It's a threat Microsoft can't let stand: the entire third world learning Linux as children, and growing up to use it. And Microsoft is going to get its way.
It comes after a sudden wave of SCO-like problems for the OLPC project. A specious patent lawsuit over keyboards. Board-member Intel thrown out of the project for attempting to convince national governments to drop OLPC purchases and go with its own (Windows) product. First, OLPC is shown what its problems will be if it doesn't cooperate with Microsoft. Then, Microsoft approaches with money and technical help - you just have to run Windows to get it.
The move is presented as enabling choice. It starts out with a dual-boot capability, provided by Microsoft engineers. Not that any work by Microsoft was really needed, Open Source firmware that boots Microsoft operating systems has existed for ten years. Microsoft says they will issue guidelines, and start field trials this month. Dual-boot sounds harmless, but Microsoft's version of choice is better stated as we'll give you choice and then make you choose Microsoft. I'm sure there will be pressure on national governments to select Windows-only loads for their OLPC purchases, or to specify texts protected with Windows DRM for classroom use.
Nobody can pretend that the world has ever been absent any choice to run Microsoft software, or that Microsoft must work with OLPC to increase choice. Microsoft operating systems are the only option offered with the vast majority of desktop and server computers. By refusing to tolerate hardware that runs another OS by default, Microsoft is working to reduce choice.
Consider how good it might have been for the third world to have a computer infrasturcture they could support on their own, without any capital and technological drain to the United States. That's what they'll be losing. But that was never the goal of the OLPC project. It's meant to bring free e-Books to students, at a lower cost than their national governments could sustain. With OLPC based on all Free Software, it was likely that those books would have themselves been under similar licensing like Creative Content. Now, it is likely that third world students will be running DRM-locked textbooks that are only acessable under Windows.
Nicholas Negroponte has always been willing to go where the wind blows: the original OLPC prototypes ran Debian, notable because it's produced by a public-benefit non-profit. Once Red Hat offered money and resources, Debian disappeared from the system. Now it's Red Hat's turn to disappear.
The folks I have the most sympathy for are those students who might have been offered a way to take control of their own destiny, and make their nation self-sufficient for the IT infrastructure they need to participate in worldwide trade. Now, they'll get less. But I also feel sympathy for the many Open Source developers who participated in OLPC, and will now see their work discarded or perverted to support Microsoft.
Microsoft's exact quote is insideous. [Emphasis mine.]
“Microsoft plans to publish formal design guidelines early next year that will assist flash-based device manufacturers in designing machines that enable a high-quality Windows experience. In addition, there will be limited field trials in January 2008 of Windows XP for One Laptop per Child’s XO laptop. Microsoft’s goal is to provide a high-quality Windows experience on the XO device; if this is achieved, then Windows XP for the XO could be available as early as the second half of 2008.”
The specs that Linux will run comfortably on, or can at least be configured to run comfortably on, aren't enough for Windows. It is too fat. So, instead of slimming down XP even more, they want to fatten up the hardware.
One of the biggest is the fact that the XO has no hard drive and only 1GB of built-in memory. The company concluded it needed at least 2GB of memory just for Windows and Office, so it convinced the OLPC folks to include an SD slot on the laptop's motherboard.
In all honesty, adding an SD slot isn't that bad of an idea. One of my biggest gripes with super-slim systems is that RAM isn't expandable. Spend the extra few pennies and make it an SO-DIMM socket instead of surface mount. Or, add a damn slot.
But Bruce is right. In this case Microsoft is the heroin dealer wanting to make darn sure they can give out free, or very low cost samples, in potential new markets. Hook 'em while they're young.
I agree that this is Microsoft's strategy. But it remains to be seen if this will work. "Hook 'em while they're young" didn't work for Apple in the 1980s. I see OLPC as the Mosaic/Netscape of this decade.
