Anything really good is worth *waiting* for, I would say, if it’s a good idea whose time has come. And I think the mobile space (vendors and consumers) is hungry for a more open platform than Apple’s hegemony over the iPhone. Let’s let the iPhone vs. Android thing slow cook and see what comes out.

  4. Zooko Says:

    My sons are fairly satisfied with their XOs. I’m requesting one for myself in the new G1G1-Xmas-of-2008-edition, and I’ve advised each of my brothers to get one.

    My son’s use separate USB-connected mice instead of the mousepads, which are too spasmodic to use.

    We haven’t upgraded the system software — we’re still using whatever came with it when it was prepped for G1G1-Xmas-of-2007-edition.

    They’ve benefited a lot from their XOs already, and I hope that their current explorations into editing and coding instead of just playing are going to accelerate.

  5. Mitchell Says:

    It’s nice that projects with future potential can survive on slow-cook, rather than vanishing entirely. This does however weaken the motivation to deal with systemic problems. Fixes can always be put off until next quarter, next year.

    OLPC is, I suspect, only on slow-cook because it hasn’t built some basic infrastructure. Similar projects have a gforge or other community development environment. OLPC doesn’t, and program development, community formation, thus suffer in the ways you would expect.

    Enthusiasm fades for many reasons. Slow-cook provides the ability to survive bad managerial and technical decisions which could kill a commercial project. But it thus also removes the urgency to address them. Which can be both good and bad.

    A key insight of OLPC is that time isn’t free — it costs 300k kids per day.

    For OLPC, a key question is: “How can we optimize a system – technical and social – that gives us the largest and most varied pool of stable, volunteer-maintained, open-source, kid-hackable educational Activities possible? Gforge, workshops, bounties, documentation, toolchains, access… through any means possible, how would you maximize the number of Activites that meet the above criteria?”


petit disclaimer:
My employer has enough opinions of its own, without having to have mine too.