Matt's Thoughts In Between - Issue #54

Forwarded this email? Subscribe here. Enjoy this email? Forward it to a friend.

What does the (UK) public think about AI?

This week I stumbled across this short and interesting paper on public attitudes to artificial intelligence. The three authors ask a representative sample of the British public about AI and its impact. In particular, they put eight scenarios - four pairs of linked positive and negative narratives - to the respondents and ask them to assess their likelihood and impact.

A number of findings are interesting. First, less than half the sample could give a plausible definition of AI. Second, two of the narratives that the researchers had intended to be positive (one about strengthening national power, one about AI becoming a "perfect friend") were assessed as "more concerning than exciting". Third, one of the biggest concerns cited was the inability of ordinary people to influence how AI develops.

The last may end up having important geopolitical consequences if democracies slow down the development of AI (and tech more generally - no prizes for guessing where this crazy study took place) to accommodate public opinion. Interestingly, fiction will play an important role in shaping public views on AI (as it did in the response to nuclear weapons). Just 14 of the 1000+ respondents made reference to AI in fiction. That suggests a big untapped opportunity for sci-fi writers who want to change the world...

If Ronald Coase knew about blockchains...

Taylor Pearson at Ribbonfarm published a long and interesting essay this week titled "Markets are eating the world" (cf. Marc Andreessen). The basic premise is that technology is rapidly reducing the minimum viable size of firms. As a result, we see far more gig work and micro-companies - and Pearson believes this trend has a long way still to go.

Pearson draws on Ronald Coase's famous theory that transaction costs determine the size of companies. High search, bargaining and enforcement costs cause firms to "in-source" work, which makes them larger. These costs have been reduced by a series of technologies, above all the internet. This enabled a wave of micro-business, while making giants of the aggregators (and even these are "small" relative to their industrial forebears).

For Pearson, the next breakthrough that will enable another big reduction in firm size are blockchains. The idea is that replacing interpersonal relationships with cryptography-based trust will reduce transaction costs again. The argument here seems a little hand wavy to me. But something like this (e.g. this - still the best crypto essay to my mind) remains the bull case for blockchains: distributed ledgers may allow us to imagine and create new kinds of organisations. And, as Pearson argues, those breakthroughs are historically rare and, when they happen, very important.

What are the long run effects of easy money?

We've now had very low interest rates and quantitative easing (QE) in the West for over a decade. That's generally been welcomed by politicians of all stripes (when in office...) and economists broadly. It seems increasingly clear, though, that ten years of historically low rates may have some unintended consequences.

A new paper by a group of economists suggests that one surprising effect is greater market concentration and slower productivity growth. One of the authors, Atif Mian, walks through the basic logic in this thread. In short, low interest rates give industry leaders a more powerful incentive to invest than their competitors. The result is another (see TiB from two weeks ago) push towards more monopoly-like markets.

And then a few days ago David McWilliams argued in the FT that QE has been a major driver of rising left-wing sentiment among the young in both the US and the UK. Rising asset (especially house) prices have caused greater inter-generational inequality, which in turn spurs demand for more redistribution. Early critics of QE worried about hyperinflation; that turned out to be unfounded, but its long run impact may yet be profound.

Quick Links

First, an apology: the most clicked link last week, on the "Ethiopian miracle" was broken. The correct link is here.

Now that the future is here: Founder of Wired Kevin Kelly's predictions for 2019, made in 1999
Move over MBTI: Long but brilliant piece on how Magic the Gathering explains personality (Thanks Hampus for the link; I think I'm Blue/Black or Blue/White...)
AWS is the best VC. Superb thread on why it makes sense for Lyft (and many other startups) to pay Amazon tens of millions a year
Way beyond the Turing test. Surprisingly good generator of puns from NeurIPS 2018 (please send me the best you find)
Still the most important tech/geopolitics story of the year. (Even) more on Huawei, China and the US

Your feedback

Thanks for reading all the way to the bottom. We're on 998 subscribers at the time of sending - so if you like TiB, do forward it on to two more people who might enjoy it. If you have comments or other feedback, just hit reply.

Until next week,

Matt