Back around 1994, NCSA Mosaic started catching on, causing Internet access to move from the physics department into every part of campus life. At the time, Microsoft was working on Project Blackbird, an online service to compete with AOL and Compuserve. Within a few years, Bill Gates (the man, not the anthropomorphism of Microsoft) realized that the Internet would embrace and extinguish all the online services, and killed off Blackbird. Microsoft quickly found itself caught in a battle to justify its OS monopoly. On the one hand, Netscape was being bundled with Windows, and threatened to replace Windows as developers' platform of choice. On the other hand, Internet Explorer (initially based on code from NCSA Mosaic) allowed Microsoft to have a foot in the door, but developers would continue to flee from the Win32 API.
A decade later, nobody writes new applications on Win32. Microsoft has made numerous attempts to stem the tide, and has some traction in C# and .NET, but primarily for developing the server end of web applications. These days, if your code isn't usable from a Mac or Linux, that's a bug rather than a platform choice.
The XO is pointing Microsoft and Intel at its future market: cheap, portable networked devices in low-income countries. Once again, it's not a competition that Microsoft can win. To have any market share, prices must be competitive with Linux. (Not free, but affordable enough to justify the cost.)
At the same time, the XO is not the actual competition. Cheap, web-enabled cell phones are. The XO just looks more like what Microsoft and Intel want to sell (a cheap laptop) than what people will really buy.
Microsoft can pay off corrupt governments and bamboozle beurocracies against a struggling non-profit. But that's not how the global cell phone market has worked so far, and it will have a lot of well-financed competition to fend off-- including Intel, which is a Linux-backer because it needs to sell a huge volume of chips to pay for its next-generation fabs.
Cell phone infrastructure is expensive, especially for large, low population density, war torn, poor countries.
But other than that, I agree with your premise. There is no way M$ can win this fight. If they beef up the hardware to run even the leanest possible MS OS and Office bundle, it will be too expensive for the intended distribution. Period, end of statement. Throw into that the huge maintenance cost of keeping a bunch of Windows laptops running, let alone the additional bugs and problems caused by a hastily 'stripped down' version, and pretty soon, even if M$ hugely subsidizes the effort to keep the price competitive, you are still going to just have a lot of not so very good bricks to shore up your mud hut walls with.
The problem with Windows installations is that they don't 'just work'. Any version of Windows is filled with bugs, many of which tend to increase self-corruption (anything with the dreaded registry). As they are discovered in use, updates will have to be distributed, and those will introduce further bugs. It's the way M$ has always operated, since DOS 3.11 (Yes, I pick that version for a very specific reason), and there is no reason to think they could change now, even if they wanted to.
It can be argued however, that a sufficiently lean Linux could be quickly developed with essentially zero bugs and zero probability of self-corruption. It will 'just work', right out of the box, with little to no software maintenance ever needed.
M$ may succeed, by these efforts, in killing the entire OLPC movement, but it can never supplant it successfully.
Cell phone infrastructure is expensive, especially for large, low population density, war torn, poor countries.
Large, low population density, war torn, poor countries are where cell phones make the most sense! No need for running expensive copper wiring everywhere, just a cell tower every few miles, run by solar power, connected by directional microwave. Does it get any cheaper?
In a sense, 3rd world countries are experiencing a technological boom based on cellular technology - they are effectively leap-frogging the 1st world in some cases by only using cellular phones, avoiding all the capital-intensive telecommunications equipment we take for granted here.
The current sugar image takes only 337332 K blocks, or roughly 33% of the available 1 GB flash.
Microsoft wants > 1 GB of flash anyway, so what is most sensible is to have it bootstrap off of the internal flash, with most of the Microsoft storage on a 2 GB SD-card ($12 in quantity 1).
So to coexist between the two, one needs more solid state memory, but since there is a built in SD slot, this is easy to add. The SD card is probably able to do 10 MB/s bandwidth (x66), might be 20 MB/s (x133), which is pretty tolerable really as the on-board flash, although faster, is still only 30 MB/s, and both are actually really good for random access.
