[p2p-research] Thanks for: Suggestions wanted for education to p2p practices and attitude

Paul D. Fernhout pdfernhout at kurtz-fernhout.com
Fri Nov 27 14:52:52 CET 2009

M. Fioretti wrote:
> Besides what I described here:
> http://stop.zona-m.net/education/schools/one-hour-xo-laptop-nepali-school
> while I was in Nepal I talked with people using OLPC there and in
> other countries. I got more than one confirmation that OLPC is used
> and doing good but not because it'ts the OLPC as Negroponte "sells"
> it: that is, people take it because it's the only cheap laptop they
> can get in volumes, but then don't really care about mesh networking,
> put their own software on it and use it a-la-teachermate, that is as a
> teacher support tool. Meaning that yes, it's doing good, but not
> because of:

My analysis of the OLPC project:
* They used the wrong processor at the start (the ARM they are choosing now 
would have made the system cheaper and improved battery life).
* Sugar was a disaster -- not because of the aspiration, but because it 
should have been an "add on", to add social networking or simplified user 
interfaces on top of Gnome (or KDE) for GNU/Linux; as it was, Sugar 
prevented software developers like me from developing software for the OLPC 
because it made the OLPC a tiny niche compared to GNU/Linux, plus Sugar was 
a moving target.
* Emphasizing Python for writing all the applications made the whole thing 
slow; OLPC should have stuck with plain GNU/Linux so apps were in fast by 
being mainly in C  (given the slow processor), or moved to a Java system as 
soon as that became "free as in freedom", or alternative gone entirely with 
Squeak (which always had an open ended education focus and was end-user 
modifiable). Even just building on Forth might have worked out better than 
Python, in which case GNU/Linux would not have been needed at all, 
simplifying the entire project to Forth plus what people built on top of 
that. Forth can be implemented in a cheap FPGA instead of a CPU, which might 
have brought hardware costs down even more.
* The hardware focus was misplaced. Software really was where it was at. A 
simple reading tutorial program and a simple math tutorial program that were 
self-paced would have probably been a bigger net benefit.
* The hardware fit with the culture was wrong (not to mention exposes 
children to theft risks). Village cultures would have benefited more from 
village PCs, so, a centralized public library-like facility for the village 
with a satellite internet connection and one hardened PC computer plus many 
cheaper terminals (even cell-phone like if needed to reduce costs).
* The decision not to sell to the industrialized world on a consistent basis 
was a mistake, because that would have been a way to lower costs through 
increased volume as well as interest more developers.

I bought two OLPCs (well, four :-) to try out for development. I did port 
some code to the OLPC (from our PlantStudio software) but it was very slow 
in Python. I might have improved the performance, but it did not seem worth 
it due to Sugar and the time it would take to Sugarize the app. Ultimately, 
Sugar was too much of a hassle to deal with constant changes with and too 
poorly designed IMHO at the time, even as it continues to improve and even 
as the ideas in Sugar are good. As a developer of free software with limited 
time doing this on my own time, why should I spend time developing for Sugar 
when I could just develop either for GNU/Linux in general or for Java? In 
that sense, the OLPC project alienated itself from the developer community, 
misjudging how developers would look at this. As I see it, having to triage 
my efforts, by poor strategy, OLPC and Sugar forced me to make a difficult 
choice -- I could spend my time making an application just for the OLPC with 
an installed base of, at most, a million materially poor kids (worthwhile, 
obviously), or I could make software (or, it turns out, content) for 
hundreds of millions of people who could run Java or tens of millions who 
run GNU/Linux, with the expectation that in a couple of years, millions of 
materially poor children when other projects just put regular GNU/Linux in 
their hands or Java based computers. It was a difficult choice. Having to 
even think about it pretty much stalled my work towards an OLPC app. I think 
many free software developers would not even have gone that far.

More than US$20 million dollars was spent on the OLPC project just for 
hardware development and software infrastructure like Sugar. I think it has 
very little to show for it in that sense (even as I think our society would 
be justified in trying 1000 OLPC type experiments, so I think more money 
should go into this area despite the problems I outline). The new quirky 
OLPC screen is rapidly becoming obsolete compared to eInk digital paper. 
That is the nature of fast moving hardware these days. The hardware design 
is already being discarded (with finally a move to ARM instead of an x86 
based processor as anyone with embedded experience might have told them at 
the start). The software is incomplete and buggy, while there are literally 
thousands of GNU/Linux applications that kids could have been using but on a 
practical basis cannot because Sugar makes that hard, and otherwise there 
might have been even more educational tools inspired by the OLPC but made 
for GNU/Linux in general if developers did not have to wrestle with Sugar 
(and its many incomplete incarnations). The OLPC pricing model failed in the 
sense that anyone in the consumer electronics industry could have told 
Negroponte that only one third of the retail cost of products is the 
hardware (the rest is supply chain costs, profit, recouping R&D, or 
advertising and so on), so pricing a retail US$600 netbook at US$200 from 
huge orders through a non-profit was no great magic. (And there were cheap 
classroom computers as well as smart cellphones even when the OLPC project 

