A lot of comments here seem to suggest that we should discount or ignore this paper because the OLPC program had other benefits (increasing uptake of lower cost laptops worldwide, giving children computer skills, etc.). This is a reasonable argument assuming that most people have only read the free abstract, but this isn't the conclusion that the authors come in the actual paper. Instead, they suggest that the program might have been more successful with increased teacher training and internet access in schools.
I was able to access the NBER version of the paper, but it looks like working copies are also available in a number of other locations:
- https://publications.iadb.org/en/laptops-long-run-evidence-one-laptop-child-program-rural-peru
- https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5391874
- https://www.ofermalamud.com/researchWhile that's undoubtedly true, is that really feasible?
Training programs are expensive, and i imagine difficult to conduct across potentially remote areas with underdeveloped infrastructure.
Internet access is maybe more doable now with starlink, but how practical was it at the time? I imagine this varries significantly with region, maybe in some cases all that was needed was LTE modem -> wifi, but if actually new infrastructure needed to be set up, that could be very pricey very fast.
Like everything its all about trade offs, if olpc did those things would they have budget for other things?
the only way to improve education is to train more and better teachers. that's even completely independent of projects like OLPC. asking if training teachers is feasible is simply the wrong question. arguing that it is expensive is the wrong argument.
education and teacher training is the only way to achieve progress in this world. and if training is expensive or difficult to achieve then that's a challenge we need to overcome, not an excuse not to do it.
There are viable alternatives to Internet access, notably offline resources as provided by projects such as Kiwix. I'm not sure to what extent the projects described in OP actively leveraged these efforts with any real effectiveness. If it didn't, those OLPC mini-laptops would've been functionally equivalent to glorified calculators, and the results would be quite unsurprising.
> because the OLPC program had other benefits (increasing uptake of lower cost laptops worldwide, giving children computer skills, etc.)...
What does that matter if food insecurity, stunted growth, low quality K-6 schools, and other critical issues remain?
From a human capital development perspective, the amount of money spent per year on OLPC could have subsidized a number of similar programs that are both cheaper and have been documented to lead to better developmental indicators.
And it wasn't like OLPC actually placed educators to teach programming at the K-10 level in most of the target regions.
On top of that, broadband and internet penetration didn't expand until the 2010s with Asian commodity telecom equipment being mass produced and exported to developing markets - so what use was a computer which had no internet to a household that was almost always in the lowest income bracket in a developing country?!?
This is why evidence-based policymaking has become the norm and why Banerjee and Duflo won a Nobel Prize.
Edit: can't reply
You (most likely) grew up in a first world country and in the top 5% of households globally.
For the target communities for OLPC, much more basic needs like clean water, school access, nutrition access, and other services were either limited or functionally non-existent.
Much of rural Peru during OLPC (the 2000s) [0] had HDIs comparable to what Laos, Cambodia, and Bangladesh today.
More critically, Peru back then used to be more developed than China [0], yet China's HDI has now outpaced Peru developmentally because local government took an evidence-based approach to developmental policymaking thanks to guidance from Stanford's REAP group [1]
I'm sure you can recognize that the policies needed in a developing country are entirely different from those in a developed country.
[0] - https://globaldatalab.org/shdi/table/shdi/PER+CHN/?levels=1+...
Why should I have gone to college?
My outcomes would be better if I were just richer, smarter and better looking.
There's a real doubt whether college is a reasonable investment these days. The costs are outrageous and the improvement, for society, seems lacking.
If you want to be richer, smarter, and be better looking, food and shelter security might go a long ways
The math of college still holds true in the US depending on what you major in [0][1].
Most non-college goers are not attending apprenticeship programs or joining union jobs - which nowadays increasingly require a college education [2].
This isn't the 1970s anymore where you can go to the local factory and screw parts by hand - manufacturing, carpentry, metalworking, and other industrial arts increasingly require STEM fundamentals which for most students they can only acquire in some form of college (be it 2-year or 4-year).
I've seen this first hand now that I've been taking carpentry courses at my local CC as a side hobby - the union track apprenticeship program that's part of the CC expects an associates degree at a minimum.
