Friday, August 9, 2013

Coursera : Universities are Dead

The last time I was this enthusiastic about a new thing it was Amazon, in the days when no-one had heard of Amazon but it had just allowed me to buy in seconds a rare book that I'd been looking for for years. (for comparison, I was mildly enthusiastic about google, and I loathed, and still loathe, facebook)

The new thing is coursera:

I loved being an undergraduate. Partly because I loved being young and brave and free and surrounded by clever girls and subsidised beer, but at least a little because every day some wise elder would reveal one of the secrets of the universe to me.

I haven't felt that rush of learning lots of new important stuff for such a long time. I hadn't realized quite how much I missed it.

The difference now is that I'm capable of dealing with it all. I know how to learn things much better than I used to, back in the days when I thought that talent was inborn and practising was cheating.

Coursera is offering, entirely for free, elite-university courses on line.

Not just the lectures, which would be no more use than textbooks are, but the exams, tests, course structure and collaborative environment which make it possible to learn.

You need to be very motivated indeed to study something alone from a textbook, no matter how good. Twice in my life I've managed it, once with the incredible Structure and Interpretation of Computer Programs, and once with David Mackay's Information Theory, Inference and Learning Algorithms.

Over the last few months I've done two world-class university courses. Daphne Koller's famous Probabilistic Graphical Models course from Stanford, and Tim Roughgarden's Algorithms I course, also from Stanford. PGM is a graduate-level maths course, and Algorithms I is a core component of Stanford's computer science degree.

Both of them have been mind-blowingly interesting and fun. Twenty years I have made a good living as a programmer, and Algorithms I has convinced me that I've never known that much about programming.

What's really freaking me out is that they're every bit as good as the best courses I remember from Cambridge, and that I have found the continuous assessment / easy-ish problems to do every week structure of them fantastically motivating. Even addictive. There are things I should have been doing, and want to do, and would enjoy doing, that I have not done because I would rather spend the time watching Daphne and Tim's wonderful explanations of concepts that, with hindsight, I realize I should have learnt about years ago. Both courses have unexpectedly turned out to have been directly relevant to my current real-world project.

So I recommend coursera very highly indeed, but I also wonder if they realise what they have done.

There is a thing called a winner-takes-all market. You don't want to be in one.

What happens is that you're, say, a man who makes his living playing the piano in public houses. There were once many such men.

You make a modest living, because there are many people who can play the piano, and many pubs, and the market is quite competitive, but you are compensated on roughly the right scale for the effort you put in to learning to play.

Or say you are a football player, who plays for his local football club. It is only a game, so you have to have a proper job as well, but tribalism being what it is, once a week, you play at your local ground for the honour of your town, and a couple of thousand people turn up to watch. Together they pay enough to keep the stadium running and pay the expenses of the club and there's enough left over to distribute round the players in salaries, and maybe even some of the really good players manage to make it a full time job.

This world is gone. Dead and no-one can remember it or believe it or understand what it might have been like.

It is as gone as the world where literate scribes copied books by hand, and only a very wealthy man could afford the abbeyful of people that it took to copy out whole books with quills.  Can you imagine a world where owning a single book would be evidence of enormous power, wealth and status? Can you imagine a world where the ability to write legibly would guarantee a man a good income for life?

What happens is that someone finds a way of making the labour of one person, which used to satisfy the needs of a few, satisfy the needs of many. A printing press, a record player, a live TV broadcast. Often these ways are referred to as media, because they stand between.

This is a great thing for the many. Suddenly the world is full of books, full of music, full of televised sport. Things that were expensive or impossible luxuries (like being a Manchester United supporter who lives in Cork, or owning Several Books, or listening to the Orchestra of the Royal Opera House in your croft in the Shetlands) become widely available and cheap. Staggeringly good news for almost everyone.

But what happens to the army of people who used to do all the writing, football playing, music playing, acting, etc, etc?

A clue. Think of a profession where there are stupidly, staggeringly rich individuals, and yet most people who do it pay to do it.

