What people get wrong about learning technologies

There is a fascinating pattern of repeated forecasting and product errors that people have been making about the impact of technology on education for over 40 years now. I know, because I made it myself on two separate occasions with two separate products.

Here is a tempting but wrong hypothesis: the best way to use technology to improve learning is by building products that try to replicate a great human tutor, i.e. automated tutors. The people who make this error start with two good premises: 
  1. Private tutors are orders of magnitude better in teaching someone than any other method (see e.g. Bloom's 2 Sigma Problem).
  2. Computers in general and AI in particular are getting better, so we should be able to increasingly emulate what a human tutor does.

This idea of building intelligent tutors has been so mesmerizing to generations of researchers and product people that there is even a book detailing the history of the various (largely failed) attempts to do so [1].  However, the same people completely miss two other things that are also true: 

  1. A lot of what a good private tutor does has nothing to do with selecting the right exercise or hint, and everything to do with providing excitement, motivation, self-confidence, and discipline (which computers can do only so much about).
  2. Given the status quo, there is a lot of far lower hanging fruit in using technology for learning, notably (a) giving people unlimited free/cheap access to high quality content, and (b) using messaging, video, and social media to put people in touch with others, be they human tutors or peers passionate about the same things.
The result is that high-tech platforms with adaptive learning "AI" engines like Knewton fail, whereas products that succeed look more like Lambda School [2] – which, if necessary, could probably be run over Zoom, Slack, and Google Docs, without any custom software at all. Another great example is a company like VIPKid, which correctly realized there was no need to build an AI-tutor, they could build a huge business by just making it easier to hire a human one.

The failure of Knewton and the success of Lambda School or VIPKid are therefore a reflection of two factors – first, building AI tutors is harder than people think, and second, there is a lot of far lower hanging fruit to be picked before one moves to the more challenging technical problems. Of course, this will eventually flip when the low hanging fruit is gone and AI has progressed sufficiently to make AI tutors more effective. However, even with the accelerating progress in AI, I suspect that for the foreseeable future, any "AI tutors" are likely to be glorified exercise banks selected and used in teaching by – you guessed it – human tutors on platforms like Lambda or VIPKid.

[1] One notable class of exceptions are domains where Spaced Repetition works reasonably well, which generally includes standardized testing and some aspects of foreign language learning, leading to success of products like Duolingo and Quizlet.

[2] See also my earlier post, Why Lambda School Works.

Mindstorms: Children, Computers, And Powerful Ideas

Okay, I'll admit it. Despite working on education technologies for several years, I never got around to reading Seymour Papert's Mindstorms, which was generally perceived as a bible by at least some part of the edtech universe. Over the holidays, I finally fixed my omission, and I'm glad I did.

The good

Let's start with the good. The book includes some great ideas. The foreword, written by Mitchel Resnick, the LEGO Papert Professor of Learning Research at the MIT Media Lab, does a great job of summarizing the key principles:

Projects. People learn to solve problems (and learn new concepts and strategies) most effectively while they are actively engaged in meaningful projects. Too often, schools start by teaching concepts to students, and only then give students a chance to work on projects. Seymour argued it is best for children to learn new ideas through working on projects, not before working on the projects.

Passion. Seymour knew that people will work longer and harder, and make deeper connections to ideas, when they're working on projects that they're passionate about. Seymour once said: "Education has very little to do with explanation, it has to do with engagement, with falling in love with the material."

Peers. Seymour gives the example of Samba schools in Brazil, where people learn from and with their peers, rather than from teachers. This, in some sense, did hint at what was to come with today's socially networked learning communities.

Play. Seymour referred to play as "hard fun". He recognized that children don't want things to be easy. They're willing to work very hard on things that they find meaningful.

The not answered

After finishing the book, I was left wondering about two main questions:

1) Why are computers the best way to put the foregoing into practice? 

Unfortunately, Seymour never really quite explains why he thought computers were the best way to put these principles into practice. There are strong arguments to be made for this – for example, that it's too hard to train enough teachers to teach this way, or that you can only teach this way in small groups, and we are not willing to pay for enough teachers to do that, so we need computers to backfill that gap. However, the book doesn't say much on this topic. I found myself wanting to know to what extent Seymour thought that computers were less effective but much more affordable private tutors vs just being better period, at least for certain teaching tasks.

2) What did Seymour think about the impact of the rest of the computer revolution on education?

