The XO Files Part III: Re-imagining the OLPC Distribution


This entry is the third in the four-part series, "The XO Files: I Want to Believe" Read Part I here, and Part II, The New 4PC Market, and its Failings.

The XO Files: I Want To Believe
The XO Files: I Want To Believe

Part III: Re-imagining the OLPC Distribution

Concern over the original distribution plan was what got me writing for The belligerent anti-pilot-project attitude, the requirement to buy the laptops in lots of 1million units, and the hushed discussions about the costs beyond the "$100" laptop. Rapid, bulk deployment is not a good model to introduce technology, particularly in a resource-constrained environment.

If you look at case studies of technology diffusion or successful ICT4D deployments (the Grameen Bank Village Pay Phone Project for example), you see the very social process of technology adoption, as people judge their usage of a new technology based not only on features and promises, but about lived experiences of their friends and networks. Duncan Watts's Six Degrees takes the network theory approach to this, and Rogers' Diffusion of Innovation drives it home with extensive case studies, and the classic "Diffusion of Innovation" graph:

Among individuals, there is a normal curve distribution of the groups in any diffusion process, which are innovators, the creators of new technologies; early adopters, those who are most likely to begin using a new technology first; the early majority who will adopt next, and then the late majority and laggards. The most important actors in innovation diffusion are the change agents, who are generally external entities bringing new technology to a community and the opinion leaders within each community, who have immense influence over the adoption of a new technology by their peers.

Bell curve illustrating the stages of technology diffusion

Not to belabor the point, what can you do with a million laptops?. It's an overwhelming number to absorb into an educational system without any prior experience with the OLPC XO, not to mention 1:1 laptop usage or computing in general. Add the fact that both the hardware and software were (then, slightly less so now) new and untested, and you're externalizing a lot of risk onto governments, which are most likely already in debt for other projects that haven't paid off as promised.

Even accepting that risk and going forward, pilot projects serve more purposes. Pilots do not just to make sure the technology "works" and that there's some return on the investment (in the form of improved test scores, attendance and participation, etc.). A pilot project can and should also develop local best practices, curricula, and on-the-ground educators with experience using the XO in a classroom. To be fair to OLPC, their plans in 2006 were to open up the distribution to a wider audience (this hasn't quite happened as yet) after the initial launch:

The key to the OLPC vision will be scale, which is why the group will initially make the computers available only to governments that place bulk orders of more than 1m units. The seven launch countries are expected to order up to 10m units in 2007. With time, this will change. "After the 2007 launch, as little as eight to 10 months later, we will open this to all non-governmental organisations, countries, states within countries, right down to school districts," said Mr Negroponte. (Limbach 2006)

Unsurprisingly, this initial approach fell through, and they've moved towards the follow-on plan. After a set of handshakes, the orders didn't materialize, and after telling people who wanted to run pilot projects "...screw you, go to the back of the line", Nicholas Negroponte and OLPC have been recently implementing much more reasonably sized pilot projects, instead of waiting for an implementation miracle.

Ideally, OLPC would have started out pitching hundreds of small pilot projects, and engaging not only governments but also NGOs, individual schools/school districts, and even communities. The project was exciting enough and the per-unit cost low enough that many of these organizations would have engaged and become (hopefully) champions, local experts, and global examples, leading to more and larger deployments in a more organic and long-term sustainable fashion.

The current situation with modestly sized pilots in a manageable number of countries is a much more sane approach. This is revealing some of the weaknesses and gaps of the technology (as there always are), and is generating the highly important localized content. Still, there's much that could be improved with very little effort.

Before G1G1, you had to be a hotshot software developer committed to helping out with Sugar projects to get an XO to work with. This helps with the technological side of the equation, but why not send some units to top-notch educators around the world to start exploring use cases, creating curricula modules, and sending feedback on classroom-level usability? The dream of the XO as a "Trojan Horse" entering the classroom as a textbook replacement but empowering children to use it at home for learning and exploration is a wonderful vision. At the end of the day, however, the teacher is the gateway to the classroom, and if he or she suspects the children are using the XO to chat during class, cheat, or so on, they will rapidly get banned. As such there has to be some teacher training on how to leverage the XO and be involved. G1G1 enables some educators access to the laptops, but some free (or at least at-cost) XOs to more educators globally could only spur more interest, open, shared curricula plans, and community efforts.

The restriction of G1G1 2007 to North American donors was unfortunate, but hopefully G1G1 2008 will be more open. Creating a global community of "early adopters" is key, beyond merely adding to the cadre of software developers (of course, whether that matters now with the move towards Windows XP is another question). A global community will find innovative ways to use the XO, contribute to debugging, and content translation and creation. At the very least, the experience will reveal remaining gaps in the software and UI through common tech support problems.

Fixing Distribution

Again, OLPC has made some great strides in addressing their distribution problems -- running smaller pilot projects; the initial G1G1, though hasslesome, increased the user community; and G1G1 2008 looks to be even better. There's even rumors of an OLPC partner in India looking at selling OLPCs with cellmodems.

OLPC needs to go further, to match innovation in distribution with their innovation in design. The goals are to improve education worldwide at a fundamental, constructivist way. The needs are high production runs to keep the costs low (ish).

An ongoing G1G1 with the ability for anyone in the world to order an XO would be a good start. OLPC might have to be careful managing the non-profit side of that, but worst case is they spin off a small, focused for-profit company that manages selling and limited support -- which can provide even further feedback as to which features users find interesting and what are common requests (though, at the end of the day, I feel strongly that the XO should retain its design focus on low-cost and education-in-developing-world needs).