On the other hand, the current OLPC software is excretably bad, and it may be unfixably bad: a paradigm which has neither files nor the notion of persistant apps like the palm is very awkward to use. Even an elementary student will get quickly frustrated by the "journal" as a linear record and the only way to access history.
Having an XO, I'd rather run WinNT 3.5 on it over the Sugar UI.
The new UI also means that existing Linux apps need so much porting you might as well call it a new OS, rather than a Linux distro, so you don't actually benefit from the existing application base that Linux has. Thus you have a gimped web browser instead of full Firefox (which is definatly worth the 40 MB footprint compared with what is there).
Not to mention suffering from Linux's still amazingly cavalier attitude about power management, which is something Windows has actually done a very good job on over the years. I would not expect a Windows laptop to drain the battery 50% in a day of almost complete unuse.
If the goal is to provide the maximum benefit to the maximum number of users, the Microsoft future tax may very well be worth paying, if this will get a usable system in the hands of more people. Because if the project succeeds, it needs usable software (something it currently does not have) and if the project fails, it doesn't matter who provides it.
If Red-Hat, Apple, Microsoft, or Satan himself is willing to provide usable software, so be it. Without usable software, the OLPC system will die, and "Self sufficient in IT infrastructure" will be the least of the concerns. (With usable software, it may still die, but at least it isn't a guarenteed death.)
I would not expect a Windows laptop to drain the battery 50% in a day of almost complete unuse.
Well, is this with the cover closed? If you have suspended to RAM, you have to keep refreshing the RAM. This would not be a Linux issue. If you have suspended to FLASH, there should be no power flowing with the lid shut unless you are supporting some form of network wakeup. Even systems with alarm clocks only power the clock to do that.
We have good small-footprint desktop software with a conventional file paradigm. It makes sense to not give kids an easy way to lose their work, but surely you can add files to the current software if necessary.
Sugar doesn't do suspend-to-flash, only suspend-to-ram (suspend-to-flash would require reserving 256 MB of flash, or 1/4 of the whole capacity), but even sleep-to-RAM shouldn't burn that much power. The design is for a very low power system, with most interfaces disabling very agressively and even when undisabled, the power budget is very low. It really should burn almost 0 power (except the backlight) unless you are banging on keys, and the hardware supports this.
But the linux distro doesn't take advantage of it. For example, it doesn't even go to sleep when you close the lid! Rather, the backlight stays on until it finally decides to shut off on its own.
Have you actually played with the current XO and its software? The hardware is brilliant (its tough enough that it would almost be Mil-Spec without even trying, and the screen is a work of art), but the software truely is awful. Its really ill-suited and awkward.
For example, kids don't actually have an easy way to SAVE their work, let alone lose it in a filesystem later. And it is a single no-ack click to nuke your work even if you can find it again (and if you remember to save it, which you have to explicitly do, again, no-ack on quitting without saving.) And its easy to duplicate-name activity, so you have to remember which TIME refers to the state you want, which makes it even easier to lose your work.
Look at it this way, it doesn't even have an easy way to save your bookmarks in the web browser, as when you quit the web browser, you lose all your state (so you have to save the activity, and restore that activity from the journal instead, rather than clicking on the web browser app).
I don't think this can be fixed, because it really is designed around a "No such thing as a file" paradigm. You'd pretty much have to change how everything works to add in the notion of files or persistent applications.
Also, by being a "new" os effectively, it loses out on many critical applications. There is no spreadsheet, no email application, no browser supporting Flash or Java, no VoIP, no word processor which support the de-facto stardard formats (as much as you hate it, .doc is a standard). All of which come for free if you install either Windows or a real linux distribution.
If the choice is Sugar or Microsoft, Microsoft wins, hugely. The only alternative is to discard Sugar and switch to Gnome or KDE, and the cost of grossly simplifying those to run on the available resources as well is going to be substantial, more than the OLPC foundation can afford.
