Why Microsoft and Intel tried to kill the XO $100 laptop

Tag : Tool For Children

Yet, 3½ years later, the laptop is clinging on to life. Itcosts around $190 rather than $100 and it is called the XO. It isno longer like a tent, but it can still be solar-powered. It is atechnological triumph. But only 370,000 are in use and another250,000 ordered. One Laptop Per Child (OLPC), the company formed torun the project, is still driven by the same old idealism, geekeryand technical brilliance. But Negroponte and his young staff areolder and wiser. They were stunned by the savagery of thecompetition they faced – competition plainly intended todestroy a philanthropic idea. “I had wildlyunderestimated,” says Negroponte, “the degree to whichcommercial entities will go to disrupt a humanitarianproject.”
For three reasons the XO turned out to be a gross provocation tothe big players in the computer industry. First, it was alwaysgoing to be cheap, undercutting the competition by thousands.Computers only cost as much as they do because the makers of thesoftware – primarily Microsoft – go to enormous lengthsto make their products necessary and expensive, and because makersof the hardware are constantly adding new features that youprobably don’t need.
In fact, electronics have plummeted in price and there’s noreal reason why you can’t get a decent laptop for a maximumof $400.
Second, the XO uses an AMD chip. The monopoly chip-maker in theworld is Intel. It has three-quarters of the market, with AMDsecond. AMD and Intel hate each other with a hatred as hard as thatof Hamas and the Israelis. For Intel, the idea of hundreds ofmillions of AMD laptops out there was intolerable. Intel could losetheir market leadership – but not if Agnes has anything to dowith it.
Third, it does not use software by Apple or Microsoft. Instead, itis run by Sugar, a free operating system devised by geeks for thelove of it. For Microsoft in particular this was also intolerable.Its Windows operating system is the industry standard.Apple’s system is much better, but Windows, through sheerMicrosoft muscle, has been made to appear necessary. The newmassive non-Windows user base threatened by the XO is the sort ofthing that seriously cuts into Bill Gates’s me time.
“This was a project that could operate outside the regularbusiness world,” says Ethan Beard, a former OLPC board memberrepresenting Google, one of its backers, “and that’snot an unreasonable expectation. But it is in some ways threateningto businesses and when you threaten businesses, especially verylarge ones, they are going to react in ways that hurt you.”
So the big boys stamped on the fingers of the XO. Intel called it agadget and then made their own cheap laptop, the Classmate, whichthey sold aggressively against the XO. Microsoft’s Gatessaid, “Jeez, get a decent computer…” and thenwent around trashing Negroponte’s earnestly well-meaningmachine.
“He said that sort of thing privately to people Iknew,” says Negroponte. “There was a fair amount ofthat. I was annoyed enough to say so, and he apologised for it– a lot of good that did.”
Gates’s reaction was especially tasteless. Apart from being– like, apparently, everybody else rich, powerful or famous– an old friend of Negroponte, he is the greatestphilanthropist in the world. But even though he’s steppeddown as the head of Microsoft, he remains almost paranoiacallydefensive of Windows.
Yet, miraculously, in spite of all this the XO is still alive,clinging to the cliff face. But for how much longer?
The Massachusetts Institute of Technology (MIT) is geek central,the greatest school of its type in the world. Along with Harvard,it dominates the city of Cambridge, across the river from Boston.MIT is a city within a city consisting of the campus but also thepenumbra of gigantic, architecturally eccentric buildings in whichlive the high-tech companies that graze on the institute’stalent. Around Kendall Square, people shuffle along the streetslost in thought. In the cafes they read and write notes on yellowlegal pads. This is where money meets IQ. It is not quite likeanywhere else on Earth.
It was here that, in 1985, Negroponte founded MIT’s Media Labwith the idea of “inventing a better future”. It was anidealistic attempt to show that the computer and communicationsrevolutions could, indeed, make the world a better place. The XO– high tech for the poor and unconnected – was theperfect embodiment of the Media Lab’s idealism.
In the OLPC boardroom in an office in Cambridge I meet Negroponte.He is a staggeringly well-connected Greek-American from a wealthybackground. The background and the connections give him a slightlyremote, slightly languid, though not pompous or grand, air. Hisstaff at OLPC regard him affectionately and with some bemusement.