Seymour's focus is clearly on immersive learning environments – his key example is teaching math and physics by having students control and program an imaginary turtle moving through space. But what did he think about everything else that computers would enable? And more specifically, what would he think about where we are today? Excited? Disappointed?

  • On the one hand, despite incredible advancements in technology, very few students learn math and physics via programming today. In that sense, computers had relatively little impact. 
  • On the other hand, one could argue that that what really enabled "projects, passion, peers, and play" was not learning how to program computers, but simply using computers and the internet in particular for two things: (1) accessing an infinite library of great content, for free (by teachers who inspired passion, and provided enough guidance to enable projects), and (2) talking to other people about it (including peers), which made it more like play.

Unfortunately, I couldn't find any talks or articles by Papert where he expressed how he felt about where we ended up by 2005, 2010, or 2015, before his passing in 2016. If you happen to know of some, please let me know.

The coming unbundling of higher education

There is a lot going on in US higher education right now. Harvard and a slate of other universities just announced that this fall, all undergraduate classes will be online only. Parents are now widely starting to negotiate college tuition discounts the same way they would haggle over the price of a car. More broadly, I suspect we are about to finally see the great unbundling of the US higher education market. 

More innovation in US higher ed would be a good thing. Under the existing model, going to college is a bundled package that includes quite a distinct set of component goods:

  1. Signaling for employers and the world more broadly, both on being smart enough to get in and smart/diligent enough to graduate
  2. Credentials for employers in fields where that matters (especially if they are regulated like law or medicine)
  3. Acquisition of professionally valuable knowledge and skills (both hard skills like programming and soft skills like getting along with other people)
  4. Opportunity to build a strong professional network
  5. A vaguely defined chance to become a more rounded person and citizen
  6. A socially-approved party circuit

For a long time, breaking up the bundle and the monopoly of colleges on providing these goods has been challenging. There is a lot of inertia in how society operates, and for higher education to change, a broad variety of stakeholders generally need to become open to some alternative and do so roughly at the same time: the parents, the students, employers, colleges, and the government (which both provides funding and in some cases enforces credentials).

The system was on an unsustainable path well before Covid-19 – tuition inflation was off the charts while many students received education with at best dubious impact on their lifetime earnings potential. However, Covid-19 just probably accelerated the change by a decade. Suddenly, a lot of people are really asking - is it truly worth it to pay $50,000 for a year of sitting at home taking Zoom classes? Or maybe $150,000 if you got accepted into an MBA program? Or would you rather unbundle some of these components and get them on better terms?

Maybe you pay $30,000 for a one-year Lambda School degree, go get a job, work for 2 years, pay off any debt, and then take a one year sabbatical of travel, reading, partying, and/or finding yourself, which you otherwise may have done over the course of a 3-4 year program? Except that at the end of the 4 years, you are left with a high-paying job and some savings rather than $200,000 of student debt and unclear job prospects?

Or maybe you go take a gap year, work on a side project, take an apprenticeship of some kind, or pursue some other path first, and wait to see how the system shakes out. I don't know yet how many viable alternative paths we will discover over the coming years, but it feels like the beginning of a Cambrian explosion in this space.

The End of Education

My final read on education from Neil Postman was his 1996 book, The End of Education. This was my third book on this topic written by him, following Teaching as a Subversive Activity (published in 1969) and The Disappearance of Childhood (published in 1985).

Unfortunately, I don't have much to say about The End of Education, other than "it's more of the same". In some sense, this book is the most disappointing of all three, since there simply isn't even much to either agree or disagree with. Other than meandering endlessly on various tangents, the book just doesn't say much.

Here's a summary of the first 100 pages: schools fail because kids are not motivated to learn. Kids are not motivated to learn because America is missing a great narrative about our purpose in the universe. Essentially "our schools are lost because our nation is lost".

The second and final 100 pages then discuss five possible narratives that would allegedly somehow fix our schools, which Postman titles The Spaceship Earth, The Fallen Angel, The American Experiment, The Law of Diversity, and The World Weavers/The World Makers.

Unfortunately, while some of these titles sound somewhat promising, the detailed discussion doesn't quite convince that any one of these great overarching narratives would make a tangible difference to the behavior of a Grade 5 teacher, or the experience of a student sitting in their classroom.

To his credit, Postman does acknowledge that he's more sure of his diagnosis of the problem that his prescription for possible solutions. However, I'm not sure even the diagnosis is that convincing - yes, perhaps in the early two centuries of the republic, we had a society that was hungrier to improve itself and its life, more unified by a narrower set of cultures and religions etc, but I'm still not quite sure that such differences made our schools worked better (or indeed, whether our schools worked better in the first place).