Microsoft would not be a disaster, for the OLPC foundation or for the world. Rather, Microsoft is the best thing that can happen right now, as Microsoft's participation would turn a toy into a tool.
Well, it sounds like they need a systems programmer. Making suspend work when you close the lid isn't a big deal if there's a lid switch. If ACPI works you can do it all with a text editor and without recompiling anything. If there's no ACPI you need to make a special driver. I do this stuff for a living.
At one point people were raving about the software. I'm sorry to hear such a bad report. There were good programmers (like Mako Hill) on the project at one time. He's moved on, maybe things aren't so good now.
My thoughts exactly.
Also, why the "no files" paradigm? Why not have them speak the same language as the rest of us?
So they get a copy of edubuntu with a custom power management thing and there u go.
Right?
There is actually a double switch (magnetic), so it can tell if it is open in book mode, computer mode, or closed.
The problem is the power management requires more than just ACPI for the XO. The hardware supports 100ms sleep/awake operations on the CPU (the frame buffer is separate and part of the LCD display controller) and sub-computer shutdown. When the power management is fully working, you should see ~2W power consumption when say, reading text on the screen, as all that runs is the dispay and the mesh network WiFi. That, and I thought ACPI was still "experimental".
To their credit, I think Update.1 fixes a lot of the power management issues.
It also comes down to that designing a GUI and full app environment from scratch or even semi-scratch (OS, dev tools, and mid level graphics primitives as the starting point) is significantly harder than designing the hardware. Further complicated by trying to reinvent the GUI didn't help either.
With such a small budget, and such a huge initial emphasis on hardware, the software has suffered. The toy apps (eg, the music aps) are a ton of fun but the real apps (browser, word processor, pdf reader, filesystem) suffer or lack completely (mail) because the environment is so limited.
In contrast, all but the screen are actually pretty well established technology, so the hardware is a good combination (an amazingly superb combination, in fact. Intel so screwed up on the Classmate hardware its not even funny) of existing technology and even the screen is a very clever spin of two existing technologies.
So while the hardware was "combining known components in a new way, with a real budget", the software is really "new and novel from scratch, and without the budget". I think there is a mistaken reliance that "open source shal provide" on the software, which to date hasn't worked. Hopefully this will change.
I'm thinking that for mine, I'm going to see about installing Ubuntu over it, because with real Firefox it will be quite cool.
..completely fork the project back to pure open source, perhaps rename it to OLPP, or one laptop per person, with no discrimination to what a person was based on geographical area, income, whatever, just a decent self powered laptop that anyone could use based on them being a human being, zero discrimination in other words. Make an adult sized version, keeping the nifty features like built in ruggedness, instant mesh networking and so on, the self powered bit, all of that, running open source, and perhaps do it as some sort of huge buyers and builders co-op, even distribute it as a kit to be assembled. Probably be easier to distribute as a box of parts anyway, compared to a finished product, less legal hassles..maybe.
.
Just a thought..it would be a big project,, but perhaps successful...food coops certainly work along the same lines. buy and order in bulk in order to get the wholesale price, a rather simple concept after all. I think the interest would be there, look how many thousands of the buy one get one deal they did at 400 bucks and just in the US. Expand this globally.... If this was just the same 400 bucks just to get one, an adult sized and "beefier" one, that you know worked flawlessly with open source because it shipped with it installed, perhaps this is doable.
In other words, a workaround for the obvious hijacking of the project by MS and others plus it addresses a lot of the complaints we heard, you can't get it outside the US, too small, etc.
Ya, you can get cheap laptops for that money from some vendors, but you can't get a rugged instant mesh network self powered linux comes pre installed and works perfectly fine sized for adult humans laptop for that. but I think it could be done as long as you don't go nuts with the specs and shop around for the components and get enough preorders to justify the cost of wholesale purchases. I know I would like one, especially if it was financed through the project with the pre-order arrangement over some few months, one step at a time, say this month we get the screens ordered, next month the mobo, next month the case, next month the ram, and so on.