A techno-utopian by nature, MIT is his natural home. And it was atMIT in 1968 that he met Seymour Papert. Papert had worked with thegreat educational theorist Jean Piaget. From Piaget’s work,Papert had developed the learning theory of constructionism. Putsimply, this means that children learn most effectively when theyare doing things rather than just sitting and listening. Negropontebecame an enthusiastic constructionist. It synched with hisworld-transforming view of technology. Computers were to be theperfect constructionist tool, allowing children to discover andmake things on their own. If Negroponte is the father of the XO,Papert is its grandfather.
“The question we were asking,” says Walter Bender,long-term Negroponte collaborator, “was not the‘how’ of computing but the ‘why’. And theprimary answer was learning.”
Constructionists tend to be sensitive creatures, primarily becausethey have been so angrily attacked. Children have to be toldsomething, say the critics: they can’t just be set free to doanything they feel like. Sane constructionists accept this, but, tobe honest, I do have a suspicion that they may be more geeks thaneducators. They want computers to work in schools because they likemachines.
Through various experiments, Negroponte and his colleagues zeroedin on the idea of the computer as the key that would unlock thepredicament of the world’s poor. In 1999 Negroponte built aschool in Cambodia.
Then, in 2001, he suggested his son go to it and, using a satellitelink and a few laptops, connect it to the internet.
“This was a very remote village – no electricity, notelephone, no TV – but the wi-fi was so well done that when Iasked myself if you look at the constituent parts they were allreplicable and in most cases the prices would scale down. The oneexception was the laptop. That became the focus…”
Yet laptops were expensive and they never seemed to get anycheaper, they just got more complicated. They have become loadedwith “bloatware” – over-featured, over-complexsoftware. “Everything becomes like an SUV,” saysNegroponte. “It’s crippling because, like an SUV, mostof the power in that machine is being used to drive the machine,not you or I.”
Pricey laptops meant that the Cambodian scheme was not“scaleable”. Cheap wi-fi connections could be scaledup, but then each child would need a laptop and there was no cheapway of providing these. This was a huge frustration to Negroponte,who firmly believes in constructionist learning through computersand that a connected world would be a better, more harmoniousworld.
But, I point out, it doesn’t seem to have worked so far. Ournew connectivity hasn’t made us significantly less evil.“It’s not working as well as it should because notenough people are connected at the moment,” he admits.
So the price of laptops was standing between this world and abetter one. By 2004, Negroponte was ready to do something about it.He asked Intel to provide a low-cost, low-power chip. “Aslong as you don’t call it a laptop,” said Intel, withthe tactlessness that seems to be a corporate policy. It meant theydidn’t like the sound of this new machine. “Did theyactually say this?” I ask earnest Agnes. No comment.
But, for months, Intel did no more than think about it. AMD, incontrast, said yes in a few hours. So, by the time of Davos,Negroponte was ready to announce the new machine with an AMD chip.“Why didn’t you give us a chance?” whined Intel amonth later. Remembering this, Negroponte laughs wryly: “Itwas like you get married and then your girlfriend comesback.”
For whatever reasons, Intel didn’t get it and AMD did.“I’m not sure,” says Dan Shine, a director ofAMD, choosing his words with the care of a Hamas spokesman at theUN, “[that] Intel viewed it quite as holistically as wedid.”
Negroponte had also hooked a spectacular range of backers: Red Hat,Google, AMD, Brightstar and News Corporation (the parent company ofThe Sunday Times). These each contributed $2m upfront and then afurther $500,000. All the companies put in money as sponsorshiprather than investment. This was, to the core, a pro-bono,philanthropic enterprise. OLPC, at Negroponte’s insistence,had been set up as a non-profit operation.
“It was probably the best decision we ever made,” hesays, “but we came this close to not doing it. I was advisedby absolutely everybody to make it a profit-making entity so wecould make lots of money and then give it away… But thenon-profit decision was important because it provided clarity ofpurpose – first, a head of state will talk to you becauseit’s about children and learning and not profit and,secondly, the best people will work for you for zero salary.”
