The Fonly Institute

Moving forward

On my way back from Tunis, laying over in Amsterdam, I have occasion to reflect on events, my writings and other peoples' reactions to both.

This blog is not about rubbishing the One Laptop Per Child project. There are some good and valuable aspects to it and I applaud the effort that is being put into it. I would not try to convince anyone not to work on it. Something like the laptop will be needed as infrastructure develops and it will be good to have the results of a lot of engineering effort put into reducing costs and setting new specifications appropriate to the new IT market in developing countries.

I cannot, however, suspend my critical faculties as I see how the project is presented and marketed. There is a real problem with the assertion (which I have had confirmed) that research is not being pursued - supposedly because MIT has done educational research for 30 years. I do not accept that  everything is now known about the world and that the challenge is to act quickly and decisively. Such assertions have been heard before and usually lead to far different outcomes than intended, at which point those who drove the process throw up their hands and blame the technology or the technologists.

The outcome in this case to be avoided is the perception that "you can't give computers to those people - it won't work for them". I can see this as a very real possibility if we all keep our mouths shut and act as if the laptop is the culmination of computer tehnology as far as the developing world goes.

In my view, design matters a great deal. Measuring computers by tonnage or disk size is illusory - what matters is what happens when people use them. Design places constraints upon how people can use computers. We need to have a critical conversation about the issue of design of products which are intended to have great consequences. Not having seen much in the way of such conversation, I have tried to lay out the issues as I see them in the hopes of fostering one.

I do not expect immunity when it comes to my own design work in this field. No one should be immune from criticism who leads a design and development effort intended to affect people in the millions. The offering of evidence, argumentation, and the appropriate adjustments in the project goals and definitions are what we need more of, not something with which we can dispense in the rush to implement.

Lee Felsenstein

Now for a message...

I came to Tunis in search of business. Fonly was established as a vehicle for engineering work related to the class of village-scale telecomm-capable systems which devolve from the system I defined and prototyped for use in Laos by the Jhai Foundation .

Since performing the initial prototypîng of the system intended for Laos, we have concentrated our efforts on solving the problem of sustainable sub-kilowatt electrical power for such systems. The Jhai system was blocked from installation in Laos for unclear reasons and the system which devolved from that design has been put into service on the Navajo Indian reservation in Arizona.

In the meantime, we're open for business in the area of eletronic circuit; device and system design. I bring upward of 35 years' worth of experience in product design qnd development to the table, along with a way of approaching problems that places the user of the center of the design process.

My discipline is electronics (analog, digital, microprocessor, control, medical) and we can assemble teams as required to address the full range of problems in any project

Lee Felsenstein
l

OK, wise guy...

So what would I do instead, if it were up to me? I've got plenty to say on the topic, since it's been my focus for a few years now.

I must admit I've fantasized about being offered the position of project engineer, but there seems to be no possibility of realizing that particular fantasy.

That's because I'd set as a condition the complete re-evaluation and re-planning of the project, not to mention subtitling it "but first make sure it's going to work". That means a simultaneous focus on 1) research, and 2) provision of infrastructure.

Research would consist of 1) ethnographic field investigation, gathering new data among the intended usership communities, 2) review of pertinent literature on education with computers, along with 3) analysis of data fro, both of the above, and 4) experimentation to develop appropriate system specifications and scenarios of use.

The part about infrastructure provision should be the focus of the balls-to-the-wall crash development effort. Fortunately not much research is needed here - there is a body of experience with telecentres (I use the British spelling advisedly here in recognition of the leading role played by practitioners in India) that shows the need for a nice, solid, reliable PC in a box that will operate in telecentre service with a minimum of setup and maintenance.

Telecentre operation involves the following capabilities:
     1) IP telephony, both
          a) through the Internet for international calling, and
          b) through a gateway to the local switched telephone (POTS - Plain Old Telephone Service)
     2) World wide web browsing
     3) Spread sheet operations
     4) email
     5) simple word processing, in the character set of the local language(s)
     6) a simple paint program (for training as well as drawing)
     7) a simple database program
     8) printing capability with the least cost of operation and greatest possible reliability.

An essential part of the package would be a source of electrical power. My approach is a 100 Watt system that keeps a 12 V lead-acid battery charged from a variety of sources, with pedal power as the backup. In addition to a robust pedal frame and an optimizing switching regulator, it needs enough microprocessor brains to run a little database system for keeping energy accounts. My experience shows that the people where it's installed will want to use it to charge other batteries as well, so the technology can't forbid that. Think of it as a small village power utility.