If there's any silver lining in this, it's that M$ will never be able to run Windows XP reasonably-well on the XO, let alone take advantage of the XO's custom capabilities (like Sugar/Fedora will). I can't even do that on a Thinkpad with better specs, that is if any necessary apps are installed (e.g. office apps). And I hear the same sort of things from others, even running one of the pared-down versions of XP floating around. Hopefully users will see the contrast between a smoothly-running OS and one that was slapped on at the last minute, that is if they're still allowed to use Sugar/Fedora by teachers etc.
Ivan, I saw one of the first OLPC prototypes at the UN World Summit in Tunis, Tunisia, in 2005. I was a guest of UNDP, and OLPC was showing in the UNDP area. It was running Debian.
Unfortunately, I don't believe that Microsoft's intent toward the OLPC project is at all benign. They will promote their OLPC software load for DRM-locked content with the help of proprietary publishers who are threatened enough by open content to throw some zero cost but DRM locked e-books to the third world. If they can get governments to commit to the DRM-locked content on your platform, a non-Microsoft OS is going to be out of the race.
Also, nobody who wants an open platform for the third world is being "religious" about it, promoting sound public policy is not religion. I'm really tired of hearing that old saw brought up, please stop it.
1/ Baylis, of wind up radio fame, dismissed the OLPC as a pile of crap, in my opinion not so much because when he was shown what they had at that early stage he thought what they had was crap, but rather because their methodology was crap and so they were doomed before they started.
2/ The link to the "OLPC Employee" blog reads like astroturfing, Bruce Perens is wrong, ner-ner, no proofs shown.
3/ I'm not a coder, but Debain is the no-brainer choice for these things, and spend the specific effort in power management, suspend, hibernate, etc, which linux generally sucks donkeys at.
4/ I'm not a coder, but to base something on Linux, in which everything is a file, and then present a UI, in which nothing is a file and nothing can be saved trivially, is the most monumental fuck up imaginable. George Bush could have done better....
5/ Negropronte is indeed a political animal, OLPC was a project headed by a committee of political animals and vested industry interests, nowhere in there was there a committed engineer or ten at board level, if you think such a set up is going to produce ANYTHING but a cock up then you have a 1k input buffer.
6/ I ran Win98SE + 98Lite in a 17 megabyte disk footprint all those years ago, so Microsoft can too, don't assume this is about porting vista, and don't assume the OS (in Microsofts case) is anything other than a vehicle to deliver the real money in other sales of other products.
7/ I'll lay money nobody at OLPC actually specified and designed the hardware actually shipped, the OLPC ended up x86 compatible at THAT was the point that the writing was on the wall
8/ Intel have been closed mouthed about this, while everyone else is bad-mouthing them, time will tell, maybe Intel pulled out simply to avoid being associated with a monumental failure...
9/ The more I read about OLPC and the more details come out, the more convinced I am that my initial opinions when they confirmed the x86 arch were correct, it is death by committee.
In terms of decent performance, wide compatibility, x86 is the only option. ARM doesn't have the software base and peripheral options, and the RISCs have largely died. There are several low power, strong performance x86 options (both from AMD and Via) that going with anything other than an x86 would have been a mistake.
Apart from the political decision to make the keyboard ONLY usable by children, and the lack of a mouse-control on the side of the screen for eBook mode, the hardware design is impecable.
The $100 price tag was a dream, but even $200 is an amazingly low price target that they were able to meet, the ruggedization is incredible, and the small touches (eg, replaceable backlight) are well thought through.
Logically, your claim is oxymoronic, the x86 arch pre-exists OLPC, ergo nobody at OLPC can have specified or designed anything x86... QED
Designing the logo or "designing" the colour scheme don't count.
