OLPC is not the good solution [LWN.net]

OLPC is not the good solution

Posted Jan 8, 2009 0:54 UTC (Thu) by Alterego (guest, #55989)
Parent article: Changes at OLPC

I have ben for 3 years in western africa for IT with schools and university.

1/ what is _needed_ is a working IT center (ala cyber center) in each school, that is 20 desktop for 1000 pupils, maybe 50 for 1000 students at university.

2/ the idea of one laptop per children is really more expensive and can only be imagined by rich individualist people who have never been in Africa. The finances does not allow such a solution, and the culture of sharing everything is really close to free-software, and opposite to intellectual property (in some african tongues, the possessive form does not exist in the language, it can only be "the", not "my")

3/ durable: i saw 486 and pentium with win95 and 32MB ram doing a great job, so don't be afraid, like for their cars they have magicians who manage to rebuild working machine with our rubish (even if it would be much better to have new low power equipment).

4/ last, bandwith and power cost are the real problem, don't even dream of 1MB/s outside of capital, unless you have your own satellite stuff.

to post comments

Better than OLPC

Posted Jan 8, 2009 1:12 UTC (Thu) by Alterego (guest, #55989) [Link] (14 responses)

Sorry i was only negative in my previous post, i must propose something better :-).

The solution which can work now (it was immature when i was there) would be something like:
- one server which does everything: firewall, routing, mail, proxy, dhcp, storage, and a local copy of wikipedia in adhoc languages, and tldp.org...
- N light clients with LTSP
- a mailing list for assistance/formation of local admins.

Of course with free software, and all possible documentation (remember, bandwitdth is nearly zero).

Microsoft neutralised free software in developing countries

Posted Jan 8, 2009 1:29 UTC (Thu) by Alterego (guest, #55989) [Link]

For example World-Links, a huge NGO, previous a World-Bank program, which did lots of things:
http://www.world-links.org/en/programs/telecenters/

Microsoft treats directly with governments, and it's nearly impossible for a private IT company to *not* be a "µsoft partner", and when they are, of course they don't do free software, i don't know why ;-).

Better than OLPC

Posted Jan 8, 2009 1:38 UTC (Thu) by dlang (guest, #313) [Link] (10 responses)

what you describe is relativly close to the OLPC plan

one 'school server' that has a fair amount of resources, some sort of connectivity (whatever is possible)

a large number of the XO laptops that contain a bunch of lightweight stuff (more than your LSTP systems, but far less than what's available on the school server)

training classes, mailing lists, and on-site support for the local teachers and admins.

Better than OLPC

Posted Jan 8, 2009 2:13 UTC (Thu) by Alterego (guest, #55989) [Link] (9 responses)

it seem i was unclear, and maybe misinformed about OLPC.

What i heard was olpc reinventing the wheel, and developping hardware, and in the end forget one very important advantage of open source = the possibility to study how it works. Thats how i got my current net+sys admin job, when i was initially a physicist.

"My" :-) plan is to use existing hardware, and customise a distro (let's say ala knoppix) to provide one DVD for the school, plus N DVD's for wikipedia and other contents, and a nice script to install it automagically.

Better than OLPC

Posted Jan 8, 2009 7:22 UTC (Thu) by dlang (guest, #313) [Link] (8 responses)

remember that OLPC 'wasting their time' creating the new hardware essentially invented the netbook market

reminding people that there are two sides to Moores law.

the side that everyone thinks of is that computers get faster and more powerful with time.

the side that people were forgetting is that the equivalent to the old performance gets smaller and cheaper (and uses less power as well)

I consider the hardware side of it a wonderful success.

by the way, one of the reasons why the hardware is over budget is that the opensource community let them down by letting things bloat so tha the inidial specs needed to be beefed up

on the software side they completely flopped. in large part they got into the mindset that they were instantly going to be the biggest linux distro ever, and as such everyone else would worry about being compatible with them and they didn't need to worry about being compatible with anyone else. as a result of the incompatibility they have much less software and far fewer people working with the system than they could have had.

I'm as much of a pack rat and scrounger as anyone, but there is a huge difference between a setup put together with mix-and-match cast-offs and a setup where you have everything the same (with spare parts and replacements available) I've done both and while I will do the first if it's nessasary, I will definantly prefer to do the second. at the scale that OLPC is working in each deployment, doing the mix-and-match approach would require a huge amount of knowledgeable labor on-site to keep things working. I can't see any school systems, let alone governments spending nearly the resources on this that they are willing to spend on modern (if slow) standardized equipment.

Better than OLPC

Posted Jan 8, 2009 8:42 UTC (Thu) by Ze (guest, #54182) [Link] (7 responses)

>remember that OLPC 'wasting their time' creating the new hardware essentially invented the netbook market

I'm sorry to put this so bluntly but that is absolutely crap. There were small factor laptops around before that were aimed at being school kids.

The reason why netbooks are so common now is simply the costs of components has come down to the right price and companies went looking for new markets. This would have happened whether the OLPC was around or not.

What the backers of the OLPC should've focused on is developing the missing technologies and getting them used by others. It'd be a far better bang for donors bucks.