And so sadly, Postman's writing on education ended up being a disappointment. However, I generally agree with the alleged quip by Lord Acton that we should judge character at its worst but talent at its best. And so it is Postman's most piercing and enduring work, Amusing Ourselves To Death, that will for me always define his intellectual legacy.

The Disappearance of Childhood

After Teaching as a Subversive ActivityI just finished The Disappearance of Childhood, another book by Neil Postman that deals with education, at least broadly speaking. The book was published in 1985, and makes a fairly simple argument:

1. At its essence, childhood is defined by the existence of secrets that adults hide from children, and which children get to gradually discover. By secrets, Postman refers to many aspects of adult life, including money, social relations, sickness, death, violence, etc.

2. Until the invention of the radio, and especially television, it was possible to hide secrets from children. Books cost money, their distribution can be controlled, they can be put on top shelf at home, and they require effort to read. Besides, there are certain things that can only really be experienced by watching them - rather than by reading about them. [1]

3. Since modern media makes secrets obsolete, childhood has largely disappeared. Or, as Postman put it himself "It means - to use a metaphor of my own - that in having access to the previously hidden fruit of adult information, they are expelled from the garden of childhood."

Overall, I think this is an interesting argument, and many people will probably find themselves agreeing with Postman on this. But to me, the book feels a bit too much like it was a nail to someone with a hammer. I don't think it's an accident that Postman's most lasting work, Amusing Ourselves to Death, was also published in 1985. Postman got a lot right in that book about the impact of media on modern society —but extending this to defining childhood feels like a stretch. My own intuition is that at least in my own case, my childhood was defined more by freedom of having both lots of time and little expectations placed on me, rather than by being naive about the world. 

Aside from the core argument, the book unfortunately feels quite similar to Teaching as a Subversive Activity. It's full of diversions and side points which do little to advance the core argument, and seem to be there mostly to show off the author's encyclopedic knowledge. [2]

[1] The Disappearance of Childhood was published in 1985, before the internet, and before the iPhone. Given how exponentially harder these inventions have made it to keep secrets from children, even compared to television, Postman would presumably say that by now, childhood essentially doesn't exist.

[2] Having said all that, I did enjoy a few of the many side points and quips, for example:

"The importance of fairy tales lies in their capacity to reveal the existence of evil in a form that permits children to integrate it without trauma."

"[The disappearance of childhood] can be seen to occur not only in clothing but in eating habits as well. Junk food, once suited only to the undiscriminating palates and iron stomachs of the young, is now common fare for adults. This can be inferred from the commercials for McDonald's and Burger King, which make no age distinctions in their appeals. It can also be directly observed by simply attending to the distribution of children and adults who patronize such places. It would appear that adults consume at least as much junk food as do children. This is no trivial point: it seems that many have forgotten when adults were supposed to have higher standards than children in their conception of what is and is not edible."

Oxbridge Admissions in 2019

Last year, roughly 25,000 different people used OxbridgeAdmissions.com to learn about applying to and studying at Oxford and Cambridge universities.

If you applied to one of these universities in the last few years, please consider submitting a profile describing your experience so that those applying in future years can benefit from your experience. You can get started here.

How to wreck your education system in 16 steps

Neil Postman's Amusing Ourselves to Death is a classic that stood the test of time. Published back in 1985, the book successfully predicted much about how new communications technologies would reshape society. So the other day, I grew childishly excited when I discovered that Postman was also an education theorist who wrote several books on education and our school system.

Unfortunately, Teaching as a Subversive Activity, Postman's first foray into education, published in 1969, ended up being quite the disappointment. The book starts out OK, with some reasonable critique of the current system. However, the critique drags on and on, with no or limited suggestions for how to do better. Worse still, the tone that seeps from the pages is one of an ivory-tower intellectual, sneering at the teachers who do the hard work without any reasonable ideas for how to make things better. This is particularly surprising, given that Postman actually was a teacher earlier in his life.

The insanity climaxes about two thirds through the book, when the authors finally reveal their 16 suggestions for how to improve the system (as listed below). Disappointingly, their suggestions are mostly the kind of radical soundbite that sound controversial and lofty in theory, and somewhat entertaining at a dinner party debate, but don't stand up to even a brief consideration of whether they might actually work in practice. Some of their suggestions are OK, many are just bad, and a few are just outright insane.