And finally there's the real eye-roller, connectivity. Felsenstein's Ninth Law states that every wire has two ends (and wireless has more), so just putting a WiFi transceiver and a directional antenna there isn't enough. We'd need to craft arrangements with satellite providers for low-cost VSAT time and work out the system structure to enable optimum use of the channel. Then we'd have to specify a low-cost earth station and negotiate a bulk deal on its production. Wifi and just plain wireline modems are the backup here.

It would be a real project - not just something to be dashed off and abandoned to its fate. When it's been shipping for six months and working stably we can then begin to bring in the laptops and be sure they'd work.

Along the way there would be a lot of unglamorous arrangements to be made with indigenous partner organizations and governments, not the high-visibility schmoozing with heads of state and the cream of the business establishment that make the OLPC effort look so appealing (and inevitable).

This one would be the steak, not the sizzle, and that's why it's fantasy to think that the OLPC project could be re-directed toward this methodology.

Lee Felsenstein

Alan Kay comes through

This morning I attended a session at the World Summit on the Information Society entitled "Creativity, Culture and Capacity", organized by Prof. Mike Best of Georgia Institute of Technology (Georgia Tech, or gatech.edu). I had first met him in 2003 when he was at the Media Lab and he was researching the kinds of systems I was building for Laos. He later went on to manage the Media Lab India, left MIT and is now editing "Information Technology and International Development", a journal published by MIT press. I will have to subscribe now.

The session was not well publicized and Best had invited certain people, myself among them, to attend. At four hours' duration, it would not draw many casual onlookers, especially with a bland title. Originally, Nick Negroponte had been scheduled for a panel, along with Alan Kay, for whom I have the highest respect. Another panel member was an acquaintance of mine, and I hoped there would be a good round debate on the OLPC.

But Negroponte had canceled a few weeks before, so instead we had a pretty good discussion on the whole issue of creativity on the internet. I particularly want to quote a comment by Alan Kay, the only one he made that was even tangential to the OLPC.

He said, "There's a hierarchy of elements, each of which is harder than the other. There's the hardware and that's easy - that's just engineering (I don't take that wrong - LF). Then there's software, and that's harder. Then user interface, and that's even harder. Then there's courseware, which is harder still. And finally there's mentoring, which is really, really hard."

I think this is an important taxonomy, and in a later session (Alan had to leave to be part of the OLPC road show) I realized how important it was. Clotilde Fonseca of Costa Rica said that "the digital divide doesn't exist. The use of the term implies there were no divides before the digital one, and there were many... the digital divide is just the old divide in a new area."

That got me thinking about how the digital divide was in fact a design failure. We should have been able to design computers and their software so that everyone could use them. Then Alan set out his hierarchy, and I was thrilled. "He's made my case," I whispered to Bob Marsh, sitting next to me.

Because you need to design from the right end of the hierarchy. PCs were designed from the bottom up, in the order Alan Kay listed, and he was a major player in areas like user interface. But we should have designed them from the top down, taking on the hardest first (hindsight is great!). So now, if we're going to redesign things (or put forward an initiative intended to change everything, as OLPC is represented), we have the ability (and responsibility) to make use of hindsight and structure our efforts the right way.

Mentoring (the human interactions which are both necessary for successful use of the technology and which constitute an important outcome) is the key. Exactly how you design them and design for them is not obvious to me, but you have to start somewhere. Then courseware, then user interface, then software and finally hardware. Each step reveals the requirements for the next.

And, of course, starting from a different set of assumptions about mentoring yields an entirely different path through the hierarchy and results in a different system. Taxonomies provide structure for thought, and the rest is mostly just hard mental work punctuated by insights.

So let's all remember Kay's Hierarchy and try to apply it in designing the next generation of everything. It won't be easy, admittedly, but we've been doing it the hard way until now.

And I rest my case (for now) about the OLPC. It's being designed as a piece of hardware, with all the other hierarchy elements pushed out into the future. You'd think we'd learn.

Lee Felsenstein

The old soft shoe

I squeezed in to the last half hour of Nick Negroponte's presentation of the OLPC today at the World Summit on the Information Society. It was so full it was moved to a larger room outside the exposition hall, in an area which I believe is inaccessible to those holding only an exposition (ICT4all) badge and not credentialed to the WSIS (which means you know someone in a delegation willing to do a favor).

What I heard when I got there was Nick's reply to someone who brought up the connectivity questions. the schools would all have servers that would provide backhaul and could "trickle" data into the laptops overnight. I first thought he was referring to trickle charging of batteries, but then figured it out.

I stood next to a fellow who next got the mike, and he asked "is there a system design?", referring mostly to software. Nick said that all kinds of software would be designed for it by others, and he used his best go-away-and-stop-botherng-me sincere voice to assure us that OLPC would do everything possible to make the right software available.

