Tuesday, August 01, 2006

Conflict from the OLPC mesh

The One Laptop Per Child project is receiving much criticism in recent days (also see a waffly article in PC Pro, August 2006 issue). I'm personally concerned about two things.

I just had the first look at the mockups of the Sugar interface on the laptop.org wiki, and many of the screenshots show browser windows displaying Red Hat promotional material - hardly suitable for educating children! I strongly believe these links must be omitted from the final version. Obviously, information is required, but not inundated with PR speak and graphics.

The second concern is about the mesh - this is an idea that I independently had a little while ago, of making direct links between mobile devices to make masts redundant. I imagined this would be useful if using a mobile in the countryside, so long as a chain of capable mobile devices was available trailing back to civilisation, where masts as well as a denser, more reliable mesh would be available. The OLPC project is intending to use this principle to give internet access to their hand-crank recharged laptops. As far as I understand at this point, the mesh goes down when the computer is switched off; trivially, it definitely goes down when the battery runs out. So what if the kids in the next village don't keep their laptops cranked, and you end up losing the link? This can either lead to the mesh collapsing over time (because if I've little hope of getting a link, I won't crank my machine) or to conflict between villages. Not that learning conflict resolution is a bad thing, but this should be addressed by the organisers when the laptops are introduced. If you sell these as a wonder pill, and then the link keeps going down, people will be hugely disappointed and toss these laptops aside. Back to carpet-weaving you go, Chaitanya!

1 comment:

Anonymous said...

Your concern about the social aspects of mesh networking is a new angle. I've only wondered about the technical side - getting it to work in huilly, dusty, rainy conditions.

But you are so right. Why string-pull (the newsest power system) if I am not assured of a connection?