I was peripherally involved with this project when it was new, and my biggest takeaways would be that:
The project placed excessive emphasis on nonessential high-effort software initiatives, like developing their own BIOS replacement, desktop interface, UI framework, and nonstandard file manager. Developing and supporting this software consumed a lot of time at the Foundation. Additionally, the project placed a lot of emphasis on unusual, experimental hardware features which they were unable to fully utilize, such as the infamous hand crank (never fully shipped), an unusual hybrid capacitive/resistive touchpad (resistive features were never used), the Pixel Qi display (monochrome mode was poorly utilized by software), mesh networking (software was never fully implemented), hardware buttons for software features that were never implemented (particularly "view source")... the end result was that a lot of time and effort went into designing/redesigning unique features which didn't provide a lot of value to end users.
However, the project placed very little emphasis on the development of boring-but-practical end-user software for the device, like educational games and tools, media viewing tools (like book readers and audio/movie players), or course management software. While they did thankfully ship an office productivity suite, it was literally just OpenOffice; it looked completely different from the rest of the OS, and was poorly integrated. The same was true of the web browser, which was a thin shell around Firefox. Given that these were probably the two most useful apps on the system, the system felt pretty disjointed overall.
Overall, my impression was of a project whose participants treated it as a vehicle for promoting their personal "pet" projects, rather than as a means to a specific educational/humanitarian goal. This severely compromised the effectiveness of the project.
> Overall, my impression was of a project whose participants treated it as a vehicle for promoting their personal "pet" projects
I'm so glad to see others speaking out about this. I'm 100% in agreement that's what "saviour" projects like MIT Asia, OLPC have all ended up being.
> the Pixel Qi display (monochrome mode was poorly utilized by software)
I've said this before about the Pixel Qi panel in the XO-1. I'm not attempting to be negative, but I actually work in the display industry and found the PixelQi display on the OLPC-XO-1 to be unusable. My perception of the display was one with very low quality color with backlight; and terrible viewing angle and unreadable contrast levels in bright sunlight. It also had low resolution. The XO-1 display I had access to had lots of pixel defects. If I am not mistaken (as I do not know actual pricing that Quanta got for the PixelQi displays) they were more expensive than LCDs and required additional parts that made managing their supply chain overall difficult. Most experts in my industry that I talked to expressed the opinion that it was a reality-distortion field type charismatic executive (ML Jepsen) who managed to convince people at OLPC to try that idea. It was funded by UN and money from developing countries budgets who were excited by the promise of being able to educate their masses. Sadly it looks as if those countries and their budget expenditures were taken for a ride by Nicholas Negroponte and his gang, rather than benefiting from a genuinely merit based idea.
Your assessment of the properties of the display specifically seem to go counter what I perceived as a convincing argument by the OLPC guy brought up in a TED talk at the time, namely, that the two biggest cost factors of a new laptop are marketing and the display unit. He then argued that the project can largely just do away with marketing altogether (plausible) and went on to say that even displays that come out of production with a single pixel error have a massively reduced price tag. From that I concluded that the project could reasonably build cheap hardware if only they went with no advertising and run-of-the-mill display units that narrowly failed quality checks.
Nothing in that talk hinted at the possibility that instead of opting for a established and proven technology like LCD they'd choose to go with RDF (reality distortion field) displays instead.
I notice this kind of response anytime I say something that goes against the faith of the OLPC believers.
If you feel something I said is factually inaccurate, then please point it out specifically.
Otherwise, I believe all my points have been proven out. Quite simply, we can all observe the fact that nobody is using a Pixel Qi display or any technology related to it. If your speaker's claims about "Nothing in that talk hinted at the possibility that instead of opting for a established and proven technology like LCD they'd choose to go with RDF (reality distortion field) displays instead." were true, then I'd have been proven wrong and the display industry today would be worshiping at the feet of Jepsen.
One Display Unit Failing QA Per Child? There aren't that many failed displays in the world! Especially in the cost, size and energy budget of a hand crank computer.
I think this is another effect of the reality distortion field.