I turned to the questioner and said "the answer is no". He agreed, we talked, and it turned out that he had some acquaintances that I wanted to meet.

Nick said at one point that it was better to have the laptops than to have nothing, as if that were the choice. He did say that the one-per-child principle was absolutely essential, and gave us the argument that he said won over the governor of Maine:

"Say there was a state like Maine where there was only a spoken language, and they invented written language. So they say that there will be one pencil per school, and some others say there should be labs with ten pencils in them. I say there sould be one pencil per student."

Well, this seems to assume that pencils would cost $100 apiece. If that were the case then it would be a good idea to keep the pencils in a room in each school and supervise their use, making sure there is equity in access until such time as the price can be brought down so that losing or breaking a pencil is not a traumatic event for the child.

Or is he saying that in that example pencils cost $1000 each, and he wanted to bring their price down to $100 so each child could have one. Could they? I really think he needs a better metaphor.

And along the way he mentioned that, of course, they wouldn't be able to equip all the children with laptops at once, but that maybe one school would be equipped, then more later. Now, let's imagine which schools would be the lucky ones...

So it seems as if, in the harsh light of day, there is backpedaling and fast footwork going on, as I suspected there would be. It's the old soft shoe, not a slam dunk, which is good news. It's the herd mentality that most frightens me.

Lee Felsenstein

Some movement on OLPC

A day or so after my post on "Problems with the $100 laptop" an interesting event happened. I'm posting from the World Summit on Information Technology, where OLPC is introducing the laptop. They have a balsa model with a keyboard and an LCD with a thick cable attached to a box under the counter, and Mary Lou Jeppsen, the LCD designer and the chief engineer right now, makes no bones about it not being ready yet. They seem to have added a crank about 6 inches long, made of flat balsa wood pieces.

I have been told that they now claim that the crank will have a 3:1 (instead of 100:1) ratio of crank time to running time. That's much more reasonable from a power calculation standpoint, but not appetizing for users who plan to leave their laptops on 24hrs/day (for mesh networking support). So I went searching for the original statement to show a friend. No dice. The only place you'll find it is in the post in this blog. In fact, there's no mention on the Media Lab web page http://laptop.media.mit.edu of power generation at all!

Now, I know I've seen it in printed articles, so they can't deny they said it. But they did a pretty good job of erasing their tracks on the web (somewhere it has to be in a cache, but caches get updated). I consider this a vindication. But it's also a validation of another point I want to make in better detail.

At a technical conference just before leaving for Tunis, I was on the panel of a session talking about the laptop. There I heard for the first time that the intended use of the laptop was to replace school textbooks. Since textbooks were said to cost up to $20, just 5 uses would break even! I made the point that having a continuously alterable textbook was not something from which I want to have schoolchildren studying. Especially when Rupert Murdoch's News Corporation is one of the funders.

Can you say "propaganda"?

Lee Felsenstein

Problems with the $100 laptop

Problems with the OLPC approach

Lee Felsenstein

Fonly Institute

[email protected]

10 Nov. 2005

 

Top down structure

 

The One Laptop Per Child (OLPC) project (http://laptop.media.mit.edu/) rests upon a fundamental assumption that the creation and widespread distribution of a single type of computer will solve the problem of the “digital divide” in the developing world. By creating a laptop computer priced at $100 each (when sold in quantities of millions), the thinking goes, schoolchildren throughout the developing world will all be equipped with powerful tools for learning and exploration.

 

The educational theories behind this approach were developed by Alan Kay and Seymour Papert starting in the 1970's, and gave us both the LOGO language (Papert) and the concept of the laptop computer (Kay's Dynabook). While their work led to important advances in the shape and use of computers, it has not been generally validated as bringing about new paradigms of child learning. Children do not go out to play bringing along their laptops, and have not been generally observed to create LOGO programs spontaneously.

 

By marketing the idea to governments and large corporations, the OLPC project adopts a top-down structure. So far as can be seen, no studies are being done among the target user populations to verify the concepts of the hardware, software and cultural constructs.  Despite the fact that neither the children, their schools nor their parents will have anything to say in the creation of the design, large orders of multi-million units are planned.

 

Distribution problems

 

This represents a particularly striking form of a command economy where a market economy is an absolute necessity. The history of command industrial economies, like that of the former Soviet Union, shows that informal mechanisms of distribution develop unbidden. Gray markets (for items legally obtained) and black markets (for illegally obtained merchandise) come into operation. For this reason, among others,  we can expect the OLPC laptops to gravitate toward other segments of the population, where money and influence may be available but where budgets are still tight enough to place standard laptops out of reach.