In terms of decent performance, wide compatibility, x86 is the only option. ARM doesn't have the software base and peripheral options, and the RISCs have largely died. There are several low power, strong performance x86 options (both from AMD and Via) that going with anything other than an x86 would have been a mistake.
Again, claiming x86 is the only option doesn't make it so, Arm most certainly does have a software base, and since they started out with debian arguably they had the most felxibility possible, my old MIPS RAQ2's managed to run a wide range of application software, you could even wipe the Cobalt code and run debian on em.
Mobile phones are quite CPU intensive nowadays, they manage to run on non x86 arch and manage to provide code by avenues other than "if it ain't on the shelves as PC world, it don't exist" mentality.
Apart from the political decision to make the keyboard ONLY usable by children, and the lack of a mouse-control on the side of the screen for eBook mode, the hardware design is impecable.
I haven't literally had my hands on one in the flesh, but I've never seen impeccable hardware design in anything portable, except maybe a rock... it's ALWAYS a trade off.
The $100 price tag was a dream, but even $200 is an amazingly low price target that they were able to meet, the ruggedization is incredible, and the small touches (eg, replaceable backlight) are well thought through.
I don't see why 100 bucks is a dream, samsung can make mobes for much less than that, with startlingly high specs, and surprisingly massive profit margins.
You're doing the "deconstructing an x-box" to prove MS just has to be subsidising each one by a thousand bucks bit, it just ain't so, you make ANYTHING in these kinds of volumes and the price per unit drops through the floor... I was at one time privy to the factory gate cost of a *very* similar device to an OLPC, figured someone must have got the decimal point in the wrong place, realised I couln't FedEx it from the factory gates to me without doubling the price... one thing you can guarantee is EVERYONE involved in this is getting paid enough for it to be worthwhile, and that is a LOT of hands being held out.
Wouldn't surprise me in the least if the volume hardware cost, minus all markups all along the line, was fifty bucks US a unit at the factory gate.
AS for ruggedisation, well, like I said, I ain't seen one, but, I'll take a bet with you I can destroy one without even trying, and show you other electronics devices that will just shrug off the same treatment, ABS is ABS is ABS, full titanium casing with space frame construction and piano hinges can get you "incredible ruggedization" or for less money you can use silicone alloy castings, not what you'd call "light" though. BTW I have worked with this stuff, including the 1kg drop ball test on screens etc etc etc... Panasonic Toughbooks cost a lot of bread, you can't even stand in the same street for a couple of hundred bucks, believe me.
In terms of decent performance, wide compatibility, x86 is the only option. ARM doesn't have the software base and peripheral options,
Whoa there - if the XO was to run only open-source software, then it running
on an ARM chip would have just as much software as one running x86. Whole
linux distributions can be built for ARM. Now if you're talking about
proprietary software, point taken, but that's exactly the point of the
original poster.
But if you want the potential to tap into the huge ecology of open source software, as well as closed source, you go with an x86. EG, how much audio stuff for Linux is designed to work with the AC97 interface?
And the AMD chip used is excellent. The Geode LX is a ~1W typical power processor at 433 MHz with an integrated display controller and floating point (most of the ARMs out there in low-power-land don't have floating point), in a .13 micron processes, with pretty big (64K/64K L1, 128K L2) caches as well. Thus the penalty for x86 at that performance point is maby, perhaps, if you are being pessimistic, 1W.
And the hardware is really agressive in supporting power management: you can shut off the CPU and turn it back on in the time it takes you to tap a key, so with the right software, you only pay that 1W penalty when you are actually doing real work.
But that 1W opens up a huge world of potential compatibility, even in the open source world.
EG, even in the Linux world, the JITs for x86 are going to be better than the JITs for ARM or PowerPC. If you are building an embedded system, yes, something like the 350 MHz FreeScale used in the Chumby would be a good choice. But not for this.
Anyone who plays with the OLPC complains about app performance. This is because most of the environment is written in python. Which do you think the people building a Python JIT are targeting or will target in the future? Uhh, x86. You could never take advantage of, say, Psycho, if you went with an ARM or PowerPC core.