[0] - https://www.ppic.org/publication/is-college-worth-it/
[1] - https://static1.squarespace.com/static/60832ecef615231cedd30...
The OLPC project clearly didn’t achieve its aims, but how would they have known that without trying?
More recently, the impact of smart phones on the developing world has been transformational, suggesting some of the ideas behind OLPC may have been good, but the specific implementation lacking. Thanks to smart phones, developing communities now have access to media in global languages, online education, finance, communication, markets (without having to travel for miles), disaster recovery, health resources and much more.
You can even now see rural villages themselves prioritise phone infrastructure over many things that on the surface seem more important - such as by fixing the phone charger before they fix the plumbing!
There’s a book, The Charisma Machine (https://mitpress.mit.edu/9780262537445/the-charisma-machine/), that examines outcomes from multiple OLPC initiatives and covers many of the failings.
A youtube talk about the book for the lazy\busy
Some of the more recent lookbacks with anecdotes and takes:
The Charisma Machine: The life, death, and legacy of One Laptop per Child (2022)
https://news.ycombinator.com/item?id=29971982
OLPC’s $100 laptop was going to change the world (2018)
I think the (overly utopian and unrealistic) ideological basis was not that students would neccesarily do better in traditional learning environments, but that they would become better self-learners and be able to teach themselves things that are relavent to their own needs which may differ from the official curriculum.
I doubt that happened, but i don't think this study would capture that if it did.
OLPC may not have helped many rural children directly, but it did inspire an entire class of computers (netbooks/chromebooks) that made computing more accessible for children and adults across the world. For that reason, I think it was a worthwhile pursuit.
Netbooks were largely a passing fad as were Chromebooks in large part, though the latter are still around to some degree. However, OLPC went beyond cheap laptops and, however well-intentioned, never really took off as a unique entity.
Chromebooks are ubiquitous in U.S. primary and secondary schools
From : https://en.wikipedia.org/wiki/Chromebook#Sales_and_marketing
"In 2020, Chromebooks outsold Apple Macs for the first time by taking market share from laptops running Microsoft Windows. This rise is attributed to the platform's success in the education market.[79][80][81]"
Hmm.
This is interesting. Searches give you different numbers.
But it looks like the number of Chromebooks sold each year is comparable, but probably lower than the number of Macs.
2020 was a special time where every child in America was given a Chromebook so they could do school from home.
They've become the browser-centric portal for the education market which is the only thing a lot of people need/want. (I probably prefer that to defaulting to tablets/smartphones.)
At the same time, as far as I can tell, Chromebooks are pretty much all K-12 focused devices at this point. Which is fine. But I'd potentially buy something higher-end if it were available. But it isn't.
Lenovo doesn't have any current Thinkpad Chromebooks, but they had at least a few models in the past. Those qualified for at least pretty nice, depending on which CPU you optioned. If you picked a Chromebook model that used mainstream CPUs (ex Intel Core), chances were good that they'd have a SKU with a higher tier cpu (i5/i7), even if the case and the screen were nothing special. The atom and arm based ones didn't really have a high tier cpu to consider.
Of course, having found and incubated a useful niche, Google has canceled Chrome OS, so Chromebook offerings are going to be trickier to find.
I buy and re-image old chromebooks to use for terminals for paperwork at a few places I volunteer, they're like $50 and easy to reimage... and nice for doing paperwork.
Small laptop computers are ergonomically useful to children who literally have a smaller body frame. I don't think there were many of those around prior to OLPC and/or netbooks, beyond specialty products like the Toshiba Libretto and perhaps the original Psion Netbook. Nowadays we still have low-end smaller ultrabooks that are effectively quite comparable to the former netbooks.
And I don't think there's much available today. I used to have a small Asus for travel (and a small windows Fujitsu before that). But there are very few <13" laptops/Chromebooks available these days.
I had a couple of netbooks, and they weren't very useful other than as basic devices to check email or SSH. The screen was too small, the aspect ratio and resolution was weird, they keyboard was too small, they were severely under powered - in that they would struggle to play a YouTube video.