 

The truth of  the assertion that through distinctive design the OLPC laptops will be rendered safe from theft or misappropriation depends upon the moral calculus of the appropriators. So long as they are capable of browser access, the laptops will have value in commercial use. If they can be had, competitive forces will create pressure for businesses to avail themselves of the new tool, as has so far happened with pirated software.

 

It is also reasonable to predict a reaction against the concept in recipient families. In developing societies children are perceived to have a place in helping the family advance, not in racing ahead and leaving the family behind. Unless it is evident that the laptop will improve the prospects of the family then support within the family may not be forthcoming, and the laptop will more likely be converted to cash.

 

It would seem apparent that serious social research must be done to determine family, village and societal attitudes before proceeding with a program like OLPC.

 

Hardware issues – power generation

 

But what of the absence of reliable electrical power? OLPC statements refer to the hand-cranked generator included in each unit, having a ratio of 100:1 for operating time to crank time. For an optimistically low power drain of 1 watt this implies a 100 watt generator.

 

A hand crank of 6 inch (15.24 cm) length operating at 2 turns per second would require a tangential force of 11.8 pounds (5.3 kg), assuming 100% efficiency of generation and storage. This would tire a strong adult quite rapidly. It would seem apparent that the figure of 100:1 was arrived at by means other than calculation.

 

Hardware issues – mesh networking

 

Questions about connectivity of the OLPC laptop are answered by referring to the wireless mesh networking capability to be built into the device. Each one will link to others nearby, which in turn will link to others until finally one links to something connected to the Internet, whereupon all of the other laptops pass their data through the final link.

 

This is a nice idea where the “Internet cloud” is reasonably pervasive and only the final 100 meters remain to be bridged. It will not work so well where the distance to the cloud is in the region of tens of kilometers and where that link is not a broadband connection and not reliable due to power outages.

 

Also, mesh networking depends upon most of the links being operational whenever connectivity is needed. Are we to assume that all of the OLPC laptops will be left running, especially when the effort of battery charging is considerable? Much more likely is that the laptops will have connectivity only in districts of cities where power is reliable and where higher-bandwidth channels are available through wireless access.

 

Infrastructure and alternatives

 

It is not advisable to implement technological systems with inadequate infrastructural support. In the case of the OLPC laptop, the lack of power and Internet backhaul capability will present a serious impediment. This problem can be addressed by the large-scale implementation of community ICT systems, which bring immediate economic benefits in terms of extending telecommunications, agricultural information, telemedicine and e-government functions.

 

In the literature of the OLPC project such suggestions have been brushed aside with the comment that “one does not normally think of community pencils”. Of course, one does not normally think of pencils costing $100 each, and which soak up money needed for traditional pencils. Dr. Theogene Rudasingwa tells of how in Rwanda his schoolteacher had the pupils write on their legs using thorns. Pencils can be quite precious, and in poor schools are indeed treated as community resources.

 

Once stationary community systems become widespread, bringing the Internet (with attendant local wireless access) and reliable low-wattage electrical power capabilities, the ground will be ready for the kinds of laptops described by OLPC. But in that case a diverse marketplace involving a range of types and uses, forms and allocation of the laptops should be in operation.

 

The OLPC methodology is far too rigid to succeed at its stated goals. If it goes forward as currently described, the laptops will most likely wind up in other than students' hands, in areas where infrastructure is more likely to be adequate. The content of material available through the laptop is likely to involve advertising merchandise to audiences more likely to have discretionary money.

 

It is sufficiently discomfiting to consider that the outcome of a massive project like OLPC may be a different form of commercial television for the developing countries. Worse yet would be the preemption of funding for many other projects designed under a community model. Future talk of computer systems for the developing world would meet the dismissive response that “it's been tried and it failed”.

 

Conclusion

 

The time will certainly come when the appropriate tool to promote economic development will be a laptop produced very inexpensively in large volume. Before that point it will be necessary to implement systems that provide infrastructure which the laptop will need, in addition to producing tangible economic benefits for their users. OLPC is to be commended for raising issues and focusing attention, and for posing some technological challenges in a highly visible way.

 

However, the “can do” approach taken by OLPC points in the wrong direction. The solution is not proven to be appropriate, and the distribution model is open to challenge.  Despite this, large sums of money are to be committed to the project in advance to fund manufacturing in deals where the customers are government ministries and not the end users.

 

It is important to begin discussions now that question assumptions and that are open to alternate approaches, lest the outcome be one that diminishes equitable development and that poisons the public trust in ICT as applied in pursuit of the Millennium Development Goals to reduce poverty.

(This is the first of a series of posts on this topic.)