In the long term though, yes, I think there will be one (or many) computers 
per child. And overall, OLPC was a worthwhile attempt, even despite it 
making just about every obvious error such a project could make (like they 
went out of their way to fail, and were still a bit of a success anyway. :-) 
We just need many, many more such attempts. To reiterate, because I said so 
much bad about Sugar, I think the goals of Sugar made sense, they were just 
implemented in an unfortunate way, to get in the way of all use of software, 
rather than as as add on.

As the "hole in the wall" project showed, and it existed before the OLPC 
project, illiterate children in materially poor circumstances can pick up 
common computer interfaces very quickly.

Repurposing old computers might make a lot of sense too. IBM and other big 
companies every year take literally millions of older laptops and crush 
them. (I had a plan to turn those into display walls and built a prototype 
at IBM Research.) Same for older cellphones that could be repurposed (some 
people edit Wikipedia from their cellphones), and maybe even reprogrammed as 
terminals with a local network. In two to three years or so, the current 
generation of smart phones just coming out like the Google Droid will be 
discarded for something new, and those might make terrific cheap education 

So, Droid is a more tempting platform to me for educational software than 
the OLPC and Sugar in that sense of a big market. :-)

Imagine, Google and Verizon could even make a promise now to customers -- 
buy your Droid through Verizon, and in two years, if you continue your cell 
phone plan, we will give you the latest Droid version and if you return the 
old one to a Verizon store, we'll send it to materially poor kids loaded 
with educational software that teaches them how to read, write, and do math. 
And with bluetooth, and WiFi, the Droid could even have some software that 
works along the lines that Sugar aspired to do, with kids collaborating 
together. What a deal -- and it might greatly boost current sales. :-) Maybe 
someone should forward this note to someone they know at Google or Verizon? 
:-) Seriously, what US teacher would not buy a Droid over an iPhone knowing 
it was going to teach some poor kid to read in two years? (Of course, Apple 
might eventually have to follow suit. :-) And that gives me and the rest of 
the free software developer world two years to write all that free software 
for those kids. :-) Of course, it might be nice if Google or Verizon helped 
some of those free software developers to write lots of cool stuff (millions 
of dollars in support for education software could just be considered part 
of their advertising budget). But it might happen even if they did not 
directly provide support, because a lot of developers might see the 
potential, as I did. And it might help Droid sales even now, for parents to 
hand their Droid to their kid who was learning to read or write or do 
arithmetic, and it would help the kid. Parents might even buy a Droid for 
all their kids, and think that in two years, those Droids would also go to 
materially poor nations. This project might even help boost the economic 
recovery in the USA. And of course, there are many Android devices beside 
the Droid, so all of those might benefit as well from educational software. 
And, the Android platform already runs well under almost any PC OS in 
emulation. So, any free software made for the Android will also run right 
now on any desktop or laptop, and likely that integration could be improved 
even more over time. I'm using an Android emulator on the Mac for playing 
around toward an Android app I'm thinking of releasing as non-free at first 
to pay the bills -- as sad as it is for me to go back to writing non-free 
software. :-( At least if I sell it myself, I can later free it, whereas if 
I write proprietary stuff for others as I have in the past, it can never be 

Here is why I feel somewhat qualified to criticize OLPC, :-) a post I made 
about a US$100 educational laptop from 2001 (yes, probably years before it 
was a gleam in Negroponte's eye :-), and I still think my design outlined 
there makes a lot of sense (except a SSD would be better, true): :-)
   "[unrev-II] The DKR hardware I'd like to make..."
"Developing and then deploying this sort of device is the sort of thing
the UN or a major foundation should fund (if they were on the ball).
But luckily, there is hope from toymakers!"

So, I should now add, "But luckily, there is hope from the cellphone 
makers!" :-)

Seriously, this sort of planning for a formal transfer of millions of Droids 
to the materially poor world in two to three years could shake the 
foundations of our global society in a good way. (A bit of a pun intended 
there too -- foundations, shake the money tree, get it? No? OK, so I'm a 
programmer, not a comedian. :-)

--Paul Fernhout

More information about the p2presearch mailing list