I had a couple in the late 00s to avoid having to lug around my 15" MacBook Pro when on-call. For that they worked great, but other than that I avoided it.
Later I got the 11" MacBook Air, and that had many of the same issues (especially weird screen size). It also wasn't that small, by today's standards, as it had a massive bezzle.
Nowadays I have to check if my 13" MacBook Pro is in my bag, as it's small and light enough not to notice. However I'd love a MacBook Air that is same aspect ratio as the 13" just smaller width and depth (No Apple, it doesn't need to be thinner), maybe with a 60% keyboard so typing isn't weird.
Yeah. I have Pro MacBook models but even with better magnetic keyboards for iPads, I think MacBook Airs are pretty competitive for travel. I used to try to use downsized travel laptops when I was traveling a lot but I'm not sure it makes sense any longer even if there were good options available--which there mostly aren't.
The screen and keyboard of your average netbook were too small for fully grown adults, but that exact same diminutive size was and is perfect for kids.
I believe there were many in Japan - including products like small laptop-shaped electronic dictionaries - but that's Japan. Their homegrown cell phones have recently picked up the moniker "Galapagos phones" because they tended to develop country-specific features and rarely leave to other markets. The same can be said for their micro/portable laptops of the early 2000s.
One could argue that OLPC made people outside of Japan aware that the form factor was possible and even optimal for some users.
An unfortunate yet unsurprising report to those familiar with the literature on cognitive ability. I too donated to similar programs. I hope better computer skills make some sort of earning impact, though the prevalence of smart phones probably makes a bigger difference.
I'm surprised about how popular these racist explanations about why the program failed, and not exploring the fact that the hardware, software and training for teachers might have been lacking
If they were good at what they set out to do, the program would've been successful and desired in Western countries (perhaps with upgraded models). But it wasn't.
I'd say the lack of ability to self-reflect on the shortcomings of the HW/SW/Infra and the willingness of the program's creators to embrace such explanations is much more telling about the probable cause of failure.
I'm sure most of these supposedly cognitively inferior Peruvian kids are on their laptops right now, playing League or Overwatch, with most of them having smartphones.
In broader strokes, and with the benefit of hindsight, I think the story of Africa is worth exploring, after a century of Western selfless efforts of trying to civilize and develop the continent, very little progress has been made. Then the Chinese moved in with far less noble intentions and a profit motive, and succeeded beyond imagination at civilization-building.
Are there any links you can share about the unimaginable civilization building in western Africa?
> An unfortunate yet unsurprising report to those familiar with the literature on cognitive ability.
The conclusion of that literature being…?
Sad if true. I was a donor, and the program had good intentions.
At least now instead of just theory we have one study of the results and a data point to use in the next attempt at a similar project. The idea is probably still solid and could be attempted again but with a more refined implementation.
75% or so of venture-backed startups fail within 5 years. OLPC results certainly aren't enough to invalidate the concept or model.
I don't think there's any country left on the globe where you can't get a smartphone or a used laptop, if you really want to.
I don't think you'll find a deeper study than this. And not everything was bad, at least one benefit was "significantly improved students’ computer skills" which may help them get employed in the future, the study focused mostly of academia but there is a whole world outside of academia :)
AI will have similar effect in general population
No significant effect except in the minority who have the drive and capabilities to leverage new technology to achieve their goals
There is always a bias on the effects of new technology because the early adopters are already highly capable people
AI has very clearly disincentivized learning. I don't think OLPC can be accused of that.
I constantly use ChatGPT to learn new things, both in terms of software engineering and more general knowledge.
I've learned a ton of new things from ChatGPT, some which might even be right.
> Following schools over time, we find no significant effects on academic performance but some evidence of negative effects on grade progression.
I still have one of the early green OLPC laptops kicking around that I got at linux.conf.au 2005, in part because they were (or were thinking of) making use of avahi as part of the mesh stuff. They are quite fun to look at.
The project was quite interesting and exciting, and I really miss the era of custom linux desktops, phones, tablets etc being viable projects, it's a shame the project never really directly worked out.
