Business Week has an in-depth write-up about the One Laptop Per Child project's first deployments in developing countries. The original plan called for building 150 million laptops by the end of 2008; it now looks like they'll be lucky to ship a million before the end of the year. It appears that a big part of the problem is that Nicholas Negroponte and his team underestimated the support requirements for the laptops. Getting laptops into the hands of poor children is good, but it's a lot better if the laptops come with training for teachers and support personnel on how to use them effectively. OLPC may have hoped to build a laptop that was so easy to use that little support was required, but the countries writing the checks don't appear to have bought the argument. Nigeria, for example, backed out of a previous commitment to buy a million laptops from OLPC, opting for Intel's Classmate PC instead. Intel's superior support was cited as the major reason for the decision.

This highlights what was so ridiculous about Negroponte's demand that other companies stop offering competing low-cost laptops. Negroponte deserves credit for pioneering the concept of producing cheap laptops for poor children, but coming up with the idea is, relatively speaking, the easy part. What's far more difficult is the execution. Technical wizardry is an important part of that, to be sure, but probably even more important are the logistical details: keeping the project on time and under budget and ensuring that the shipping project has adequate support. There are a million ways for things to go wrong, which is why it's a good to have a bunch of different organizations working on the problem in parallel. By his own admission, Negroponte is more a visionary than a strong manager, which is precisely why he should have welcomed the entry of a company with Intel's logistical prowess into the market. It may not be as personally satisfying for him to have a for-profit company finish the job he started, but if the goal is to help poor children, then he should be happy to see them being offered more options.

From the outset, one of the oddities of the One Laptop Per Child project has been the tension between its organizational philosophy and its software platform. In his famous essay, "The Cathedral and the Bazaar," Eric Raymond contrasted two organizational philosophies for developing software. In the Cathedral, software projects are organized in a top-down fashion, with the development process following a plan carefully developed by the project's leaders. In contrast, the philosophy of the Bazaar is to "release early and often, delegate everything you can, be open to the point of promiscuity." The OLPC project was a strange beast because it was clearly organized on the "Cathedral" model, yet it was developed around Linux, the open source project that Raymond used as the poster child for the "Bazaar" style of development. And its broader vision of empowering third-world kids to use the laptops without a lot of central support, is clearly more Bazaar than Cathedral.

I think many of the problems we've noted with the project stemmed from this fundamental conflict of visions. Nicholas Negroponte's vision for the OLPC organization has always been the model of the Cathedral: produce a perfect laptop on the first try and sell it in batches of a hundred thousand to the world's governments. Negroponte's plan left little room for the kind of development growth, bottom-up participation, and trial-and error that characterizes the Bazaar. Indeed, even when customers were beating down the door to try out Negroponte's product, he resisted selling it to them because it conflicted with his vision. And of course, he absolutely hated the idea of his customers having other options to choose from.

This tension was never sustainable, and indeed there are increasing signs that OLPC's innovative technologies are being steadily liberated from the Cathedral. In January, we noted that one OLPC alum was starting a new firm to commercialize the OLPC's display technology. Now CNet notes that another OLPC alum, Walter Bender, is starting a new software spinoff to license OLPC technology to a variety of laptop manufacturers. Bender's decision to start a new company was presumably sparked by Negroponte's decision to run OLPC more like Microsoft, which one engineer claims involved demoting Bender in favor of someone with less technical expertise.

It seems that the folks who have left OLPC have a more Bazaar-like vision for their companies, licensing their technologies to a variety of companies. In contrast, Negroponte seems to be doubling down on the "Cathedral" model. He's reportedly considering a switch from Linux to Windows. That would be oddly appropriate given the apparent similarities between Negroponte's management philosophy and Steve Ballmer's.

The Christian Science Monitor has an interesting story looking at the rise of dirt-cheap laptops and the potential impact these laptops will have in developing countries. It gives a fair amount of attention to the One Laptop Per Child project, which was obviously one of the early players in this space. I've had my share of criticisms of the OLPC project, but one thing I do have to give them credit for is that their XO laptop seems to be very competitive with the laptops being offered by commercial companies. Most of them, such as the Asus Eee PC, are priced in the $299 to $399 range; it appears that no one has yet figured out how to produce a full-featured laptop at that magic $100 price point. The thing this article does highlight, though, is that OLPC is operating in an increasingly competitive market. OLPC head Nicholas Negroponte says "I don't want to compete with anyone," but he's going to have to compete whether he likes it or not.

One of the most intriguing competitors is Ncomputing, which is trying to resurrect the dumb terminal model for people on a shoestring budget. Ncomputing uses a cheap ($350) PC as a server to drive a bunch of ridiculously cheap ($70) terminals. Dumb terminals are almost as old as the computing industry itself, but getting the terminals to be this cheap certainly opens things up to new markets by bringing hardware costs within reach of that magic $100 price point. Of course, these dumb terminals won't be as portable as an XO laptop, and they likely require more tech support. Schools in developing countries will have to weigh those disadvantages against the XO's higher price and decide what will serve their students best. And that's the way it should be: more competition means that end users will be able to choose the computing solution that best fits their unique circumstances and budget.

In the latest blow to the OLPC project, the organization's security chief, Ivan Krstić, has resigned over philosophical disagreements with the organization's direction. The nub of the dispute seems to be chairman Nicholas Negroponte's belief that now that it's out of its startup phase, the project needs to be run "more like Microsoft." Krstić complains that the organization's previous president, Walter Bender, was demoted, and Krstić was asked to report to "a manager with no technical or engineering background who was put in charge of all OLPC technology." Now as we've said before, it's healthy that the OLPC organization is beginning to realize that they face many of the same challenges as for-profit technology companies, and might be more successful if it adopted some of their methods. But bringing in non-technologists for senior leadership positions and adopting a rigidly hierarchical org chart might be taking things a little too far. Culture matters in technology companies, and it probably matters even more in an organization like OLPC that depends on having employees willing to go above and beyond the call of duty for relatively modest pay. More orderly management is a good thing, but not if you cause your best people to jump ship in the process.