Ivan Krstić Says Negroponte's Wrong About Sugar and OLPC 137
Posted
by
timothy
from the who-is-this-krstić-guy-anyhow dept.
from the who-is-this-krstić-guy-anyhow dept.
Not many days ago, we mentioned ZDNet's interview with Nicholas Negroponte, in which Negroponte had some harsh things to say about Sugar and its connection to the slower-than-hoped uptake of the XO. Ivan Krstic (formerly head of the OLPC's security innovative subsystem) responded to Negroponte's claims, which he says are "nonsense." Among other things, he mentions that Sugar "was the name for the new learning-oriented graphical interface that OLPC was building, but it was also the name for the entire XO operating system, one tiny part of which was Sugar the GUI, and the rest of which was mostly Fedora Linux."
Feature creep killed the XO (Score:4, Interesting)
In particular, trying to cram yet more hardware into it to meet the demands of the Microsoft lobby.
If they'd just made the widget, put it into production, and focused on the sales, they would have made a difference.
Re:Feature creep killed the XO (Score:3, Interesting)
Re:Alternative (Score:3, Interesting)
The DS is indeed awesome, I had my first encounter this year with it by buying a DSi for myself with several games for an extended work trip with many hours (with a colleage) in a car. I recall the days of the original game boy and how it didn't change for an entire decade, and this is light years from that... However, do you use something else to browse besides the opera DS browser? That is one software I found severely lacking on the DS. On a fast wifi connection, it's just painful.
And yes, I further set it so it doesn't download pictures, etc. And consider the latest DSi has 4x ram (and more CPU power) than the original DS/DSLite - 16mb vs 4. Maybe we go to different pages, but if Nintendo spared an extra dollar, it probably could have been bumped to 64-96mb ram and been a much better and actually viable browsing experience.
Otherwise, I'm happy.
I'm not sure that's fair. (Score:5, Interesting)
I'm not a big fan of Negroponte, but both Intel and Microsoft went out of their way to kill this project -- telling outrageous lies to potential developing-world customers in order to put them off it. When did either of them make a product with a fraction of the innovation and convenience that the XO exhibits? Negroponte did a deal with the Devil in order to keep the thing afloat, and it went wrong on him, as deals with the Devil usually do. But the fact that two gigantic for-profit corporations were so greedy that they were prepared to kill a charitable little startup just on the off-chance that it might deny them a few low-margin sales, is simply disgraceful. If they'd had any heart at all, they'd have said, "Great! How can we help?" and turned it into a big PR bonus for themselves.
Mind you, it doesn't surprise me coming from Microsoft; I've had dealings with them in the past. But I thought better of Intel.
OLPC is a success (Score:5, Interesting)
Thanks to OLPC, we have soon 50 million netbooks in rich countries.
Thanks to OLPC, children have soon millions of cheap lower power laptops in poor countries.
Thanks to OLPC, the PC/Laptop industry's interpretation of Moore's law has totally been reshaped, every 18month now PC/laptops will be half the price instead of 2x more powerful and with 2x more bloatware.
Sure, I would have been happier, and so would most other Linux geeks if OLPC had shipped 100 million laptops to poor children by now, and not just 1 million units. Reason for that not happening yet in multi-hundred million scales though are several:
1. Intel will do anything it can not to be killed off by a non-profit laptop technology revolution. Including abusing of monopolistic situations and corrupting politicians.
2. AMD is not much interested in helping OLPC succeed in lowering the cost of laptops and PCs. Lower cost also means less profits and margins for AMD, and AMD has enough problems with profits and margins as it is.
Looking forward, to reach those 100 million poor children sooner rather than later:
1. OLPC needs to find an alternative to AMD as soon as possible. VIA is planned for XO-1.5 which could hopefully ship a few millions of units in a few months time, if VIA supports this move of OLPC creating a cheaper and lower power market using their processor. XO-1.5 could reach the $150 pricepoint soon and enable dozens of commercial netbooks using the VIA processor and also copying on the way OLPC is using the VIA processor.