XO-1's resolution was 692×520 (1/3 of 1200×900 which is the black&white resolution). Apples to apples means RGB XGA vs 692x520 bayer or whatever layout Jepsen claimed was better than industry standard.
There were many XGA 7" and 6" panels in 2006 from Panasonic, Sharp and others.
Why do all of the pages I’m finding with a quick search say it’s 1200x900? Some wires must be crossed somewhere or some massive limitation must exist for your 1/3rd res to be correct, that I’m missing
It's 1200x900 subpixels. In transmissive (color) mode, this was effectively 600x450, since it took a full 2x2 elements to display a color (it used a Bayer-style RGBG matrix, iirc).
I think the OLPC software still treated the display as 1200x900, but this meant that there were visible color artifacts on any sort of fine detail, and color rendition was pretty awful.
> it’s 1200x900? Some wires must be crossed somewhere
1200x900 in XO-1 is tft pixels not RGB pixels like XGA. In the simplest 1d layout, a 1x1 resolution RGB display is actually 1x3 tft pixels. That's why I'm saying apples to apples comparison.
I have an XO-1 in my closet and the display was really the only thing I loved about that laptop. I wish I had the knowledge to create a display driver for it to hook it up to a raspberry pi that could make a fun little netbook.
A display that you could actually read outdoors was super useful for me. I used the XO-1 for a few months as my actual laptop after my macbook died and I bought a macbook pro. Those were literally unusable outdoors and for a while I would carry the XO-1 around with me instead of my laptop just because I could actually use it. Honestly if they keyboard had not been complete crap I might have never replaced my macbook.
Even today my laptops 500 nit display is only usable outdoors at 100% screen brightness and destroys my battery life.
Well Pixel Qi seemed like a magical display at the time. I wonder if given a few more revisions it could have been more competitive? It always felt like a shame that we lost the option of having this display.
I have seen really cool stuff like Mirasol and CED displays come and go and it just sucks that we have seen so many nifty things never really make it. I am holding out hope for CLEARink ePaper and playNitride microLED.
> Well Pixel Qi seemed like a magical display at the time.
Not to me. I had the chance to carefully look at an OLPC XO-1. It was terrible even by the lowest standards of the industry at the time.
> I wonder if given a few more revisions it could have been more competitive?
This is like saying Theranos' drop of blood test could laso have become successful if we just gave the founders a few more billion and a few more decades.
> I have seen really cool stuff like Mirasol
Yes, it is sad that we still can't make interferometric displays work reliably. Some day though. Some day. I still have some hope.
I'd say this is a fair assessment of the project. It came off as overambitious with a number of high-risk project management decisions made at the expensive of a minimum-viable initial product. And this was just looking at it from an engineering point of view, ignoring whether it did actually provide value over other options for educational spending in developing countries.
Funny enough it seems like the hype it generated was its main contribution, which led to the netbook product segment and ultimately Chromebooks. I'd say its spiritual descendent is the humble grades 7-12 school chromebook with much more conventional hardware and software choices. The killer app for school Chromebooks is the combination of Google Docs and whatever Google's classroom management software is, provided cheap to schools in a long-game move by Google to hook a future generation on their office suite. Ironically, despite the very agressive price points of sub-$200 Chromebooks, their current market is limited to the US and the wealthier Commonwealth nations.
> Ironically, despite the very agressive price points of sub-$200 Chromebooks, their current market is limited to the US and the wealthier Commonwealth nations.
Most likely, Chromebooks are too dependent on the "cloud" to be useful in middle-income countries with comparatively flimsy Internet connectivity. A general-purpose x86 laptop will simply go a lot further there. (And if you put a lightweight OS on it, you will almost never be limited by performance for basic workloads, even on old hardware.)
This. I'm in the military, so keenly interested in how much you can do without network. I have two XO-1 laptops and I was a Cr-48 test pilot (their label for their beta testers). I took the Cr-48 with me to Japan for Fukushima in 2011. ChromeOS was not all that useful. Why run ChromeOS when Xubuntu is available? Lubuntu? Pick your favorite lightweight Linux distro. After about a week I cried uncle and installed Xubuntu. Depending on network for the ability to type a document is non-sensical. If I recall, Google Docs didn't have offline mode yet, but due to that experience I have keenly attended to offline mode. It improves things, but it's not a panacea.