Negroponte then went out to sell the machine. Connected as he is,he decided to use a top-down approach. He sold straight togovernments and heads of state. It seemed to work like a charm. Asif by magic, he conjured up promises to buy millions of laptopsfrom Argentina, Brazil, Nigeria, Thailand, Pakistan and Libya. Itwas, in publicity terms, a brilliant coup. From nowhere thisnot-yet-existing machine seemed to be conquering the world. Thepress lapped it up. Negroponte was on a roll.
Unfortunately, none of the orders materialised. “He would gofrom prime-minister meeting to president-of-country meeting andthat was his sales model,” says Rebecca Gonzales of AMD, whonow advises OLPC. “And it didn’t work, absolutely not.As we have learnt in the business world, just because you have ahandshake from the president or the prime minister, itdoesn’t mean you have an order.”
In fairness to Negroponte, his strategy had at least earned OLPC ahigh-publicity profile, and some of the non-orders were due tounforeseeable events – there was, for example, a militarycoup in Thailand.
“There’s nothing I regret about this strategy,”he says. “It created enough hype and pictures of Nicholasshaking hands with heads of state that, back in Taiwan where 250engineers were working on it, people felt part of something.”
Meanwhile, industry-changing technologies were being applied to thedesign of the computer itself. In its finished form there are threethings this computer has that all laptops should have butdon’t.
The first is its screen. This was created by OLPC’s chieftechnology officer, Mary Lou Jepsen. It is, first of all, cheap.Jepsen points out to me that the screen is the most expensive itemin any laptop and yet, for some reason, it is not normally includedin the hardware costs, so it gets overlooked. Secondly, it issuperbly readable in any light. It isn’t glossy orreflective. It is probably the best laptop screen in the world.
The second thing is mesh networking. This means if you have 10 XOsin a room, they can all talk to each other directly without goingthrough the internet. So even in an African village without wi-fi,the people could have their own intranet. Mesh also means that whenthey do have a wi-fi connection, its range can be massivelyextended as the mesh picks up the signal and rebroadcasts it. TheXO has probably the best connectivity of any laptop in the world.
And third, it probably has the lowest power consumption of anylaptop, essential in environments where power is at a premium.
The hardware, in short, is superb. The software, however, hasproblems. Sugar is a development of Linux, a free, “opensource” – more of that in a moment – operatingsystem.
It is tuned specifically to the needs of children and they, I amtold, pick it up quickly. Adults who have any experience ofcomputers, however, will find it hard. It has also been quitebuggy. And the user interface (UI) – the way you interactwith the machine – is very hard work compared with Windowsand Apple. “It can take years to get a UI right, so startingfrom scratch for me was a question,” says Ethan Beard.
All of which was bad news for the computer when it first appeared– it went into production in November 2007. A review in TheEconomist in January said “the implementation of thetechnologies is terrible” and described the Sugar operatingsystems as “cumbersome”. The Economist being read byprecisely the sort of people who might buy this machine inquantity, this was catastrophic. It was also not, given the qualityof the hardware, entirely fair.
And the truth of the design is that it had to be done that way toget the price down. Both hardware and software had to be rethoughtfrom the ground up. This is most difficult with the software in thetime available, because good user interfaces takes many years todevelop. Apple – the industry leader in this area – isonly as good as it is because of three decades of development. ButSugar was free, and buying Windows off the shelf – Apple doesnot license its system – would have almost doubled the priceof the machine.
There is a further element to all this which is crucial tounderstanding the idealistic, visionary zeal of the geeks whoworked on the machine. From its beginnings in the 1970s, thepersonal computing revolution has been suffused withcountercultural idealism. Apple was born of the conviction that the“people” should have computing power and IBM, the thenbig player, was the corporate beast that was not going to provideit. Even now it, too, is a corporate beast, Apple still marketsitself as the countercultural alternative to a Microsoft-poweredmachine. But the ultimate countercultural gesture is “OpenSource” software.
