Power and Progress (Book Review)

A couple of weeks ago I participated in Creative Destruction Lab’s (CDL) “Super Session” event in Toronto. It was an amazing convocation of CDL alumni from around the world, as well as new companies and mentors. The event kicked off with a 2 hour summary and critique of the new book “Power and Progress” by Daron Acemoglu and Simon Johnson. There were eleven of us charged with summarizing and commenting on one chapter each, with Daron replying after 3-4 speakers. This was the idea of Ajay Agrawal, who started CDL and is a professor of strategic management at the University of Toronto’s Rotman School of Business. I was thrilled to see a book given a two hour intensive treatment like this at a conference, as I believe books are one of humanity’s signature accomplishments.

Power and Progress is an important book but also deeply problematic. As it turns out the discussion format provided a good opportunity both for people to agree with the authors as well as to voice criticism.

Let me start with why the book is important. Acemoglu is a leading economist and so it is a crucial step for that discipline to have the book explicitly acknowledge that the distribution of gains from technological innovation depends on the distribution of power in societies. It is ironic to see Marc Andreessen dismissing concerns about Artificial Intelligence (AI) by harping on about the “lump of labor” fallacy at just the time when economists are soundly distancing themselves from that overly facile position (see my reply thread here). Power and Progress is full of historic examples of when productivity innovations resulted in gains for a few elites while impoverishing the broader population. And we are not talking about a few years here but for many generations. The most memorable example of this is how agricultural innovation wound up resulting in richer churches building ever bigger cathedrals while the peasants were suffering more than before. It is worth reading the book for these examples alone.

As it turns out I was tasked with summarizing Chapter 3, which discusses why some ideas find more popularity in society than others. The chapter makes some good points, such as persuasion being much more common in modern societies than outright coercion. The success of persuasion makes it harder to criticize the status quo because it feels as if people are voluntarily participating in it. The chapter also gives several examples of how as individuals and societies we tend to over-index on ideas coming from people who already have status and power thus resulting in a self-reinforcing loop. There is a curious absence though of any mention of media – either mainstream or social (for this I strongly recommend Martin Gurri’s “Revolt of the Public”). But the biggest oversight in the chapter is that the authors themselves are in positions of power and status and thus their ideas will carry a lot of weight. This should have been explicitly acknowledged.

And that’s exactly why the book is also problematic. The authors follow an incisive diagnosis with a whimper of a recommendation chapter. It feels almost tacked on somewhat akin to the last chapter of Gurri’s book, which similarly excels at analysis and falls dramatically short on solutions. What’s particularly off is that “Power and Progress” embraces marginal changes, such as shifts in taxation, while dismissing more systematic changes, such as universal basic income (UBI). The book is over 500 pages long and there are exactly 2 pages on UBI, which use arguments to dismiss UBI that have lots of evidence against them from numerous trials in the US and around the world.

When I pressed this point, Acemoglu in his response said they were just looking to open the discussion on what could be done to distribute the benefits more broadly. But the dismissal of more systematic change doesn’t read at all like the beginning of a discussion but rather like the end of it. Ultimately while moving the ball forward a lot relative to prior economic thinking on technology, the book may wind up playing an unfortunate role in keeping us trapped in incrementalism, exactly because Acemoglu is so well respected and thus his opinion carries a lot of weight.

In Chapter 3 the authors write how one can easily be in “… a vision trap. Once a vision becomes dominant, its shackles are difficult to throw off.” They don’t seem to recognize that they might be stuck in just such a vision trap themselves, where they cannot imagine a society in which people are much more profoundly free than today. This is all the more ironic in that they explicitly acknowledge that hunter gatherers had much more freedom than humanity has enjoyed in either the agrarian age or the industrial age. Why should our vision for AI not be a return to a more freedom? Why keep people’s attention trapped in the job loop?

The authors call for more democracy as a way of “avoiding the tyranny of narrow visions.” I too am a big believer in more democracy. I just wish that the authors had taken a much more open approach to which ideas we should be considering as part of that.