I don't think a laptop without network is ever going to be as awesome as one with network, but there's a lot of room on a laptop to do more than run AJAX applications in a browser (do they still talk about AJAX?) that offload all the smarts to a server-side app.
The CR48 was over a decade ago wasn't it? I was also a cr48 beta user and I recall being a teenager, which I have not been for a while. Comparing chromebooks today to the cr48 feels like comparing... idk, anything from 12 years ago to anything from today?
ChromeOS can now run Android apps and a full linux environment.
It's not fundamentally different though. I had the cr48 too and it was just a toy compared to a normal Linux distribution. It didn't displace my similarly underpowered Asus netbook, since the later ran Debian (I think that's the distro I ran back then).
Chromebooks of today are more powerful, which they need to be because web developers of today waste more memory and CPU cycles than ever. But fundamentally the concept and limitations of these "online first" devices are similar. The ability to run Android apps is certainly a step forward, but these things are really designed to lock people into Google's cloud products first and foremost.
It is so, so fundamentally different. It has been over a decade. Chromebooks today run debian. I've got a debian shell right now where I'm working with intellij.
> anything from 12 years ago to anything from today
IDK, I have access to pretty much any hardware I want (cloud GPUs, rack-mounted DGX-type systems, actual supercomputers, yada yada) and in the evening I still sit down to use old ThinkPads. My oldest daily driver is an X201 which actually predates the Cr-48. I don't find my brain working any faster using a 16" macbook. I do find during the workday that the massive amount of screen real estate of 2 XDR monitors is handy to quickly refer across 2 or 3 applications (e.g. scribing for a document under discussion in a video call, while looking at the references under discussion, and checking my calendar with a glance)
What's different is that ChromeOS was an alpha OS 12 years ago and obviously is radically different today. One major difference is that you can run a full debian environment.
> middle-income countries with comparatively flimsy Internet connectivity
Most middle-income countries I've been in had faster internet than the city I grew up in in Canada, and for about $10/mo.
Sure, internet goes down perhaps once or twice a month for half a day or so, but it often does in rural-adjacent parts of Canada too. And in middle-income countries unlimited or near-unlimited LTE phone data for tethering is also very cheap.
I never truly relied on wired internet in low-income countries, so I can't speak to those.
Unless you consider Thailand or Ukraine low income. Although my internet in Belarus and Albania was fast and reliable, and I guess those are low-income countries.
I don't think you'd find many Chromebook deployments in rural Canada, either. Having an IoT thin-client device as your main computing platform turns "internet is down/dog slow" from a mild nuisance into a total showstopper for the day.
> A government-supported private telecommunications cartel
What's the point? Why not simply have a fair market with competition?
> Also, in Nova Scotia, a government-supported private electricity monopoly.
I think Quebec has the same no? That's actually one instance where I was told the government was doing a good job. My colleague told me he pays something like 0.05$ per Kw/h. That's insanely cheap. I assume Nova Scotia does the same.
You would not believe how connected to the internet people from middle income and low income countries are. Everyone has internet.
No, it’s mostly that phones are more convenient/cheap than computers. Everyone just has a cheap smartphone, computer feels redundant and expensive for what it can do. And computers are too expensive for educational segment.
AIUI, the basic architecture of the "Sugar" environment is quite Emacs-like, though not literally built on Emacs. That's why they thought it would be important to have a "view source" feature.
