Presentation at the University of Toronto (With long conversation afterward ...

Spent the afternoon doing a presentation at the University of Toronto. Basically an introduction and request for developers. The Networked Systems group in particular showed me a couple of already-built applications that might be useful for the project.

They pointed out that their research (which focuses on (electrical) power reduction from software optimisation of network communications) isn't getting applied in bandwidth-rich environments, but that they might be applicable for a satellite, modem or similar constrained-uplink situation.

About the projects (team leads, if you're reading this and think one of the projects sounds useful, contact me and I'll put you in contact, otherwise I'll root around to find the right contact points):

One is designed to reduce total network bandwidth by using a Firefox plugin along with a Squid proxy for retrieving web-site images. The plugin allows the child to request a higher-resolution version of the image, algorithm then prioritises higher-resolution version of the image for other children.

It would mean that on image-heavy sites where most of the images are basically useless the children would only download extremely low-bandwidth versions of those. With enough children looking at a page, the percentage who need to request higher-resolution images drops until most of the time a given child doesn't need to increase resolution at all, the interesting images are already adapted. Requires a Squid proxy and a plugin installation on the laptop itself; pretty elegant, simple solution using standard components.

Another is a (completed) Master's thesis looking at delay-tolerant networking mechanisms, particularly using abstraction layers that allow for mesh-based delivery of things such as email with connect-time simulation to make it appear the events happened while networking (email being the given example). Also interesting work on tracking point-to-point contacts among users with analysis to allow for history/statistics based routing (this guy normally talks to someone who normally talks to someone to whom I want to send a message, so I'll give him an encrypted message for that person). Interesting (sideline, from what I understood) work on reducing the "active mode" operation of the WiFi device to optimise power consumption using bandwidth shaping.

Also a finished Masters thesis with a Firefox plug-in originally targeted at mobile phones, PDAs and the like. It's an interface to allow a user to customise a site (think of a news site) in such a way as to strip out the less-desired elements and leave only the "core" of the site. It might be usable to make news sites fit well within the e-book mode of the laptop (i.e. just show the "content" column of the page). As with the first project, it has the advantage of being a plug-in that could be downloaded/installed with little effort (and this one does not use a server at all, so it's entirely a child's choice).

The Human Interfaces group was also interested, though their work is largely on the high end of the spectrum (they work on the BumpTop project, for instance). Anything of that character is probably a revision two project (they're pretty much all OpenGL), unless we look at something along the lines of the physical simulation rendered into 2D using PyGame or something similar.

Biggest question, as always, "how do I get one (some)". Requests from all over to get single machines for trying out projects on them, multiple machines to run classes that focus on hacking on them, and ones to just play with. Also a rather smart question "where do the developers in Toronto get together to work on it".

Since Myles and I all show up at Linux Caffe daily-ish and Mel drops in quite frequently... I suppose that's where we're meeting these days, but the idea of having a formal weekly meeting for interested developers is probably a good one. Of course, I'm disappearing to Italy and Albania for three weeks later this month, but in the mean time we could try to get something started.

One thing that kept coming up was OLPC Nepal. Because I've heard from them more than anyone else I find myself referring to their work far more than any of the other countries... which is weird, because IIRC they haven't yet got a commitment from their government.

One comment by Jing, which really stuck, was that until he actually felt the laptop in his hands he was rather non-committal and skeptical about the project. It really is a very tactile experience to hold it and know that there are going to be millions of them, and that they are going to change the world.