Postman kept writing about education for the rest of his career, going all the way to 1995. Here's hoping that the next few books are better. Teaching as a Subversive Activity gets an F.

The authors' 16 suggestions for how to improve education:

1. Declare a five-year moratorium on the use of textbooks.

2. Have "English" teachers "teach" Math, Math teachers English, Social Studies teachers Science, Science teachers Art, and so on.

3. Transfer all the elementary-school teachers to high school and vice versa.

4. Require every teacher who thinks he knows his "subject" well to write a book on it.

5. Dissolve all "subjects", "courses", and especially "course requirements."

6. Limit each teacher to three declarative sentences per class, and 15 interrogatives.

7. Prohibit teachers from asking any questions they already know the answer to.

8. Declare a moratorium on all tests and grades.

9. Require all teachers to undergo some form of psycho-therapy as part of their in-service training.

10. Classify teachers according to their ability and make the lists public.

11. Require all teachers to take a test prepared by students on what the students know.

12. Make every class an elective and withhold a teacher's monthly check if his students do not show any interest in going to next month's classes.

13. Require every teacher to take a one-year leave of absence every fourth year to work in some "field" other than education.

14. Require each teacher to provide some sort of evidence that he or she has had a loving relationship with at least one other human being.

15. Require that all the graffiti accumulated in the school toilets be reproduced on large paper and be hung in the school halls.

16. There should be a general prohibition against the use of the following words and phrases: teach, syllabus, covering ground, I.Q., makeup, test, disadvantages, gifted, accelerated, enhancement, course, grade, score, human nature, dumb, college material, and administrative necessity.

What happens to homeschooling after COVID-19?

In the last two weeks, I have had numerous conversations about how forcing millions of people to homeschool their kids may permanently shift the popularity of homeschooling. Although I think that will be true at the edges, I remain skeptical about the impact on the bulk of the population. 

Why? For better or worse, my sense is that if we were to be honest about what schools actually provide to society, learning of the curriculum would be the plurality, but quite possibly not even the majority of the value. Schools these days seem to provide at least the following six values to society [1]:

  1. Teaching of the curriculum: maths, science, languages, history, etc.
  2. Teaching of intra-personal skills: patience, ability to pursue goals, discipline, ability to manage boredom, etc.
  3. Teaching of inter-personal (social) skills: interacting with many different adults, and more importantly, one's peers
  4. Friendship formation (and later on, dating)
  5. Watching the kids (= childcare)
  6. Feeding the kids

In theory, homeschooling options could provide all of the above. In practice, I think this is hard for all of the above, but especially for 3 through 6. The reason is that all of those are intrinsically group activities, which get exponentially better and/or cheaper with scale (up to a point).

In theory, homeschooling doesn't have to mean just a parent teaching their own kids. Rather, we could of course see a whole ecosystem of home schools emerging, where one or more parents run a school for a dozen to a few dozen kids. This would make it easier to deliver on 3 through 6 above, but then we run into a whole host of new issues:

  • Reaching consensus: anyone who has ever tried to get a group of parents to agree on anything will appreciate the daunting nature of the task of getting everyone on the same page about anything from curriculum to what will be served for lunch. This gets substantially easier in narrow communities with an existing set of beliefs (e.g. specific religions), which is why this works and will continue to work better there.
  • Inverted U-Curve in Teaching Quality (1/2 above): unless you as a parent happen to have a particular knack for teaching, the most likely reason for why you would be better at 1 and 2 above is not that you're a better teacher, but that you care more. For all the griping about school teachers, most people who go into the profession do so because they care (at least a bit), and for all the griping about their insufficient/out-dated education, they have been taught some useful things, and have learnt some through experience. So an amateur teaching their own kids will do OK, or a professional teaching a group of other people's kids will do OK. What are the odds that an amateur teaching a group of other people's kids is the worst of both worlds?
  • Inverted U-Curve in Network Quality (3/4 above): when it comes to inter-personal skills and friendships, the economies of scale do not stop once you hit 20 kids. If you care about friends, or sports, or dating, the quality of your experience generally gets better as you go from 20 to 100 students. Now, there is a point at which diseconomies and anonymity kick in, but I'd guess that's somewhere in the range of 200 to 400 students, way above the group size that's realistic in homeschooling.
  • Inverted U-Curve in Operational Quality/Cost (5/6 above): my sense is that the cost and difficulty of providing the infrastructure of running a school follows an inverted U-Curve. If you're teaching your own kids, you need a room in the house, and the ability to cook a lunch for a few people. If you're running a big school, you have a building and dedicated kitchen staff. But where do you base a school for 25 people, and how do you get lunch for everyone?
  • Funding models: finally, unless we get school vouchers or something similar, the only parents able to spend out-of-pocket for such a school will be the wealthiest one, who on average have less of a burning need for this, since their kids probably go to a better school already.