2. OLPC needs to implement the worlds best ARM processor based laptops for XO-2 working with Google to implement the so called Chrome OS on those. Cloud computing can work also for places without stable internet access, HTML5 supports offline web apps and offline databases. OLPC needs to push Google to make it work on WiFi Mesh networks as well. XO-2 can start at $100 when released and reach the $50 price point, when manufactured using any of half a dozen ARM processor companies chips. All of TI, Qualcomm, Marvell, Freescale, Nvidia and Samsung, all those ARM processors should fit in the XO-2 design. Competition will bring the prices down faster.
Re:OLPC is a success (Score:4, Interesting)
Though OLPC is launching XO-1.5 based on the VIA processor in the next few weeks or months as you can see in the videos on my http://olpc.tv/ [olpc.tv] Using this new lower power VIA processor, OLPC can speed it up 4x as well and still lower the cost and lower the power consumption.
You complainers about Windows support need to learn that it's BECAUSE OLPC is an open platform that Microsoft is able to port Windows XP for it. You are completely ridiculous not understanding that for OLPC to not support Windows XP, they would have had to build a closed proprietary system. Since specs of XO are opened, and it's X86 based, Microsoft is obviously able to read the specs on the Wiki and build a port of Windows XP for it. It's just plain stupid to keep asking for OLPC to somehow block Microsoft.
Give 1 Get 1 program was not a failure at all. Tens of thousands of laptops were given for free in dozens of countries. To create those dozens of hundred or thousand-laptop OLPC pilot projects. Those projects would not have been financed if it wasn't for the G1G1 program.
Now sure, you can critisize OLPC for not having found more money if you want. I find it that considering they are just a 30-employee non-profit, finding $200 million to fund those 1 million first XO laptops is pretty decent achievement no matter what. Sure, I'd prefer if they had access to billions of dollars to help millions more children get laptops. People in rich countries are greedy, they only care to pay for stuff that they can get for themselves.
Re:A lot of things combined to kill the XO (Score:2, Interesting)
Mesh networking is crucial to OLPC:
- Children in poor areas with NO internet connection can still collaborate on projects, share data.
- Children in poor areas with LITTLE internet connection, can all share the same hotspot thus providing much cheaper Internet access, down towards $0.20 per child per month. This works.
ARM Processors consume ALOT less power than X86. With ARM you are talking milliwatts of power used to run the laptop, not watts.
Re:entirely not the problem (Score:2, Interesting)
I heard Negroponte speak at a conference a couple years ago. One thing was evident: the OLPC was a lot more about Negroponte's ego than anything else. Nearly the entire speech was a list of all the powerful and famous people he hangs with.
The entire thing just seemed too complex to really work. All of the special built custom parts and new GUI around Linux seemed to add little to the over all value.
Just my 2 cents worth.
Ivan agrees with Nicholas, I don't get the fuss (Score:1, Interesting)
Where in Nicholas Negroponte's interview does it say he thinks that the core Linux hardware/software development was the mistake?
Where in Nicholas Negroponte's interviews does he say he thinks Windows support on the XO is better than optimized Linux?
Talking about working for the evil empire, I'd say Ivan Kristic working for Apple should not have too much to brag about.
He's a genius for sure, and the work OLPC engineers have done for XO-1 was simply amazing considering the very small amount of engineers employed by OLPC, but I simply don't get why Ivan doesn't simply recognize that an open platform like XO simply cannot and should not try to block Microsoft from doing whatever they want if they want to port Windows XP for the unit as well.
Simply put, how can Ivan be working like this on an OPEN X86 based project and then demand that Microsoft not be allowed to port their Windows OS to it?
Re:entirely not the problem (Score:3, Interesting)
Re:Feature creep killed the XO (Score:5, Interesting)
I think that I must be the only person who actually purchased an XO, because all of the reasons given for the poor sales are not pointing out the one problem that the OLPC XO had on launch: it was (and still is) sluggish. It is a pain in the ass to use, since doing things like reading PDFs is slow as hell.