Another option is to refactor the OLPC as a "base of the pyramid" style technology, selling it through financing and providing training to small businesses and entrepreneurs interested in bringing a portable cybercafe to their village, or in creating citizen media reports.

In short, OLPC needs to look at new ways to get the XO into more hands - more users, more educators, more children, more entrepreneurs, more developers. This creates a better infrastructure of content/guides/curricula for current and future deployments, helps in bug-testing, translation/localization, and creates a larger community like we already see in the global grassroots organizations and learning clubs, and on the OLPCNews forums. These might not all march to the tune of the 1CC drums, but might just be the crowd the XO needs to thrive.

Related Entries


Hear! Hear!

These suggestions are very close to my own feelings.

1) Get XOs at subsidized prices to anyone who is in the position of doing the project some good -- educators, developers, community leaders.

2) Get XOs at or close to 'government' prices to anyone who wants to do something positive with them -- self organized pilots, school computer clubs.

3) Let _anyone_ easily buy one, hopefully in a localized version. If not, in a US or 'international' version, with the promise of localized keyboards as they become available.

4) Start realistic pilots. Pilots at the end of five day donkey treks in schools with five amps of power, no infrastructure, and with participants speaking indigenous or minority languages might seem philosophically pure. However, pilots near a big town, with internet, power and possibly a sophisticated faculty would seem a more practical way to run a first pass at figuring out what works and what doesn't.

5) Make the infrastructure more available. Easy kits to self create schools servers, virtual community school servers. Get the whole experience out there, not just one isolated XO.

Thinking about that bell curve...

The number of children in the world ages six to twelve (OLPC's target) has to be somewhat north of one-half billion. This puts the 'innovator' segment at around fifteen million! I've seen figures that put OLPC shipments at about 55,000 per month at this time. We're not even putting much of a dent the innovator population.

This just makes it seem (even more) to me that NN's sell by the millions approach, which seems to be aimed at the fat part of the curve right now, isn't going to get much traction.

His hope was for about ten million machines out there at this point. If that had happened, we would have closed the innovator window and be ready to go for real adoption, but we're not there yet.

Robert - the problem with the bell curve here is twofold. First, the innovators in this model are the folks in Boston and the communities springing up around the OLPC, which is not really a percent of the children 6-12 worldwide. The important audience for deployment are the early adopters - savvy decision makers for organizations as well as people like us who become evangelists for the technology.

Regardless, we strongly agree on the need to better support a more organic diffusion of the technology - I don't think either of us think top-down mass purchasing and distribution (at least without initial pilot projects) is a sustainable method to really getting good results from the XO.

Jon, you make a lot of good points. However, you need to understand that NN's initial mass deployment strategy, while flawed in many ways, was nonetheless essential if the project was going to get off the ground at all.

There were a number of reasons for this, but let me mention just two. One was that there was an unspoken agreement in the laptop industry to not produce the sort of inexpensive machines that were needed. The oem's, along with Microsoft and Intel, engaged in this conspiracy because they knew they could make a lot more money this way. To overcome that NN had to get the world thinking this was going to be a huge market, thereby getting one leading oem, Quantum, to jump in.

The second reason was the display. NN had to get a display producer to make a radically new design, and no one would be interested unless sales in the millions were anticipated.

I think you also make a larger mistake. You imagine how things could have been done in an ideal world, but often the best workable plan is far from perfect. That being the case, the real question is whether or not a plan will more things forward, and by this more reasonable criterion we should judge NN's plan a remarkable success, far greater than that of a many other people who have tackled the problem of developing world education.

All that said, let me repeat that I think you have a lot of good ideas about where things need to go from now on.

I still say that the best ground for these machines is not the third world. It is the developed world which is "infrastructurally" capable of running these kinds of projects and has the means to pay for the full ownership costs.

I am all for poor countries but they may not be the ideal places for being the testing ground of this technology. It should be proven technology what we give them. No wonder most of them are wary to start XO projects.


I realize he had to achieve huge scale for the price and simply get in the door. What I don't see is why OLPC had to externalize that constraint on government purchasers. The primary target market is huge, with big buyers likely in many cases (governments and aid agencies), but there are secondary (and perhaps larger) markets in the industrialized nations with equally large potential buyers (entire school districts/states and even some national-level programs), as well as interested users, price and travel/rugged -conscious users and organizations (I could see it replacing a PDA in many data-collection tasks). The point I want to make is that other computer sellers (admittedly more established) don't force large contracts on potential buyers -- they just have large markets containing buyers who will inevitably make large orders. If the million-minimum-order was a white lie to get in Quanta's door, fine; but it seemed at the time a hard and fast rule (that's since broken down). The OLPC XO penetration and user-base, and even adoption globally would be better today if it had started out more flexible.

I don't agree with your second statement. The OLPC XO is in a few pilot programs. I don't see that as really moving the needle in global education (yet). There've been countless attempts to change education at every level. OLPC has produced a lot of buzz, but the jury is still out on its results, especially if you're looking at a global scope.

I'd argue that focusing on realistic, proven diffusion-of-innovation models is the way you leave the "ideal world" and take advantage of real-world social networks (the offline kind) and organizational structures to encourage adoption at the local level.

At the end of the day, though; this is all crying over spilled milk, and with the "Post-1CC" activities of user groups and Sugar Labs, it's time to focus on how to best move forward.

Anyone who ever wanted to play around and/or write programs for the XO-1 could always find them on eBay. Making it sound like G1G1 is the only *easy* way to get an XO-1 is wrong and misguided.