TiB 113: When AI writes tax policy; a unified theory of innovation; can central banks learn to play politics; and more...

Welcome new readers! Forwarded this email? Subscribe here. Enjoy this email? Forward it to a friend.

Salesforce’s AI unit published a fascinating piece on training a reinforcement learning model - the “AI Economist” (sorry) - to design optimal tax policies. They run a simulation of a mini-economy where the agents learn to maximise income and the AI Economist learns to optimise the tax rate. This is both an interesting technical problem (see the full paper) and has potentially important real world implications.

One common trade off in tax policy is between output and equality. Higher tax rates (sometimes) reduce work, but also reduce inequality. Salesforce claims that the AI Economist’s tax policy is better at managing this trade off than any real-world policies it tested against - and indeed is strictly better than the US tax code, which it beats for both output and equality.

It’s interesting to look at how it achieves this, though, as the AI Economist’s optimal tax code is odd, at least compared with what we’re used to. For example, compared to the US tax code, almost all high earners would pay a lower tax rate and most low earners would pay a higher one (before redistribution, importantly). Of course, Salesforce is at pains to emphasise that this is a toy model, but it’s nevertheless an interesting special case of the problem of explainable AI: even if AI delivers better results, are we willing to trust it if its methods conflict with our political intuitions?

One of the best known ideas in innovation is Clayton Christensen’s distinction between “sustaining” (or incremental) and “disruptive” innovation, which are usually seen as completely distinct processes. Jerry Neumann has a superb new piece in which he argues this is a mistake: they’re the same process - it’s merely that an optical illusion causes them to look different.

That optical illusion, Neumann argues, is the power law: if the impact of innovation is power law distributed (Neumann, as it happens, wrote the definitive essay on power laws in venture), the big outcomes will stand out because they’re big and the small ones because they’re common - and they’ll look like two different things (The same, Neumann says, applies to Kuhn’s distinction between “revolutionary" and “normal” science). Neumann proposes a model to explain why we should expect the power law to apply: if you imagine technology as a tree of layers of technologies, improvements in the most fundamental layers ripple up through many more high level technologies and so have much more impact. It’s worth reading this piece on “discontinuous progress in history” for some (loose) empirical backing. 

The problem is that there are strong incentives for scientists and technologists to work on incremental rather than fundamental problems, as we discussed back in February (Neumann cites the paper by Bhattacharya and Packalen that we discussed then). If we can fix that - and we’re starting to see more meta-innovation in the space - the upside is enormous. 

A couple of weeks ago, we discussed the end of the era of apolitical central banking. As I said then, the critiques of the Fed and its counterparts are clear - but how will the central banks respond? This week I came across this fascinating paper by Benjamin Braun (via Adam Tooze) on the challenges central banks have in engaging and influencing the public.

The argument is that for several decades successful monetary policy - and therefore central bank legitimacy in the eyes of the public - simply meant keeping inflation low. But following the global financial crisis, central banks took on a broader mandate - and public scrutiny increased. This posed a challenge: the public’s understanding of monetary policy - what Braun calls the “folk theory of money” - is based on largely incorrect premises, but during the Great Moderation there was no incentive to educate the public. “Strategic ambiguity” in communications - such as this amusing example - very much suited Greenspan et al.

Now, however, correcting the “folk theory of money” is firmly in central banks’ favour if they want to reassure the public that “money printer go brrr” won’t lead to runaway inflation. Braun points to two videos the Bank of England made in 2014 that try to do just this. This sort of thing does, though, feel like bringing a knife to a gun fight, given the form of central banks’ critics. Ironically, if central bank leaders don’t learn to play politics, central bank independence may not last long. 

If you enjoy Thoughts in Between, I’d love you to forward it to a friend or, even better, share it on Twitter / LinkedIn.

I’m always happy to hear from you: you can find me on Twitter or just hit reply.

Until next week,

Matt Clifford