I (also peripherally involved) suggest OPLC had a unicorn-or-dead business plan, but stalled unexpectedly long in not-quite-dead but not-unicorn-yet, and didn't adapt well. Intel and Microsoft/Gates put a lot of effort towards dead. A unicorn story of "succeed (massive country-wide deployments), and they will come (developers, content creators, educators, etc; also network and power infrastructure, next-gen laptops, maintenance and deployment infrastructure, supply chains, etc)" means you're doing a minimum viable unicorn launch. So for illustration, when you're bouncing along, running out of runway, and Guido (python BDFL) publishes clueless criticisms, and the python open-source community isn't "showing up" to help, do you spend scarce funds on hiring a community outreach person to nudge that, or focus on making your next bounce just that little bit higher enough to ignite your unicorn rocket engine, and make training wheels and bootstrap wings a non-issue? Especially when a critical bottleneck isn't technical, but Bill flying around offering politicians Microsoft and Gates Foundation bounties on unicorn eggs.
So I suggest the goal was to trigger mass deployments, supporting subsequent refinement. I'm unclear on what/which technical improvements might have been sufficient to alter the deployment outcome. But people have been critiquing the unicorn-pathfinder-prototype-alpha-preview-0.0.3 unicorn corpse ever since. :)
> However, the project placed very little emphasis on the development of boring-but-practical end-user software for the device, like educational games and tools, book reading tools, or course management software.
OTOH, I recall that media interest in the OLPC project helped pave the way for the earliest "netbook" class devices, which could readily accommodate these features, and in turn found plenty of use as educational tools in middle-income countries. (Though not in the poorest countries nor in the earliest grades of primary education; in either case, this would have called for rugged subnotebooks - much like the OLPC itself or, more to the point, the Toughbook series which was very costly on the market.)
The need for '3-4 year old proof' devices is still acutely felt in the first world, where many of the solutions just involve adding a larger back and sides shell and hoping nothing major happens to the screen.
It really doesn't help that touchpads generally have oversimplified interfaces and poor support for any sort of peripheral that would aid in greater input speed or acuity. (Bluetooth doesn't count, I hate current wireless so much.)
Because there is actually quite a lot of educational material available. It took a bit of curation but my son's time with the screen is generally positive. Of course it's one of many activities, but screens are not inherently evil.
Strongly agree, all my friends give their kids phones just because they think they deserve some time off from them, I think that's wrong.
Never gave my baby a phone, ever and I think it's a good decision, she's not even two yet so how is that affecting her brain?
How is a child supposed to learn how to interact with people if you leave them with a device that might in fact teach them something, like counting and stuff, but that's not the case.
Until the age of 4-5 your focus should be primarily to make your kid socially acceptable so they can participate in as many activities and get a chance to join as many conversations in the future as possible, you want to help them be social. Giving them a tablet will not help them become socialized at all, in fact it will just turn this around. You don't want that for your child. You want your kid to be liked, otherwise it will end up being miserable, think about it.
Spend every minute of your spare time with your baby, that's the way to do it, help them enter the society, by showing them how to interact with people.
I have a young child. I love hanging out with her, but it’s very repetitive. She’s literally entertained by me holding a rubber giraffe in front of her for hours on end. I don’t blame people for getting bored after days / months. But sure, pass judgement all you want. I really don’t care.
Every single parent I've known will, in honest private conversations, admit to the same thing. Loving your child doesn't mean that the repetitive mentally unstimulating parts of raising them magically become non-tedious. It means you do it anyway, because you care for them.
I think this is a consequence for our generally nonsensical arrangements for raising children.
The 2 parents (or maybe 1 single parent) are supposed to do it mostly by themselves in what is called a nuclear family. That is a quite recent development. Earlier it was common to live in larger groups (like extended families) so the work of raising kids was spread out over more people.
The other thing is that is is no longer culturally acceptable to let kids roam by themselves. You always have to have them at arms length when they’re out. If you like to criticize parents for too much screen time, well, what are those kids supposed to do instead?
It takes generations to build up the sort of tightly-knit social capital that makes it generally feasible to cohabit as an "extended family" or to leave kids lightly supervised. Our arrangements for such things follow on from a general way of life that discounts long-persisting social ties as quite unimportant.
I think tedious may be wrong use of word to describe it.
My daughter who is almost 3 is definitely exhausting.
Non stop, hyperactive at night goes to bed at midnight lately.
We play/do activities for like 12 hours a day.
