Shifting from Pure Innovation to Implementation Pragmatism


There is a massive reorganization happening at One Laptop Per Child. It started when Mary Lou Jepsen left with these key parting words:

"My job was simply done," she says. "I was responsible for the hardware, and I got it into mass production." She also says that working on a project like OLPC takes a toll. "The OLPC is like the Peace Corps. You go in for a couple of years, and it's really hard, but really rewarding."
Empty seats at OLPC HQ
The change then accelerated when Nicholas Negroponte both went searching for a CEO and reorganized the organization into four operating units; technology, deployment, market development and fund-raising, and administration.

Now we have Ivan Krstić resigning with a suggestion of Water Bender's "demotion from OLPC presidency," and a Board of Directors shakeup that leaves the organization without its key creators or a MIT feel to its leadership.

While some may feel this means the wheels are coming off the cart, I think we're finally seeing a shift from pure innovation to implementation pragmatism. And I only hope it's not too late.

OLPC needs to become more focused on the messy, difficult, and dull problem of getting laptops in the hands of kids, and less time on the easy, fun, and exciting technology innovation of the past two years. Or as Charbax said in a comment:

OLPC needs to become a Microsoft, giving every child a laptop and making of it a successful revolution in the worldwide education system, implementing new digital curriculums successfully, supporting students, parents, teachers and schools the right way. Speeding up the software development work, speeding up the availability of educational software and content for the laptops.

This requires many more employees at OLPC then 30. The little non-profit that is OLPC today needs to become at least 10-200 times larger in terms of number of employees, all the while continue to engage even more enthusiastic software developing volunteers around the world. So definitely OLPC needs restructuring.

We're in the middle of that process right now. Gone will be the dream-makers like Jepsen, Krstić, and maybe even Negroponte himself. In will come the bean-counters and the functionaries that make any large organization efficient.

OLPC will not be as exciting, but hopeful much more effective in its current task - laptop distribution.

Update: Here's a step in the implementation pragmatism direction. Walter Bender just uploaded the OLPC Deployment Guide onto the wiki.

Related Entries


I think the assumption that OLPC is moving from "pure innovation to implementation pragmatism" is a bit premature. The move may just as easily be to extinction whatever the rationale might be.

As much as the XO has right, there are some things it has serious wrong not the least of which seems to be Sugar. From my modest experience and reading the comments of others, Sugar seems to be in alpha stage; not anywhere near ready to release to the public. That's too bad because the hardware design sets a new standard for portability and convenience.

Wayan wrote:

"OLPC needs to become more focused on the messy, difficult, and dull problem of getting laptops in the hands of kids, and less time on the easy, fun, and exciting technology innovation of the past two years."

I strongly disagree, Wayan.

The currrent task should be to fix the many problems with the XO and to come up with some sort of implementation plan. And, if both things are not possible, at least making the laptop fully functional and relatively bug-free should be OLPC's current and top priority.

Concentrating on distributing this "alpha" version is irresponsible and will only create more problems. Imagine distributing hundreds of thousands of units only to have to recall or service them on the field in a few months?

OLPC needs to get in touch with reality. The sooner the better.

As one of the enthusiastic software developers, I agree: OLPC needs to be pro-actively cultivating a community of developers, preferably ones who have done (or are currently doing) real-world software development and understand the difference between creating a nifty hack and making a useful tool.

But OLPC's biggest challenge is probably going to be translation, not coding and not deployment. As the suite of applications grows and as the number of locales to which the XO ships increases, the problem of localization will grow proportionally.

One way to limit the damage might be to publish a "vocabulary list" for programmers: use only these words / phrases / terms in your user interface unless you absolutely can't. If the number of words needing translation is limited somehow, then the translation job will remain managable. And having a standardized list of terms (for example, "not understood" instead of "illegal format", "incorrect data" or "syntax error") will make the software feel like it's SUPPOSED to go together.

If OLPC survives the transition from "blue sky" to "ohmygosh we actually have to ship boxes", the Sugar interface WILL improve, the bugs WILL get fixed and the code base WILL grow. The "market" (yes, even a non-profit, educational organization has market forces to contend with!) will not accept "alpha-quality" software forever.

I'm not sure I agree with your assessment of needing 300 - 1000 people to pull this off, but I agree that 30 people is probably too small an organization for what they're attempting to do. But I wish them all the best.