But if you want the potential to tap into the huge ecology of open source software, as well as closed source, you go with an x86.
If you stress the "as well" part, you're back to supporting the original poster's point, which is that x86 is better for proprietary software compatibility. I maintain it is a rough wash for open-source.
EG, how much audio stuff for Linux is designed to work with the AC97 interface?
There's enough - googling arm & ac97 shows up several supported chipsets.
And the AMD chip used is excellent [low power]
Well OK but that's not a software compatibility issue. There exist high-performance, cheap, low-power ARM chips - they almost define the genre.
Which do you think the people building a Python JIT are targeting or will target in the future? Uhh, x86.
That is a fair but rather unquantified point. And anyway you originally claimed compatibility, not performance reasons.
Even just the open source world, the x86 is the best supported. Anything that touches low level, be it a JIT, a dev-tool chain, garbage collection, or anything else targets x86 first and everything else second.
Yes, linux "supports" ARM and PowerPC. Which do you think is more tested? More robust? More used? How is that not worth, AT MOST, 1W during active computing (and nearly nothing when idle)?
Saying the XO's use of an x86 is a political decision is ridiculous. Rather, I would seriously question the project if they did NOT use an x86.
Building an embedded device? Yeah, Freescale all the way, they have some nice integrated ARM systems (although be damned if they give you power figures in their datasheets). But a computer to run computing software?
The only real "Microsoft meddling" is the SD port, because Microsoft wanted more then 1 GB of flash. And the SD card is the right decision anyway, because it adds a lot of flexibility as removable state for very low cost and respectible (10 MB/s) bandwidth.
The OPLC strikes me as a quintessentially utopian experiment. It's a tabula rasa, a blank slate. In Steven Pinker's, "The Blank Slate: The Modern Denial of Human Nature", Pinker discusses the ways in which people deny human nature for various utopian reasons, and the ways in which they all fail.
I have always thought Negroponte was a utopian. His OLPC is going to be the force that saves the world (and one thinks, brings down Microsoft..) The third-world children that are his beneficiaries are unsullied by the modern technological culture. Rousseau would have identified these "noble savages" immediately. I wonder, however rationalized, if that was the real reason why the OLPC's were not intended to be sold in the first world let alone America. No noble savages here, we're already corrupted by civilization.
These "noble savages", I mean, children, won't need teachers; they'll teach themselves! They'll fix laptops themselves--without even knowing what a screwdriver is (or so the hype makes it seem)! (I ripped apart my jack-in-the-box toy when I was 3, but I seem to recall reading LOTS of books when I became one of these tinkerers. Where are the books for these noble savages?)
The hardware is brilliant, but I admit I was turned off when I saw that crank in the early mockups. The AMD chip is low powered, but there is no way that it can be powered by a crank. And if you've seen one of Bayliss' radios, the crank-spring-generator assembly is rather large, and that for a relatively low-powered cheap radio. Not a single techie pointed this out during the OLPC hype. I wonder if that turned off Bayliss, rather than (or in addition to, his management concerns.)
The hardware is brilliant, but I admit I was turned off when I saw that crank in the early mockups. The AMD chip is low powered, but there is no way that it can be powered by a crank. And if you've seen one of Bayliss' radios, the crank-spring-generator assembly is rather large, and that for a relatively low-powered cheap radio. Not a single techie pointed this out during the OLPC hype. I wonder if that turned off Bayliss, rather than (or in addition to, his management concerns.)
Baylis is, if he is nothing else, a practical engineer.
When we was invited to see a "demo model" he expected to see a crude, working, prototype.
What he got shown was a bunch of mock-ups.
Being an engineer with a LOT of experience in getting electrical power out of clockwork devices, he KNEW that it simply wasn't technically achievable to get the levels of power that they ended up deciding they would be using, not even if the OLPC was a 900 lb gorilla whose body weight was 90% biceps and triceps.