Linux-based phones and tablets are more viable than ever, though. We even have entirely new device classes today, such as e-paper devices that may turn out to be especially useful in an educational context.
I think the interesting thing is they saw significant improvement in computer skills, but no significant improvement in academic performance. It suggests to me that the academic program, or at least the measures from it, didn't factor in computer skills, which seems a mistake given their relevance.
If raw cognitive skills didn't improve, but productivity improved due to improved digital skills, then that's a national-level win.
A similar NPR report from 2012:
One Child, One Laptop ... And Mixed Results In Peru
https://www.npr.org/2012/10/13/162719126/one-child-one-lapto...
I'm not sure why people here call it a failure. The children who got a laptop have reached superior skill in using computers, while it seems like not sacrificing any other capability. That seems like a great result, skilled computer use is a highly valuable skill.
Yes, it was a clear success. It not only did they improve the education of several children and gathered valuable information on how we can try it better next time, they also scared an oligopolistic industry into diversifying their products and supplying several needs that were ignored.
That said, OLPC was extremely ambitious. I don't think they achieved any of the project's objectives. They get a lot of criticism because of that, and it's all ridiculously unfair.
Many people thought that giving kids computers would make the smarter and more generally educated, not just give them better skills with computers.
Many people though that giving kids entire new ways to access knowledge and cooperate on their projects would make them smarter and more generally educated.
And honestly, if you think that's stupid, I'm not really interested on whatever else you think on the subject. It happened to not work for several reasons, some of them mistakes from the OLPC project, but insisting it's an inevitable result is just uninformed blabber.
> but insisting it's an inevitable result is just uninformed blabber.
There’s one failed program indicating that it didn’t work. Are you aware of any successful programs showing that it could?
Well, the BBC had a whole program around getting schools into computing and it worked out fabulously well, so yes. It indirectly ended up giving us a novel processor that is used in just about everything as well.
Are there any studies about the long-term educational effectiveness of those programs? They could definitely be effective wrt. teaching kids simple coding in BASIC or LOGO, but what about other kinds of outcomes?
They bootstrapped the UK IT industry with this I would think that the outcome was successful beyond measure.
The level of technical aptitude the average westerner has just by proxy of being surrounded by electronics is one of those things like running water that gets taken completely for granted. Even a lifelong ditch digger is going to benefit from learning how to send and receive email, and in 2005 it was not a given that he would have.
The tens of millions per year spent on OLPC could have been better applied to programs that have demonstrated tangible positive impact on human capital development in developing countries, such as free meal programs [0], early childhood developmental screening [1], and other evidence-based policies.
Heck, most policymakers in LDCs panned the program at the time as well not actually prioritizing the aid that was needed [2]
[0] - https://econ.berkeley.edu/sites/default/files/Bonds.pdf
[1] - https://pmc.ncbi.nlm.nih.gov/articles/PMC5859813/
[2] - https://web.archive.org/web/20170210165101/http://edition.cn...
Those effective strategies were developed through the same method of research and development as OLPC. At one point, we didn't know about those benefits; should we have not experimented with those strategies?
The nature of research is that some things succeed/fail to different degrees than others, and some that have not sufficiently succeeded will in the future, or will inform other successes. If we already knew the answers, it wouldn't be research.
The issue was there was no robust quantitative research done before OLPC was created.
The programs I gave as examples above all had previously been tested in control groups via RCT before they were rolled out en masse. On top of that, these initiatives were done in coordination with local stakeholders.
This is why JPAL@MIT [0] (Banerjee, Duflo) and REAP@Stanford [1] (Liu, Wang, Rozelle) have had significant success in helping raise HDIs in the states in India and China respectively that they worked with.
On top of that, OLPC (and similar initiatives) took a significant amount of oxygen from the philanthropy ecosystem, with programs and initiatives that had a better strike rate being looked over simply because "it's Negreponte". Even Negreponte's MIT Media Lab largely failed from an outcomes perspective, and was buoyed becuase of donor relations.
These are great points; thanks.
> done in coordination with local stakeholders
That reminds me of my first impression of OLPC when I first read about it - a typical patronizing rich-world kind of aid, 'let them eat cake'.