Given the above, I'd expect the main mode of school delivery to return to normal once we have dealt with the virus itself. Having said that, what I suspect will change is the acceptability of using remote/homeschool-like tools at the margins. I don't know exactly what these uses will be, but I'd guess stuff like dialing into class over Zoom when sick, moving more of the communication online, perhaps increased focus on learning from peers, etc.

Thanks to my mom, who was a teacher for more than 30 years, for reading drafts of this.

[1] In addition to the six values listed above, schools also provide credentialing. However, at least in the US and in the UK, this has already been unbundled from the schools themselves and is now delivered mostly via independent bodies such as the SAT (College Board) or the various exam bodies in the UK (AQA, OCR, Edexcel, etc).

Why Lambda School works

After the Massive Open Online Courses (MOOCs) mania fizzled out some time in 2015/2016, there was a period of 2-3 years when the education industry (yet again) mostly soured on education technology startups. But after a brief respite, there is a new darling in town now: Lambda School

While journalists continue being skeptical as usual, the data on salary increases before and after people go through the school is impressive. We're told there are hundreds of people whose yearly salary jumped $50,000 or more after completing the course. Some student stories are inevitably a bit more reserved, but the overall picture is impressive.

Back in 2014, as someone who was involved in and spent a lot of time thinking about the education business, I wrote a post called Why it's so hard to make ourselves better. In the post, I discussed the reasons for why we didn't see more big companies being built in online education. I wrote:

The problem with the “customer value proposition” of most education and healthcare products is that they offer:

    • an uncertain, often unmeasurable benefit in the future
    • in exchange for effort (learning) or giving up “easy pleasure” (not eating that desert) today

and then clarified this further by pointing out that:

The challenge with education and healthcare is that often, they can offer neither:

    • Measurability: although we’re making progress in quantifying both domains, isolating the impact of a new product remains difficult.
    • Immediacy: as it turns out, (for now), our bodies and brains just need some time to change.

So why does Lambda School work? Among other things, they figured out a way to overcome these exact challenges:

  1. Lambda made the reward huge — learning most skills that MOOCs offered, for example, wouldn't really move the needle on your expected income in a few years. Lambda picked a domain where a highly valuable skill is both in massive shortage and can be taught in 12 months. So the reward became $50,000 per year of extra income. For virtually every student, that's life changing.
  2. Lambda made the reward measurable — Lambda made incremental student income after completing the school one if not the primary metric they focus on and market. This is very different to most MOOCs or other online education companies, who market a much more vague value proposition. Making the reward not only measurable, but doing so in a "currency" that people care about as much as money massively increases motivation.
  3. Lambda brought the reward forward — most education startups teach you skills that may be valuable at some point in the future, often years out. Lambda found a domain where you could teach someone something extremely valuable in 12 months.
  4. Lambda reduced the financial cost  instead of paying upfront, Lambda uses Income Sharing Agreements where students don't have to pay anything unless they get a better job that pays more money. This removes two problems for students. First, they don't have to worry about where to borrow the money for tuition if they don't have savings. Second, but perhaps even more importantly, the students don't have to worry about wasting their money on a useless course.
  5. Lambda reduced the effort cost — by building a strong curriculum and tools such as peer-to-peer support, mentors, recruiting help for finding the next job, etc., Lambda School reduced the effort of going through the program.

In summary, Lambda offers a big, measurable reward that you can get rather soon, and they significantly reduce both the financial cost and the effort cost of achieving that reward. No wonder it seems to be working.

New updates on OxbridgeAdmissions.com

We're continuing the rollout for the new Oxbridge Admissions site with the launch of the Oxbridge Admissions Blog. These highly curated posts are meant to help perspective Oxford and Cambridge students with everything from potential interview questions, pre-interview test prep, and first-person reviews of stuff like accommodation and food at Oxbridge’s most popular colleges.

To see all posts, visit the Oxbridge Blog archive or start here for General tips for Oxbridge interviews. If you’re looking for something in particular, here’s a breakdown of topics.

Cambridge Courses

Cambridge Colleges

Oxford Courses

Oxford Colleges