My OLPC is sitting in my office unused because, as much as I wanted to use it to read PDFs and browse the web, Sugar is slow and doing things like moving from page-to-page in the reader take a looong time.
On the development side, did the Sugar APIs ever get mature enough that the documentation was stable? Is it really ready for third parties to write software in Sugar without having to worry that large sections of their code will have to change on the nest upgrade? Looking right now at the docs, there are still parts of the code that do not have stable APIs.
How can you take a sluggish device with moving APIs and expect to sell it to countries on a large scale? Will governments really be willing to spend millions of dollars on something that is clearly unfinished design-wise and second-rate?
Microsoft and Intel did not kill the OLPC. The OLPC was enough to kill itself.
Re:OLPC is a success (Score:3, Interesting)
Thanks to OLPC, the PC/Laptop industry's interpretation of Moore's law has totally been reshaped, every 18month now PC/laptops will be half the price instead of 2x more powerful and with 2x more bloatware.
Halving the price of computers every 18 months is simply a fantasy, it's just not a sustainable rate. Costs do go down, but I don't think it can go down anywhere nearly that quickly for long periods of time.
Moore's "law" was really an observation not on power or software, but just silicon complexity, you can't just take the time period and just randomly apply it to some other technology and assume it will work at the same rate.
I wonder if the netbook idea is approaching the problem from the wrong direction anyway. Instead of somewhat bulky notebooks, why not cell phones? That way you can start cheap and expand the device rather than start expensive and work the price down. Heck, Palm used to sell PDAs for $100.
I think what he means (Score:3, Interesting)
Is that deciding that it MUST be free software is the problem. There's a big difference between wanting things cheaply, which free software might be able to do for you, and insisting that it be all GPL'd. For a project like this, OSS zealotry had no place. The emphasis should have been on low cost. Whatever gets you what you need at the lowest price point is what you go with. That may well be OSS in a lot of cases but you use it because it does the job well for a good price, not because you are an ideologue.
So I agree with the GP, I felt the same way. When I heard that it was going to be all GPL I said "Well this is fucked." Reason is that means they weren't being pragmatic about it. They were putting ideologies on source code before the goal of getting computers to kids in poor countries. Also when you start making the GPL a focus, it bring out some of the worst zealots, you get people like RMS who want to tell you what it is going to be. That is NOT what your project needs.
What they needed to do was take any and everything that did what they wanted at the lowest cost. If that was MS or Apple provided, fine, if it was GPL, fine. They needed to say "Here's the requirements," and take what met them the best. If people or companies tried to play politics they needed to say "We aren't interested in that, you provide us with what we need or you don't, we aren't going to do something your way for political reasons." You take the real pragmatic approach of doing what you can to deliver a product that your customers want. This also means focusing on the needs of children, not the needs of geeks. Geeks may think a hackable kernel is a must, children in a 3rd world country probably just want something that is easy for them to get a book on.
That's why various netbooks are succeeding in first world countries. They are NOT concerned with ideology, they are concerned with making a product people want. What people want in 1st world nations is different, I'm not saying you could ship an EEE PC to Africa and have it succeed, but it is the design idea behind it. They give people what they want. They don't care what that is. You want Linux? Fine. You want Windows? Fine. You want a cheaper computer? Fine. You want a more capable computer? Fine. You get to have what you want. They aren't saying "No, this is the One True Way(tm) that netbooks must be done!" They figure out what the customers want and do their best to deliver that.
Had that been the focus of the OLPC project, it might have worked. Instead they played politics heavily, and not just in terms of OSS, and they came out with a POS that nobody was interested in.
Re:Feature creep killed the XO (Score:3, Interesting)
IIRC, they didn't do _that_ much for Microsoft and pretty much just added the SD card
That, and used an AMD chip that was more expensive, slower, and used more power than the ARM chip they were considering.