> Heck, most policymakers in LDCs panned the program at the time as well not actually prioritizing the aid that was needed [2]
I don’t have any insight as to what sort of aid would have been more effective, but quite frankly some of the criticisms were ridiculous when you consider the majority of people in these countries had a cheap mobile phone in their pocket a decade later.
https://www.pewresearch.org/global/2018/10/09/majorities-in-...
A smartphone allows you to both use the Internet and make calls.
OLPC only let you use a computer without internet in a number of areas where broadband and cellphone penetration was nonexistent until the 2010s expansion because of Japanese, Korean, Chinese, and Indian commodity telecom infra.
The main draw of these devices does appear to be telecomm; the Pew article is from 2018, so the numbers have probably changed by now, but back then the majority of users were using dumbphones. I can remember watching videos as early as 2014 showing nomadic tribesmen in Africa using flip phones for mobile payments.
I was under the impression that these devices were Wi-Fi enabled; I take your point that penetration rates for broadband were nowhere near as high back then, but I still think a lot of the criticisms were misplaced. The penetration of telecomm into these countries is going to have massive upside in the next two decades, and computer literacy plays a part in that. I suspect there are compounding network effects involved here that don’t really exist for linear problems like healthcare (though I could just be underestimating the immediate benefit of $1 in medicine vs. $1 in digital literacy).
> I can remember watching videos as early as 2014 showing nomadic tribesmen in Africa using flip phones for mobile payments
Yet internet penetration in Kenya was only 43% [0]. Additionally, countries with significantly higher HDIs (ie. Significantly higher developmental indicators) like Thailand had lower internet penetration in 2014.
Internet penetration was extremely useful in building out infra, but it was just one piece of various other pieces of social infrastructure needed to build human capital.
> The penetration of telecomm into these countries is going to have massive upside in the next two decades, and computer literacy plays a part in that
Most households in developing countries don't have computers [1], so assuming internet penetration implies computer literacy is a fallacy, as most households globally instead use a cellphone as their primary computing device [2].
This is one of the reasons why OLPC failed. Steve Jobs was correct that the smartphone user experience is the best experience for non-technical users.
The organization would have realized this if they tested their hypothesis first, but they didn't. Even Bill Gates called them out for this when they were trying to fundraise in the early 2000s [3]
[0] - https://www.internetlivestats.com/internet-users-by-country/...
[1] - https://datahub.itu.int/data/?e=AGO&i=12046
[2] - https://data.worldbank.org/indicator/IT.CEL.SETS.P2?most_rec...
[3] - https://www.cio.com/article/254451/consumer-technology-bill-...
> so assuming internet penetration implies computer literacy is a fallacy, as most households globally instead use a cellphone as their primary computing device
I was using computer literacy as a stand-in for technical literacy.
> Steve Jobs was correct that the smartphone user experience is the best experience for non-technical users.
We’re speaking about this in retrospect; mass-market smartphones were still in their infancy when the project launched.
> Even Bill Gates called them out for this when they were trying to fundraise in the early 2000s
Gates doesn’t mention anything about smartphones in this article.
Bit like the argument that we shouldn't have gone to the moon
Not a bit like, exactly like. It's a false dichotomy.
It’s not a false dichotomy at all. There is a finite set of resources that can be deployed at any given time. $1 spent on a laptop is $1 that isn’t going to medicine. This usually a curve rather than a straight line (so usually you’re better off with some combination of both), but this doesn’t really apply to a situation where your limiting factor is dollars rather than the factors of production.
You're building your arguments on a false premise, that ALL resources are distributed from some magical mutually exclusive abstract bucket. While it's true when applied to some slices of reality (be it on micro or macro levels) it isn't true in general, where it's parallelized, distributed, discrete and largely independent instead. Doing it your way would inherently lead to socio-cultural diversity collapse. Which is what's actually happening currently, not due to failure of distribution, but due to resources being siphoned and accumulated by hostile entities.
> You're building your arguments on a false premise, that ALL resources are distributed from some magical mutually exclusive abstract bucket.