Let us say: George Clooney versus your nearest friend who does amateur dramatics, or (I do not know the names of any current good football players but I predict their existence and characteristics with confidence) versus the best football player you know personally, or Bryn Terfel versus the world's tenth best interpreter of Wotan, or JK Rowling versus anyone you know who's paid to publish their own book. On and on and on this list goes, covering the huge number of professions and people that used to provide widely used services on a personal basis for modest reward but where someone found a way for everyone to enjoy the output of a few people who were very good at it.

As I say, this is very very good news indeed for almost everyone.

Clever inhabitant of a little village in Tanzania whose parents can't afford to educate him past primary school? Welcome to Stanford.

Middle aged computer programmer who'd quite like to do a second degree for fun but has never quite wanted to commit the three years and £50,000 that it would probably cost in total to a project which may not be quite as much fun as he imagined? Screw that. You can take your pick of the interesting sounding courses from all the best universities in the world, and do them in the order you want, and try stuff and bin it if it's boring.

Hell, I even thought I'd try a literature course, just to see if there was anything to it. The lectures were fun and the lecturer inspiring, but I still think it's the singer not the song. They know lots of stuff, these literature types, it's just that none of it is true. Which is why they're so into post-modernism I guess. You would be if you built on quicksand and kept finding your house had sunk.

But what about the universities? I can't imagine how anyone could beat Tim Roughgarden's charmingly macho Algorithms I course, and I'm certain that he's one of the best teachers of that subject anywhere in the world. But I bet there's someone somewhere who can do it even better.

And if that guy or girl puts in the months of work to make a better version of the course, everyone will want to watch that instead. There is no room for two people doing the same thing in a winner takes all market.

And what about the sea of relatively uninspiring no-hopers in the world who have dedicated their lives to teaching important things to the young?

These people are fucked. And they do not deserve the fucking, because they are good people who have spent years doing good things for good reasons.

But who will want to pay £30,000 to do something they can do much better for free on their computer?

A shame. I will miss the universities. I hope the places who do research and consider teaching a distraction from that will weather the storm and maybe even benefit.

I wonder if any of them have seen the tsunami that is coming for them yet. I talked to a very eminent academic the other day and he thinks that coursera's an interesting experiment that might do some good. It isn't. It's an onrushing wall of death. I give the universities ten years, tops.

Schools have a child-control function. They're somewhere for society to imprison the young people it has no use for. But I suspect that quite soon they'll be places where people look after children who are doing on-line courses rather than places where people teach.

As I say, excellent news. For almost everyone.


  1. I'm not so sure you're right. Isn't this a further evolution of the modular degree concept? The universities remain providers of the modules, the differenence being that students can now curate their own courses across multiple sources. I can imagine a corresponding evolution with acreditation bodies - they will develop the capability to assess these customised courses. All the exciting action is cross-disciplinary nowadays and I suspect many of the more forward thinking universities are already integrating across disciplines and courses. Integrating content from many universities is a natural next step. I think you might also underestimate the demand for diversity in style, content and personality with these courses.

  2. I agree with all of that. I just can't see anyone paying £30000 to listen to lectures given live anymore.

    As for diversity, yes, there might be room for five or six basic algorithms courses covering similar material in different style. But not several hundred. And one of the six will be the overwhelmingly popular one.

  3. I'm on my 4th MOOC (Algorithms Pt 1 at Princeton).

    For those of us who are (ahem) past traditional university age, it's a huge opportunity. For 18-year-olds.. not entirely sure. Can they replicate the experience of living on-site/in halls at a major research university?

    Of course, for the 2nd and 3rd rank universities that don't really do research and don't have a proper campus.. could be trouble.

  4. Yeah, there are going to have to be support groups for MOOC addicts. Let me know how the Princeton Algs I course go. It seems less theoretical and more Java-oriented than the Stanford one.

    And yes, there are still going to be places where the elite hang out to do research, and there will still be a need for certification bodies so that people can prove that they have learned things, and there are still going to be places where clever young people go to drink and fuck. But whether those have to be the same places is a different question.

    In hindsight I should have gone to college at about 12 years old. The period from when I was 20 to about 25 was the only time in my life when I wasn't obsessed with maths and computers. I had more interesting things to do.

    Cambridge must be one of the worst places in the world to learn maths. It's infinitely interesting, but with a couple of honourable exceptions who show how good it could be, the lecturing is terrible.