One Laptop Per Child: Just what sort of content do you load onto these puppies?

I’ve taken to thinking about the One Laptop Per Child project in terms of three tiers: hardware, software/content and usage/support. In describing my enthusiasm for and concerns about the project to both people working on the laptop and people critiquing it, I’ve flippantly offered an observation: there’s been roughly ten times as much thought about the hardware as about the software, and roughly ten times as much thought about software as about the challenges of rolling this device out to schools around the world. (There may well be a fourth tier – disposal and recycling of these machines – and I’m open to the argument that that’s received only a tenth as much thought as the third tier.)

This hierarchy of project attention makes a certain amount of sense: it’s hard to start designing software for a machine until you understand what the hardware will make capable, and even harder to get teachers thinking about how that application will change how they teach until they can touch it, play with it and understand it. But there are other reasons as well. Talented hardware and software designers have flocked to the project, but it’s been harder to get experts on deploying technology in the developing world involved. There’s fewer of us, and most of us are working in the developing world rather than wandering the streets of Cambridge. There’s skepticism from some of the best people in the developing world geek field, and even those of us who are enthusiastic about the project realize that there’s much less control over phase three of the project than the first two phases: many of the decisions and much of the hard work of rolling out the laptop is going to made in the schools and education ministries of the nations where the laptop premieres. I can sit in a conference room in Cambridge and offer advice on using PCs effectively in West Africa, but I would have very little control over whether the Nigerian government takes my advice.

There’s another reason as well – constructivism. Alan Kay and Seymour Papert have had a great deal of influence on the design and goals for the One Laptop project, and they’re both closely associated with educational constructivism, specifically with Constructionism, a theory of education Papert developed in his work at MIT. To massively oversimplify a movement I know very little about (invoking blogger’s license), constructionism practices “learning by making”, encouraging learners to build either in the real-world or in simulation, and asking teachers to work as facilitators of learning, rather than as imparters of knowledge.

(To get a sense for just how strongly constructionism has influenced the OLPC project, try referring to the laptop as a “teaching tool” within the OLPC offices. The phrase tends to inspire the sort of embarrased silence that you generally hear in the wake of someone loudly passing gas.)

Because the laptops are devices that enable students to explore, discover and learn – either on their own or with the mentorship of talented educators – much of the focus of tier one and two has been on creating devices that function far outside the classroom. When I’ve expressed my concerns that the devices are subversive to teacher’s authority in the classroom and that some teachers might be led to actively thwart laptop use in classrooms, OLPC software and content President Walter Bender has responded, “That’s why it’s a laptop.” In other words, if the schools are broken pedagogically, at least curious learners will have the chance to explore at home with their human-powered, self-meshing machines.

(This is also a possible response to some of Knut Foseide’s criticisms of the project. Yes, a lab of up to date Pentiums would cost less than equipping a school with OLPCs. But that would give students only a few hours a week of computer instruction, rather than giving them a tool they can take home and explore on their own. Much of the “inferiority” of the OLPC Foseide critiques is neccesary to make the laptops portable devices which each student can own.)

In a little more than a month, 3,500 laptops will be distributed to schools in the nations who’ve agreed to pilot the laptop. This debate about constructionism versus more traditional educational models will be informed pretty damned rapidly by the questions, concerns and feedback offered by teachers and students in the field. One of the people thinking hardest about this question is SJ Klein, OLPC’s director of content, who I spent much of Friday with, in a multiparty conversation about what content needs to ship with the One Laptop.

When we’re talking about shipping content, we’re talking more about servers than the laptops each student carries – the current version of the laptop has 512MB of flash RAM storage, which doesn’t leave a lot of room for large content libraries. But some subset of these laptops will be attached to external hard drives, and likely to some form of internet connection – while they’re otherwise unmodified OLPCs, they’ll act as caches of content and will share their connections via the wireless mesh – despite the fact that they’re peers to the other computers, it seems logical to call them “servers”.

To explain my interest in the topic: getting the content right for OLPC has potential impact far beyond the OLPC effort. OLPC is a very effective lever in getting publishers of educational content to think about releasing their content under an open license. Because so many publishers are interested in being associated with OLPC, SJ has been able to take a very strong stance: unless you’re going to release your content under a license like the GNU documentation license, it isn’t worth our time to talk with you. If this leads lots of people to release their content under these licenses, and if collections of content can be indexed and made accessible through repositories, lots of useful content could be available both to students lucky enough to have a OLPC laptop, and to those who live in schools approaching other strategies towards computerization. A school with a tuXlab or a SchoolNet computer lab could have access to a rich and wide array of books; a school that couldn’t afford a full computer lab, but had internet access and a printer could create textbooks and learning materials in the spirit of the Internet Bookmobile. Even if OLPC fails, an open educational content movement sparked in part by OLPC could have important implications for education in the developing world.

There’s at least six families of content that SJ and crew are thinking about:

Textbooks are the hook Negroponte and crew have used to make OLPC make sense to beancounters in education departments around the world. If students can own a laptop and use textbooks as e-books, the cost of the laptop can be offset by the cost of paper texts over five or more years. From SJ’s perspective, textbooks are largely the recipient nation’s problem – they’re generally chosen by the school system or the education department, and aren’t likely to be replaced by open texts imported from outside – if Nigeria, for instance, decided to start using an open-licensed US history book, it’s not hard to imagine a strong backlash from Nigerian authors and publishers who currently make money in the textbook industry.