Gentlemen, I could not agree more - software needs to be at the operational level of the hardware, no matter its innovation.

One aspect of implementation is listening to your customers (the Ministries) as much as your target audience (children) and conforming your product and services to their needs. A strong Sugar code base and an easy translation process, are two aspects of that conformation. So would a realistic maintenance plan, integration of Activities into national curriculum, and an honest conversation of the costs and resources required.

Let's hope all of this and more is happening.

"Amateurs study tactics, professionals study logistics" - Omar Bradley

Moving Walter Bender and his enormous talent into the Deployment operating area makes a lot of sense.

I agree with all this Wayan. OLPC only has has 30 employees? That's absurd when you're talking about something that is supposed to sell millions of units and change the world. Is sheer lack of manpower behind the troubled rollout of G1G1?

I think I was more floored with what Teknico wrote on the News Forums:

"A source within OLPC, who asked to not be identified, reports that people within the project have a low opinion of Krstic's work, and that it is gradually being reworked, including BitFrost, replaced by something simpler called Rainbow. They basically say: 'Good riddance!".

So it looks like things ARE being worked on (for the better side of things). Its only a matter of time... in the meanwhile, the non-programmer people need to work on what Wayan was talking about in the article. Not getting bogged down in technicalities, but actually getting this out to children in 3rd world countries. After all, a global education project is the main mission (and hopefully priority). Bugs or not, nothing is perfect. And the journey of a thousand miles begins with the first step.

I ordered a XO 31 Dec 07 which has not arrived. From all the reading, etc I have done, it appears to me that OLPC needs someone experienced in marketing. I help many villages at with farming but many are involved with education and need the XO.

Ars Technica has a doom and gloom post around Ivan's departure and the OLPC shakeup:

"The resignation of Krstić seems to indicate that OLPC's attempt to reorganize is in deep trouble, if hasn't failed. The future of the project will likely be placed in jeopardy if additional key participants decide to leave their posts. At this stage, it seems unlikely that OLPC will be able to fulfill all of its ambitious goals."

"A source within OLPC, who asked to not be identified, reports that people within the project have a low opinion of Krstic's work, and that it is gradually being reworked, including BitFrost, replaced by something simpler called Rainbow. They basically say: 'Good riddance!"

Unless someone with at least a basic understanding of security "says" this, this is the first step into disaster.

Security is hard. Security for elementary school children who cannot write is even harder.

I can understand why "Bitfrost it too complex" for XP. XP has only cosmetic pseudo-security layers of cruft available. So any "real" security would certainly be too complex.

But that means the children will be as safe as the owner of a rooted XP box: Not at all.


I don't put much credence in unnamed anonymous sources, especially one who joined the OLPC forum specifically to post that accusation. I hate to admit it but Irvin called it. Krstić left and they say the same thing they said when Intel left, "we didn't need him anyway."

I think I'm a little more cynical about this. Problems with Sugar and shipping aside, didn't Ivan use the word "ethics" in his "I'm outta here" post? When I hear a word like that used the way Ivan used it, I think there's more going on.

In short, I have a bad feeling about this.

In long, I agree, partially, with what Wayan says, though. I think OLPC does need to move more into "reality" and focus on making the XO more practical. I agree with the Sugar naysayers, to a certain degree. That OS needs to be less buggy, fast, or all the XO does is introduce what we all had to deal with back in the early 90s. The difference is, it'll be kids who will immediately lose faith in the machine if it gives them trouble. However, I think replacement is a bad idea. OLPC needs to grow, not lose it's heart and soul (aka Ivan and Walter).

The number of employees being 30, that's what I think I heard at one piont in a video or saw in an article I cannot remember when. And I think it's only half of those that are the software developpers. So I think the Sugar OS is being made by 15 employees and 1000 volunteer contributors. They had to start doing the OS kind of from the bottom up, cause no one else has ever made a super light weight Linux distribution for actual mass production.

From my experience of the Sugar OS, I don't think that it's actually buggy even though it does sometimes freeze if too many software and tabs in Opera are opened at the same time. So I guess some equivalent to ctrl+alt+delete to kill processes toclear memory could be nice, basically pulling up the main screen and more easilly killing processes then having to right click and click on the x button. I'd see it perhaps a good idea to provide an option to arrange the surrounding hotspots in a different way to have a different type of overview of surrounding hotspots, although that might be less reelvant when children mostly are just connecting automatically to the mesh networks. I'd think a button only pull in of the sidebar might be better than it showing up every time the mouse is by the border of the screen.