Plus work full-time.
She has a tablet with learning activities, very smart for her age and helped her advance (reads at "3rd grade" level).
We let her play 30-60 minutes a day if she wants to, or we need a quick distraction while I make dinner or something.
Everyone I know, including my grandmother whose tended to 30+ kids agrees she is simply exhausting.
She's very interested in alot of things. Technology one of them.
Many of the most valuable, precious, and important things in life are tedious, boring, difficult, and painful. The unpleasant costs should underline the immense value we place in the things we're willing to suffer for. Very little is withheld when parents suffer for their kids, and it's ok to be honest about it.
The disjointed efforts & lack of care for the most common use modes are jarring. The inability to get hardware promises really delivered sucked.
There were some really good & powerful underlying OS concepts that I think we ought have deep reverence for. Even if the hardware-layer 802.11s wifi mesh didn't work, the systems-layer link-local XMPP mesh, used to create Telepathy Tubes, was ultra-advanced systems engineering that just worked, that was supremely well engineered by discerning sharp choices.
There's so many examples of really smart really sharp visionary engineering. To tick off all the places a complete rebuild of the computing experience failed is cheap and easy. Of course there were a lot of pet project hanger-ons, a lot of non-core engineering. It's just a pity we live in such a scarce, chased society, where such a novel engineering project failed to garner the love & attention it deserved. If there's a reason OLPC failed, it's as much because the rest of the world failed to show up for some bold visionary open source engineering. And in my personal view, it's because open source was already chiefly a husk, a shell, for industrial software development. I don't think the world ever really engaged OLPC at the level it deserved, we all failed to explore a delivered viable greatness & make it our own (a difficulty amplified how long it took for Sugar builds to be available on other platforms). The positive powerful lessons XOPC had for us weren't understood or appreciated then, and we're still today hooked on sad story of failure. Rather than anyone bothering to understand the victory & success & power that was built. Out of best of breed open source systems[1].
Effectiveness could have been higher. But this project chose great tech, & it feels as much like an indicator of failure by the geeks who should have seen further. It's hard to form open source community, but there were so many good ideas here, and just a little more getting-the-ball rolling would have lead to a radically different song we sing about this. There was an incredibly interesting platform for software development in a class-room scale, extremely well envisioned, and we all failed to rise to the high standards of possibility this new, vanguard capability set out for.
If I may riff and provide a potential view of a tangential alternative force that was at play that made the project destined to be doomed (via this approach you mention about un-implemented features across lower cost legacy hardware)
The declining costs of the hardware (screens, capacitive touch, battery, processor, memory, etc etc) via a Wright's Law/ Moore's Law type curve meant that over-optimizing around lowering the costs by choosing 'imperfect' hardware meant it was very difficult to ever get this going as a viable platform - which if we imagine a world of these costs declining 1/4 the rate (4x slower), this ecosystem would have more likely found enough backers and funding to get some more roots and sprouts going.
Ultimately, the cost curve declines of the other equipment meant that smartphones/Netbooks/tablets were bound to leapfrog it well before this OLPC 'formulation' was able to take off and be sustainable on its own.
I have two; a friend of mine had one. We both threw Debian on them and relished in the glorious Pixel Qi display and the nigh-indestructible nature of them. Great for backpacking and road trips in dirty areas, at a then-unbeatable price.
> The new pine e-ink thing gives me hope, with a keyboard, but bah
Give it some time, they've just about gotten started on the low-level graphics support for that particular device. The next step is coming up with meaningful UX, probably no easy task given how different e-paper is from other sorts of visual output.
It's not e-ink, it's a transflexive LCD where you can remove the colour and turn off the backlight to get higher res, as the subpixels are setup like a camera (square of RG/GB) instead of the normal 3 vertical subpixels (RGB)
At that time (~2003) I didn't realize the benefit - India govt asked MIT Media lab to leave India due to their high salaries and I guess lack of accountability..
> The project placed excessive emphasis on nonessential high-effort software initiatives, like developing their own BIOS replacement, desktop interface, UI framework, and nonstandard file manager. Developing and supporting this software consumed a lot of time at the Foundation
How many of these were funded PhD thesis vs software from salaried devs?