There is nothing magical about the idea that a resource used for one thing is unavailable to be used for another thing.
> it isn't true in general, where it's parallelized, distributed, discrete and largely independent instead.
Can you provide any concrete example of where the premise does not hold true?
There actually isn't a finite set of resources when we're talking about human capital. Motivation and inspiration and incentive come into play, and as it happens different people are inspired to work on different things.
> There actually isn't a finite set of resources when we're talking about human capital
Ever heard of something called a budget?
Teachers, aid workers, doctors, and others need to get paid. Their suppliers need to get paid. Infrastructure needs to be built.
All of that costs money.
So what, there's only so much money? Or does money represent the amount of work people are willing to do?
My point is simply that the amount of productivity a given population can exercise is not bound to the amount of money in circulation.
> My point is simply that the amount of productivity a given population can exercise is not bound to the amount of money in circulation.
It is bound by other constraints.
> There actually isn't a finite set of resources when we're talking about human capital.
Wrong.
That has an Effective Altruism feel to it though, which is unfortunately tainted due to SBF's involvement and other drama surrounding it.
A lot of people seem to have thought the progression would be "1. Give computers; 2. ???; 3. Better at everything".
Inspired by the computer in the wall in 1999 Sugata Mitra in which yes, just having a computer exist as a local curiosity lead to kids teaching themselves
https://waack.org/2010/04/27/put-more-computers-outside-the-...
https://www.ted.com/talks/sugata_mitra_kids_can_teach_themse...
What did anyone expect? really, many children especially in the Andes suffer from lead intoxication from mining, fruits and vegetables for export are premium quality while the ones for local consumption are bathed in pesticides and other chemicals, kids in the amazon suffer from parasites and other illnesses, tons of regions have malnutrition... etc.
Then there's the fact that Peru is massive (compared to Europe) and most of it lacks internet connectivity, only just recently Starlink started giving newer phones SMS capabilities outside of coverage, but that's as far as it'll go and only for modern expensive phones.
Those kids are not going to open a laptop and have a good time attempting to learn something (if they can even reach the internet), I myself have ADHD and it took me so long to even search and discover khan academy (let's not even begin with how much worse sites like Khan Academy are when you don't have the primary instructor, giving said lessons), add in low energy/inability to concentrate or want to do academic work with memory issues due to intoxication and or malnutrition and... yeah, good luck with that.
There is high amount of talent in Peru, you can tell because there's government programs to give gifted students from public schools free scholarship to private universities and they all end up top of their class, the hardest university to enter into in Peru happens to be a public university too (national university of engineering), it's exam is pretty fucking hard for someone who just graduated high school and even then the vast majority of students who enter come from poor backgrounds.
OLPC? like, the Negroponte thing from way back? That was dead in water within 5 years or so? (As sad as that was) Just surprised to see a study/report on it this many years later.
Well that's sort of the point, does it have a positive effect for that cohort of kids years later? The OLPC hardware was looking rough several years ago and Uruguay's Plan Ceibal and other projects aren't using their hardware today. For reasons unclear to me, Sugar Labs (their OS / UI) just got nonprofit status last year?
They got way ambitious with the interface (GUI). I didn’t understand how to do anything with it.
The user interface was an experiment in itself. It was a lot of python which was way too heavy for the machines delivered, it presented a new platform, and thus a barrier to entry to anyone in the field with general computer skills.
If it had have been delivered using the same Linux base with any number of off-the-shelf minimal window managers it would performed much faster, and the body of people who knew how to make software for it would have existed many of whom would have been happy to make what they needed.
Instead they got a "We've chosen this for you" of something that had not proven its worth.
I still think there's scope for educational systems in all parts of the world if they facilitated development of software that does things that the teachers ask for.
I'd love to see a grant program that gave schools money ear tagged for development on open source projects. The schools get to pay for things that they decide they would like, but it has the requirement that it has to be open source because if one school wanted it there's a chance another school will too. They can choose to pool resources and produce even better things.
I still think a modified Nintendo DS would have done much more for the same price; if they could get them onboard.