With Apple remaining a minority brand, Open Source is the biggestthreat to the dominance of Microsoft. Buy Windows Vista Ultimate– the latest Microsoft operating system – at PC Worldand you will pay £230. Vista is not popular, but hundreds ofmillions have to have it. Linux, the Open Source operating system,is, if you are geek enough, free. No wonder Microsoft said from thebeginning that they couldn’t back an Open Source machine likethe XO – they would be promoting their biggest weakness.
Linux is not just software, it is a countercultural movement whosemost fervent adherents believe in the overthrow of the Microsoftmonopoly. Sugar is based on Linux and Sugar’s greatest loveris Walter Bender.
I meet Bender in the Media Lab. There is something shy about him. Asenior figure at the Media Lab, he left OLPC. He was its mosthigh-profile departure. The reasons are disputed. Some say he wasjust too awkward to work with. But his reason is clear. OLPC haddecided to produce a new “dual boot” version of themachine – this means it can either run Sugar or Windows. ForBender, this was a betrayal of the Open Source faith.
Open Source allows users to change any or all of the software, to“drill down” into the very depths of the machine. ForBender, this makes it more true to the constructionist faith thanany proprietorial software. Children can remake the XO from thebottom up, impossible with a Windows machine. “I left becausethe future of Sugar was going to be bigger and bolder than justbeing confined to the OLPC laptop,” says Bender.
The question raised by Bender’s departure was: is OLPC anopen-source crusade, or is it a project to spread computing to thepoor by any means available? In practice, OLPC has answered no tothe first and yes to the second. But, if it is just about spreadingcomputing to the poor, then is the XO itself that important?Wouldn’t any cheap laptop be just as good? It is thisquestion that lies at the heart of the most spectacular crisissurrounding the project: the war with Intel. And here we come to mynew best friend Agnes.
Microsoft may have used words and a refusal to co-operate as itsweapons against the XO; Intel used brute force. The companydominates global computer hardware in the way that Microsoftdominates the software. And, like Microsoft, it is a fierceprotector of its ascendancy. So fierce, in fact, that the FederalTrade Commission in the US has recently opened an investigationinto its alleged anti-competitive practices designed to shut outAMD. On the academic side of the OLPC project, they were shocked bythe ferocity with which Intel attempted to kill their product. Onthe business side, they just shrugged and they all said the samething: “It’s in their DNA.”
Intel’s response to the XO was the Classmate. It is nothinglike as radical a machine in that it is, basically, astraightforward Windows laptop. Intel will tie itself in knotsrather than admit its laptop was a response to OLPC’s.
My Intel spokesperson, Agnes Kwan, seems to exist to evade theissue. I played e-mail ping-pong with her over several days. Shewas trying to avoid giving me any dates that would show theClassmate came after the XO. This included sending me a bizarre andbarely literate “ethnographic” study of computing inthe developed world. In the end, all she would say about thetimeline of the Classmate was: “It’s hard to pinpoint astart date with the nature of ethnographic research in whichethnographers collect data over a long period of time.”Sorry?
Many in the industry says the Classmate was intended to be an XOkiller and that’s how Intel behaved. Their formidable globalsales operation charged into any market in which OLPC might get afoothold, trashing the XO and pushing the Classmate. Nigeria, whereNegroponte had one of his handshake deals with President Obasanjo,was a typical example. In August 2006, Craig Barrett, Intelchairman, wrote a hard-sell letter to Obasanjo asking for a meetingin which he could explain their World Ahead programme, “whichis chartered to extend PC access to the world’s next billionusers”. This programme had been launched in May 2006, 15months after the OLPC announcement at Davos – bit of a deadgiveaway there, Craig. Barrett’s letter was backed up bydocuments listing “the shortcoming of the OLPCapproach”.
These documents having been leaked, they became a significantembarrassment to Intel. Here was a mighty company trying to crush aphilanthropic project. In May last year they seemed ready for atruce and a deal was done. Intel would join the OLPC board, invest$6m in the company, there would be moves to put an Intel chip inthe XO, and there would be no more slagging off of the XO in themarketplace. The deal failed with almost Middle Eastern speed andfinality. Intel attended only one board meeting and Intel salesmen– “it’s in their DNA” – carried onslagging off the XO. Intel also tried to parcel up the world intoeasy markets for Intel and hard ones for OLPC.
