Raw

Last night I had a little play with the One Laptop Per Child project's software. The instructions were near-enough, though made it seem harder than it ultimately was. I did spend quite a while lost in a virtual Fedora install. Essentially after installing Qemu and pointing it at an image (olpc-redhat-stream-development-build-303-20070312_1734-devel_ext3.img I'd downloaded the night before) it just works. Unfortunately on this iBook (1.42 GHz, 1.5GB) it was way too slow to get a very good impression - Tetris didn't seem to move, though the Lunar Lander demo in Squeak worked fairly well. Screen size was also an issue - it seems to use 1024x768, the size of this screen, and I couldn't see a way of getting true full screen, icons at the bottom weren't visible, though I did get the tooltip pop-ups. (I didn't go very far down the remote X over ssh path - that might sort the screen).

This interview with Alan Kay is worth reading, and although I disagree with certain points, for most I think he's on the nail. PCs are pretty Neanderthal really, assuming the OLPC project is a good idea it would be shoddy to supply the children with software that's evolved in an environment of various competing business models, where the goal is generally corporate gain rather than computer utility.

I must confess I was rather disappointed with the Sugar environment. While it is pretty it didn't really seem that much different than regular MS/Mac/Linux window managers, a straight desktop paradigm-based setup pretending not to be. A variation on the usual mix of icons and menus. Oh look, that must be a word processor icon. (Not exactly intuitive given this familiarity either, took me a little while to realise that to get the menus up you have to move the mouse point to the edge of the screen). I didn't get anywhere with the social/networking bits, maybe they are revolutionary, but on what I have seen I'm not overly optimistic. Touch more Englebart and Kay (and plenty of Berners-Lee) please, hold the Gates and Jobs.

Aside from simple curiosity, my reason for having a look is that I find the notion of a constrained, consistent life/work/dev environment very compelling. In the region of the text editor best practice - pick one and learn it well (which will be great, when and if I find one I'm actually comfortable with - whatever Mr. Pilgrim says).

But I'm afraid the OPLC stuff seemed a traditional mishmash. If I read this correctly, apps are either native C, Python or Smalltalk based. I couldn't find any dev tools for Python within the running setup, though I assume Squeak had its integrated tools.

To be honest I think the developer's efforts to date would have been much better employed working on making a 21st century version of a Lisp Machine, (complete with integrated HTTP client/server & RDF store) something more deeply consistent than what's effectively Yet Another Somewhat Obscure Linux Distro. A Python-based system would be cool, though Squeak Smalltalk is already 80% of the way there. (Or even Java - OS kernel, Java VM, slap NetKernel on top). The fact that it isn't possible to quickly download and fire up a usable running reasonably equivalent system (complete with integrated dev tools), on a standard machine is downright negligent. If they are going to tweak a Linux distro, they could at least make it reasonably hardware agnostic. But hopefully the open source community will fix that in good time.

On the social question of whether OLPC is a good idea - I'm conflicted. In principle I lean towards the positive, but part of me is frightened by the prospect. May just be my age ("ee lad, sticks and mud for toys? Bloody luxury..."), may be the idea of homogenising culture (at least it's not One TV Per Child), may just be Negroponte's scary family associations. There are a thousand science fiction stories, both utopian and distopian that could start with OLPC.

A tangential question - why are they targetting the developing world from the get-go? Wouldn't it be better running a trial closer to home, in a situation that could be more easily monitored? Somewhere with significant poverty and relatively poor education standards, hence (potentially) still highly beneficial. Texas?

I'm not convinced by the counter-argument on grounds of relevance, this is potentially equivalent to a massive education programme, which could be a Very Good Thing. Give a man a fish and he will eat for a day. Give a man a pair of red braces and access to the stock exchange... Worst case the recipients can always flog the machine on eBay to western gearheads to pay for food & clothing.

[Did I see a post in the past few days somewhere featuring NetKernel & semweb stuff? Not sure if I dreamt it...] 

@en