Finally getting instant resume from sleep mode, and going to sleep mode when losing the laptop or when pushing the power button. And more journal to online blog synhronization features might be a good idea.

So I don't think Sugar OS is alpha at all. It's more reliable then my Ubuntu 64bit desktop installation, and it definately is much more reliable and stable then my Windows XP laptop.

OLPC probably needs to hire 30 software developpers, implementors and translators in every country that OLPC is going to be deployed in. This way each local government also feel they have full time employed and respnsoible support teams provided by OLPC locally at each OLPC country around the world.

Another couple of software things I would suggest, is a USB to VGA output adaptor of the screen at 1024x768 or 1280x1024 resolution (perhaps it already exists, I didn't check yet), an FTP software, Adobe needs to be pulled by the ear to provide a better more optimized Flash 9 support. At least somehow playback the Flash videos from Youtube smoothly in full screen mode, so focus all the processing on decoding the Flash video in full screen and not necesarilly to display it embedded in webpages. (so when streaming flash is detected display a "Play the video in Full screen" button). I also think it should have a super easy to use software repository manager so all the users can easilly add/remove applications without using the Terminal and without needing to have root access to the machine. So all the software available in the customized and localized repositories would be digitally signed for security.

I forgot to say that if those few software features are be added and the overall software constantly optimized, then a 25 year old video-blogger like me could most probably be able to replace ones 2000 dollar Windows XP laptop with a commercial design of the OLPC laptop. In fact I typed my previous comment here using the XO keyboard so it's perfectly usable also for adults, but an adult design of the laptop could have a more regular keyboard without the waterproofiness. Otherwise I can also turn the screen around and use a foldable pocket keyboard and pocketable mouse to have this laptop feel exactly like my 2000 dollar Ubuntu 64bit desktop. So when I'm home I'd plug in a 19" screen cause I can afford it.

Anyways, even though the laptop isn't targetted at adults in the commercial developped countries, still making it suitable for business people, thus suitable for the politicians who make the decisions to import large quantities of the laptop, is one way to also make the project even more believable.

Nicholas Negroponte once said that he planed to replace his 3000 dollar Windows XP business laptop with an XO as his full time business laptop. I wonder if he is able to do it yet. It needs an aeroplane mode software or keyboard combination switch to be used on airplanes and just a few software features and I think it's there.

To be honest, I don't understand why a ramping up of production and distribution must be done at the expense of innovation, development and any existing 'brains trust' within the organisation. Can the organisation's goals not all be fulfilled by keeping all the development staff in some kind of a skunkworks capability, maintaining the core philosophies in which they believe by imbuing them in the qualities of the product itself, while another part of OLPC deals with the 'dirty' task of actually getting XOs to the kids on the ground?
Personally, I think they can.

The fact that there seems to be an undercurrent of hostility to the manner in which Kristic left, and all this conjecture about paradigm shifts within the organisation point towards a fundamental shift in philosophy and, indeed capability within the organisation. To some degree, my feeling resonates with Maddie's view that there's been some pretty large obstacle that has been encountered. Whether that be practical, ethical or personal, I can't really guess at, but inin any of these cases, I don't think that any radical surgery that the organisation goes through to continue operating efficiently need undermine the viewpoint or power of either position. Any future structure of the organisation should value the input of the people who have been running it up until this point. I can think of a few models under which such a compromise could be achieved.

Having said all this, I know very little about the project; I only recently became interested in it. Maybe the people who left just felt that they wanted to move on, maybe these rumblings aren't indicative of anything sinister.

Whether or not something big IS going on, and in either structural scenario if it is, I just hope that OLPC continues to run smoothly and well, and continues to do the good work that it has up until now.

@ Charbax
Regarding Professor Negroponte's aim to use the XO in lieu of an expensive business laptop.
At the AAAS conference in Boston, he mentioned in passing that he was often able, in meetings with various people, to pick up Wi-Fi signals where other people were not. I presume that, if he's using it in meetings, then he's probably using it fairly often elsewhere.

Having said that, maybe he's just doing so to make an impression.