OLPC is the computer world's equivalent of the Segway / ginger / it.
Or maybe the moller flying car.
OLPC only survived on hype because it was "open source doodz" so all the fanboys got behind it, most of the distaste now is simply because the anti-christ bill g is going to borg assimilate the trekkies OLPC.
This is gloomy, but I'm not sure that it marks the disintegration of the OLPC goals. I agree that there will be pressure from Microsoft on governments to opt for a Windows-only configuration when purchasing, this will likely come in the form of credits, cost reduction and (possibly) even ad shares.
The fact remains that the pains of running XP on something designed with sparse operation in mind will quickly be realized. The pains of DRM laced texts will soon be felt as well.
There is absolutely nothing preventing anyone with the background, credentials and support to produce DRM free texts in open formats that can be used on any OS. Regardless of what OS is on their OLPC, kids using them *will* have access to quality, free texts and information.
Microsoft can concoct lots of logical straw men that lead people to believe Windows is somehow better after getting lost in an angry fruit salad of fallacy, there is nothing that we can do to stop them from doing that because you can not document a negative and they know it.
You can not use this tactic against a quality textbook that passes peer review. A text is either accurate or its not.
The GNU/Linux operating system got its foot hold from people getting fed up with Microsoft products and installing something that worked better for their needs. OLPC will be no different, its just going to take longer due to the fact that a class full of students need to be using the same OS.
Bruce, I completely agree with everything that you've said, however I don't think we've come anywhere close to the end of this saga. Don't (yet) discount things like ReactOS, or the (near dozen) tightly engineered versions of Debian/Ubuntu that are being produced to run on these things.
In 2 years, you'll have a choice of a dozen operating systems that will run agreeably on an OLPC. Most of them will run rather well. Then you'll have Windows XP. I fail to see how this can't mean that people will be 80 - 90 times more likely to opt away from Redmond.
It may not happen as fast as we'd like, however I'm sure that OLPC will finally settle as a free and open solution for kids all over the planet.
Why Microsoft Must Control One Laptop Per Child
The move is presented as enabling choice. It starts out with a dual-boot capability, provided by Microsoft engineers. Not that any work by Microsoft was really needed, Open Source firmware that boots Microsoft operating systems has existed for ten years. Microsoft says they will issue guidelines, and start field trials this month. Dual-boot sounds harmless, but Microsoft's version of choice is better stated as we'll give you choice and then make you choose Microsoft. I'm sure there will be pressure on national governments to select Windows-only loads for their OLPC purchases, or to specify texts protected with Windows DRM for classroom use.
Nobody can pretend that the world has ever been absent any choice to run Microsoft software, or that Microsoft must work with OLPC to increase choice. Microsoft operating systems are the only option offered with the vast majority of desktop and server computers. By refusing to tolerate hardware that runs another OS by default, Microsoft is working to reduce choice.
Consider how good it might have been for the third world to have a computer infrasturcture they could support on their own, without any capital and technological drain to the United States. That's what they'll be losing. But that was never the goal of the OLPC project. It's meant to bring free e-Books to students, at a lower cost than their national governments could sustain. With OLPC based on all Free Software, it was likely that those books would have themselves been under similar licensing like Creative Content. Now, it is likely that third world students will be running DRM-locked textbooks that are only acessable under Windows.
Nicholas Negroponte has always been willing to go where the wind blows: the original OLPC prototypes ran Debian, notable because it's produced by a public-benefit non-profit. Once Red Hat offered money and resources, Debian disappeared from the system. Now it's Red Hat's turn to disappear.
The folks I have the most sympathy for are those students who might have been offered a way to take control of their own destiny, and make their nation self-sufficient for the IT infrastructure they need to participate in worldwide trade. Now, they'll get less. But I also feel sympathy for the many Open Source developers who participated in OLPC, and will now see their work discarded or perverted to support Microsoft.