None were funded PhD theses -- a mix of salaried devs, donated engineering time from partners like Red Hat, work done by volunteers in purchasing countries like Uruguay, and general open source contributions.
I assumed the research component was bigger since the project originated at MIT, and that it was one of the reasons they were pushing for a system that didn’t exist at the time instead of porting a known Linux distribution to it.
> Overall, my impression was of a project whose participants treated it as a vehicle for promoting their personal "pet" projects, rather than as a means to a specific educational/humanitarian goal. This severely compromised the effectiveness of the project.
Be that as it may:
1. The research projects were kind of cool.
2. I suspect that OLPC paved the way for the netbook/chromebook/cheap laptop era.
3. And cheap tablets are even cheaper and work great for accessing khan academy and wikipedia as long as you can get free wi-fi somehow.
4. I still think it's a great idea to make computing (including programming) accessible to everyone, be it via web browsers, raspberry Pi, micro bit, cheap laptops/tablets, or whatever.
You know, they were funded because of these novelty. If you spend more time with international charity activity, you'll see that no one truly want to help the poor, they are more interested in showing their kindness.
I don't think it's quite that. I think it's that the prospect of making reducing poverty your life's work and the end result is that you reduce the world poverty rate by 0.02% is just too oppressive to face head-on. So you either get people who devote themselves to the process rather than the outcome, or you get outcome-oriented people who look for shortcuts and miracle fixes. But that doesn't mean the motive is insincere.
The funny thing about "pet projects" is that successful projects were also "pet projects" – linux, python etc. Nothing wrong with feeding your "pet project" per se.
But the point is that it doesn't work the other way around: not all pet projects become successful projects. In fact, very, very few pet projects ever ship.
Honestly, your criticism sounds like it could be extended to non-profit/volunteer work in general; not just OLPC.
Without the "meet customer demands or starve" dynamic that places pressure on for-profit organisations, a whole lot of critical but unsexy work just doesn't get done.
That’s why I like effective altruism: there’s a lot to be criticised on a purely utilitarian view of help, metrics are often wrong and all that, but any form of self-criticism is such a step-up from the usual white-saviour complex. "We stop this unless we reach that goal" feels horrifying to people on the ground who can’t accept why beneficiaries would be punished, until they realise that, no: you are taking away an unhelpful service. Potentially better options now have a budget. It’s messy and goes in circles but at least it doesn’t stay someone ineffective.
There is nothing wrong with the product vision. Remove the part about the handcrank and monochrome screen, and you are basically describing a modern day Macbook. Unfortunately they did not have the budget of Apple to deliver such a device.
The project placed excessive emphasis on nonessential high-effort software initiatives, like developing their own BIOS replacement, desktop interface, UI framework, and nonstandard file manager. Developing and supporting this software consumed a lot of time at the Foundation. Additionally, the project placed a lot of emphasis on unusual, experimental hardware features which they were unable to fully utilize, such as the infamous hand crank (never fully shipped), an unusual hybrid capacitive/resistive touchpad (resistive features were never used), the Pixel Qi display (monochrome mode was poorly utilized by software), mesh networking (software was never fully implemented), hardware buttons for software features that were never implemented (particularly "view source")... the end result was that a lot of time and effort went into designing/redesigning unique features which didn't provide a lot of value to end users.
However, the project placed very little emphasis on the development of boring-but-practical end-user software for the device, like educational games and tools, media viewing tools (like book readers and audio/movie players), or course management software. While they did thankfully ship an office productivity suite, it was literally just OpenOffice; it looked completely different from the rest of the OS, and was poorly integrated. The same was true of the web browser, which was a thin shell around Firefox. Given that these were probably the two most useful apps on the system, the system felt pretty disjointed overall.
Overall, my impression was of a project whose participants treated it as a vehicle for promoting their personal "pet" projects, rather than as a means to a specific educational/humanitarian goal. This severely compromised the effectiveness of the project.