Better than OLPC

Posted Jan 8, 2009 10:04 UTC (Thu) by dlang (guest, #313) [Link] (6 responses)

prior to the XO the small format laptops were the most expensive laptops available. they had the highest performance components that could be squeezed into them.

when the XO was being designed it was being ridiculed as being too limited and too slow to be any good. as it neared full production other companies jumped on board.

Better than OLPC

Posted Jan 8, 2009 13:54 UTC (Thu) by man_ls (guest, #15091) [Link] (1 responses)

I think that the ideas behind OLPC and the Eee are very different: one develops exclusive technology for education, while the other uses standard, outdated components to build a small form factor laptop. Whether the OLPC concept was an inspiration of any kind for the Asus team I don't know, but the components were there and ready to assemble. The small flash disk is certainly an enabling technology, but that is hardly an OLPC achievement (and not used in later models).

The announcement of a $199 laptop (as compared to the "$100 laptop" from OLPC) might seem an obvious ripoff. But not necessarily so. Remember the sub-$1000 PC craze of a few years ago? Everyone was selling those systems at the time.

Better than OLPC

Posted Jan 8, 2009 20:58 UTC (Thu) by luya (subscriber, #50741) [Link]

"The announcement of a $199 laptop (as compared to the "$100 laptop" from OLPC) might seem an obvious ripoff"

Take account of economical reality (lower value of USD) when it comes to the cost of components and materials.

Better than OLPC

Posted Jan 10, 2009 0:51 UTC (Sat) by Ze (guest, #54182) [Link] (3 responses)

>>prior to the XO the small format laptops were the most expensive laptops available. they had the highest performance components that could be squeezed into them.
There were other netbooks/UMPC/whatever you want to call them this year before.The Apple eMate 300 based on the newton comes to mind.

The reason why we are seeing a plethora of netbooks now isn't because of the OLPC but rather because the price of components has come down to an affordable level and they have an acceptable amount of power. We are doing a great disservice to the OLPC project by crediting them with something they aren't responsible for , hiding the real gems they've been responsible for.

>>when the XO was being designed it was being ridiculed as being too limited and too slow to be any good. as it neared full production other companies jumped on board.

Yet they chose faster processors and they aren't choosing the OLPC's screen,wind up power source, or durable casing. IMHO those are the gems of the OLPC. The casing isn't new but the screen and wind up power source are novel.

Better than OLPC

Posted Jan 10, 2009 0:59 UTC (Sat) by dlang (guest, #313) [Link] (2 responses)

given that the OLPC does not have a crank on it (and therefor no wind up power source) this post shows that you don't know what the origization is actually doing and shipping (as opposed to what they thought would be possible and practical several years ago)

Better than OLPC

Posted Jan 10, 2009 3:37 UTC (Sat) by Ze (guest, #54182) [Link] (1 responses)

You've got me there.

So I went to see exactly what they were shipping and the lack of wind up power was the only thing different to what I expected.

IMHO that's a killer for a lot of markets where power isn't available all the time. That to me was one of the killer features of

it , when you combined that with a mesh network , you can end up with quite a nice distribution model for new areas without a permanent net connection , by piggybacking on the part of the population is mobile in different areas and different radii. A good distribution model like that is important if it's supposed to meet it's educational goals.

I think Mary Lou has the right idea with the screen , if they can reduce the cost of that by mass production and popularising it in the general mobile market ie tablet pcs,netbooks,mobile phones,etc. Then the chances for long term success of a cheap education tool become much better.

Better than OLPC

Posted Jan 10, 2009 4:17 UTC (Sat) by dlang (guest, #313) [Link]

this question came up a lot last year as the machines hit mass production. I was at a presentation by Mary Lou Jespin to techies (the USENIX conference) and the question was asked there.

the answer is that the crank ended up being fragile, and not that efficiant, so what they did was to make the laptop _extremely_ forgiving on it's input power (something like 7-30v tolerance and charging best on 9-15v), so it can get power from very dirty sources.

the other thing is that a person cranking can produce more power than a single battery can efficiently absorb, so the emphasis shifted to charging several batteries at once, and produce a large number of different chargers.

they have string pull chargers (think lawnmower)
bow chargers (think old western fire starter moving the bow back and forth)
solar chargers
as well as many other charger designs (wind, water, animal power, etc)

as far as the screen goes, my understanding is that it's cheaper to produce than normal LCD screens (it has 1/3 the pixels of a conventional screen of the same nominal pixel count while having ~3/2 the effective resolution in color mode), it should hit the market in other devices Q1 or Q2 of 2009

Better than OLPC

Posted Jan 8, 2009 9:35 UTC (Thu) by meuh (guest, #22042) [Link] (1 responses)

See the ADEN project:
http://www.africaden.net/spip.php?rubrique2
http://www.africaden.net/spip.php?article30

Better than OLPC

Posted Jan 17, 2009 0:28 UTC (Sat) by Alterego (guest, #55989) [Link]

Yes, that's it. It seems simple and do the right things.
It just lack a local wikipedia and tldp.org mirror ;-)
Thanks for the link, i'll to get in and contribute.