Supplements are materials teachers and students can use to make a curiculum richer. This could include materials intended to complement textbooks, like exercise workbooks, as well as standalone texts. In other words, a Nigerian english textbook isn’t part of this category, but Shakespere plays and sonnets are, as are Chinua Achebe’s novels, or a guide to teaching biology using local Nigerian plants. Because these supplements are produced by smaller publishers than textbooks in some cases, and by individual educators in other cases, SJ sees a much greater opportunity to build libraries of this material. He’s already working with Project Gutenberg to select a set of 100 public domain books which would be most useful for OLPC environments.

Media includes current events and news information in text, video or audio format. Most of this information is published directly to the web – OLPC’s interest in this category of material is making sure that it can be archived and distributed on the servers, and in making it possible for content that might be locked to be unlocked to an educational audience. (You might imagine the New York Times or the Economist making their content accessible to OLPC schools, for instance.)

Arts and Culture includes open cultural content which could be distributed with the laptop. This could include music or video suitable for remixing, or to inspire creativity with the laptop devices. (This is the subject area we talked least about, and I may be misrepresenting SJ’s thinking about this area.)

Unique Capabilities of the OLPC machine need to be showcased so that educators and students will use the full capacities of OLPC, not just treat it like an electronic book. To get a sense for what some of these unique capabilities might be, it’s worth looking at eToys, a Squeak/Smalltalk-based multimedia tool, which is one of the applications pre-loaded on the laptop – OLPC is trying to get developers to build eToys for distribution with the computer, but really great applications are likely to need the cooperation of teachers in the field with developers who feel comfortable in the Smalltalk environment.

Manuals and Instructional Material include everything a laptop user – student or child – will need to know to use their laptops, as well materials for a network administrator and anyone repairing the laptops. This needs to include documentation for the operating system and all the applications, localized into the appropriate languages. The team sees four levels of neccesary documentation: end user, teacher/parent, first-tier tech support (the folks dropping off the laptops) and second-tier support (escalation on more serious technical issues.)

Based on my experience in developing nations and the experience of cybercafes and educational projects, I suggested that grouping parents and educators together probably wasn’t a good idea – some of the parents are going to be a much lower level of literacy than educators. I also suggested that projects like SchoolNet Namibia had done a good deal of work in figuring out how to make tiered support work for school computer labs – SchoolNet maintains a phone support line that network administrators can lean on when they get into trouble. Learning from these existing projects and building systems like this so that people’s first experiences with OLPC are positive ones is going to be a major challenge for the project.

This outline helps me get a sense for just how vast the content challenge alone is for OLPC – solving the “supplemental” problem alone would probably be a good multi-year project for an NGO. To do it well, you’d need to interview groups of teachers and students in each country, figure out what resources are needed, then start identifying materials that already exist in repositories, as well as negotiating with content owners for critical materials that need to be released under open licenses. But entirely separate teams will need to focus on documentation and rollout, as well as on building content that showcases the coolest bits of the OLPC project.

While thinking about the logistics and support neccesary to roll out even the first 3,500 machine gives me hives, I had an experience yesterday which went a long way to increasing my confidence about OLPC. I played with one of the Beta machines through the meeting and got significantly more comfortable with the Sugar interface than I’d been in previous encounters. After about an hour’s worth of play, I could write in the word processor, open sites with the web browser, find wireless networks and start – though not quit – most applications. But there were things that were driving me nuts – when I started an application, nothing indicated to me that the application was opening, which led me to open half a dozen windows. The Sugar “Desktop” is accessible by moving the mouse to the edge of the screen (kinda like widgets on my mac), but it requires finer mouse control than I have to access it – and the key on the keyboard that brings it up doesn’t work.

Around 5pm, Mako Hill brings in a USB key with the latest build of the OLPC software. I shut down the machine, plug in the key, and three minutes later, have an OLPC running on a fully updated OS. It’s a surprisingly smooth process, which is a good thing, as this is also the restore process for a busted laptop. Better yet, some of my main complaints have been fixed in the new release – when I launch an application, get immediate feedback that it’s loading; my menu button works and I can quit the web browser without crashing the machine. Clearly it’s still alpha software, but it’s changing quickly, and it sounds like the developers are listening closely to the concerns and complaints of many of the people testing the machine.

I’m still not sold on the need for an entirely new operating environment for the OLPC machines. I share many of Mike Hearn’s enthusiasms about the changes Sugar invokes – moving to a full-screen system instead of tiled windows seems very smart, and using a journaling filesystem to do away with saving and opening files is hugely overdue. But this leads to another worry about OLPC – many countries are investing so heavily in the project because they want students to be computer-ready and able to move into IT-intensive industries. Using a computer that doesn’t have windows, menus or many of the other trappings of the PC as we know it is going to require something of a learning curve… or might spark the rise of Sugar as a professional PC environment in countries where OLPC takes hold. I think the new interface is likely a terrific step in making machines more usable for kids, but there’s going to need to be a way to “outgrow” Sugar and work in a more conventional environment for some of the older students.