“You mean,” says Negroponte of this phase,“Ethiopia is mine and Mongolia isn’t?”
At the same time, Negroponte was demanding Intel stop marketing theClassmate. Intel refused on the basis that there was room for aplurality of solutions to the “digital divide”. On thisissue – says Agnes – the deal collapsed and Intel leftthe board in January. Even the departure was contentious.Negroponte said there was a deal to say nothing until there couldbe a joint announcement. But, of course, Intel went ahead and spoketo the press anyway.
“It’s quite obvious,” says an OLPC spokesman,“that they waited until very late in the day to make itnearly impossible for OLPC representatives on the East Coast to gettheir side of the story in the ‘first stories’.”Bruce Sewell of Intel e-mailed Negroponte to apologise, saying“instructions were misunderstood internally”.
I put all this to dear Agnes. No comment.
Destructive as all this sounds, it represents a kind of success forOLPC. First, whatever Intel tries not to say, it is almost certainthat the OLPC inspired the Classmate and cheap computers fromothers. Furthermore, as many on the business side of OLPC pointedout, the very fact that giants like Microsoft and Intel werebothering to trash the XO indicated the power of this idea to getunder their skin. “If Nicholas hadn’t said what he saidin January ’05,” says Dan Shine at AMD in Austin,Texas, “this machine wouldn’t be here and a lot ofother technologies and discussions wouldn’t be here. Heaccelerated people getting access by probably years.”
And, finally, however “impure” it may be to the opensourcers, putting Windows on the XO was a huge breakthrough in thecomputing industry because Microsoft has let them have Windows XPfor $3 per computer. One of the previous industry certainties wasthat Microsoft never ever sells anything cheap.
So, whatever happens to the XO, OLPC has changed the industry. Thequestion then becomes – what will happen to the XO? A newOLPC machine, which is configured more like an electronic book, isdue out in 2010 and, meanwhile, the XO is making inroads in LatinAmerica and there should soon be one million in the hands ofchildren in 16 countries. Sweet Agnes can only say that theClassmate has sold “tens of thousands”. If the war withIntel is to be won by sales, then OLPC is well ahead.
Palo Alto is the Californian equivalent of Cambridge’sKendall Square. This time the great IQ warehouse is StanfordUniversity. The streets of the town are, in a Californian way, lessintense than those around MIT. But the corporations are here, notleast Facebook, the vast social-networking site. In one of theirmany buildings, I meet Ethan Beard, an alarmingly young, alarminglyvigorous man. When at Google, he was sent to oversee theirsponsorship on the OLPC board. Now, looking back, he thinks moreradically than anybody else about the future of educationalcomputing for the developing world.
“They could keep on coming up with innovations and licenseout the technology, take the money and fund OLPC. Or they couldopen-source the entire design of the computer.”
Or, he suggests, the whole system could be put into a“cloud”. Cloud computing means your machine does verylittle except contact the internet – everything else is takenfrom applications and storage in cyberspace. Beard thinks the wholeOLPC project could live in the cloud, freeing it from the bonds ofthe heavy hardware earth through which it trudges.
But whatever future emerges from the heads of Nicholas, Dan, MaryLou, Ethan, Rebecca, Walter or even Craig, Bill and the divineAgnes, the simple fact will remain that OLPC has been a nobleattempt to do something the industry would never have done withoutprovocation.
Computers are like drugs, literally. If the drug companies wantedto do the most good in the world, they would divert all investmentfrom the illnesses of the rich – cardiovascular disease,cancer, diabetes – to the much more catastrophic ailments ofthe poor, primarily malaria, but also Aids. But they don’t;they sit comfortably on their high-margin drugs. Equally, if thetechnocrats really believed in the human value of universalconnectivity – and all of them say they do – they wouldfind ways of wiring southeast Asia and Africa. But theydon’t; they sit comfortably on their high-margin laptops.
Or they did until Nick Negroponte, supreme prophet of digitalconnectivity, revealed a strange, tent-like object in January 2005at the World Economic Forum in Davos and, at a stroke, gave Agnesher job description.