The bad economics:

Peters, O. The ergodicity problem in economics. Nat. Phys. 15, 1216–1221 (2019). https://doi.org/10.1038/s41567-019-0732-0

This will not really be an RI. There's a good one-page response given by Doctor et al. here (with supplementary material here). I'm making this text post to give some context to this paper, and repost Ben Golub's twitter thread.

Ergodicity Economics

Peters' presents a decision-making theory that is meant to overcome the flaws of expected utility theory. His main, illustrative example is a coinflip game which he describes here. The game consists of multiple periods where wealth either goes up 50% or down by 40% depending on a coinflip. Basically, this is a gentler version of double or nothing. The key question is: how should we play this game? Peters first applies expected utility theory to address this question. Maximizing EU implies that we should invest a positive amount into the game (with log utility, you should invest like 25% into the coin game I think). He believes that this result is nonsensical because the coinflip game has a negative "growth rate." That is, (log(1.5) log(0.6))/2 < 0, which means that repeatedly playing the coin game will eventually cause your wealth to go to zero. From this, he argues that modern economics has failed. In response, he develops an alternative algorithm for decision-making that maximizes growth rates. This alternative algorithm and its goal of maximizing the growth-rate of wealth are entirely teleological and not based on psychological factors, which he argues is a positive quality.

Ben Golub's RI

Golub summarizes his and Doctor et al.'s response here.

This thread gives my own gloss and expansion of some points Doctor et al. raise. Peters and co think there is a hidden assumption of economic theory: specifically, they think expected utility theory secretly assumes a mathematical property called ergodicity. This is false.

Expected utility theory makes 4 assumptions, which are stated precisely and concisely in every graduate textbook. Ergodicity is not among them. EU is not the kind of theory that can hide assumptions: it is like Newtonian mechanics, not like Freudian analysis. Indeed, basic Expected Utility Theory does not need to make any assumptions about time at all, because it is a static theory of decisions under uncertainty.

EE can't deal with uncertainty at all: it wants to be all about time. For infinite income streams, they think there is ONE TRUE UTILITY FUNCTION: the long-run time-average growth rate. Their criterion applies only to the rare cases where that rate is deterministic: again, they can't think about uncertainty (~MuLtIpLe uNiVeRsEs~) at all. In any case, the grand theory they propose is a small special case of the (extant, rich) theory of dynamic choice, and mostly boils down to the Kelly criterion, which is a good idea from the 1950's and well-known in economics and finance. Kelly and contemporaries understood this nice criterion could not accommodate either the full diversity of the dynamic choices people face or of the preferences they have. Thinking through this led to the development of modern decision theory.

The EE crew, in contrast, are stuck in a grandiose exaggeration and misunderstanding of one 1950's idea, thinking that it falsifies all economics since. Some of them, led by @oliver_b_hulme, even think they have run an experiment falsifying EUT in favor of "EE." Doctor et al. demolish this misconceived and confused experiment, beginning with the point that the authors apply static EU in a dynamic context. Doctor et al. also point out that it is trivial to falsify the predictions of ergodicity economics if you want, which is what you might expect with a theory that posits ONE TRUE UTILITY FUNCTION. More fundamentally, Doctor et al. give Peters et al. the basic economics lesson they sorely need, pointing out that the dynamics of, say, household decision-making are more diverse and complex than can be accommodated in their zero-parameter theory.

Doctor et al. have done a generous thing, though unfortunately the learning will likely be lost on the EE crew itself. They are very committed to the bit, and the idea that their magic bullet will not restart all of economics is too bitter a fact to swallow. In their commitment to the hope that they will redirect a mature field with a simple, known idea (and without engaging with current work on the same issues), they embody the main feature of scientific cranks. While it was wrong of @NaturePhysics to give a big platform to such work without soliciting an expert critique, I'm grateful that @jasndoc and coauthors have contributed their time to setting the record straight.

Many thanks to @ShengwuLi and @PietroOrtoleva. Though they're not implicated in anything I say above, I am in their debt for helping me think through both what to say and how to say it!

Peters' Rebuttal to Doctor et al.

Article here, which I've screen-capped as well.

From the first paragraph, Peters doesn't seem to understand where Doctor et al. disagree with him. Also, nothing in his rebuttal actually addresses their points, so I have to wonder if he's even read their paper. Firstly, ergodicity is completely unnecessary for EU - this is specifically addressed by Doctor. But, Peters begins his response by bringing up ergodicity again even though it's irrelevant, Secondly, Peters claims that "entities will often act to maximize the long-term growth rate of their wealth," although he has only ever had one piece of non-evidence (this paper by Meder et al.) which he cited in his original Nature article. The response by Doctor et al. clearly states (Appendix A of the supplementary material) that the Meder et al. paper tests EU with an incorrect experimental setting. Peters does not address this. Furthermore, Peters goes on to claim that people maximize the long-run growth rate of their wealth. This is almost trivially false, since people care about consumption/labor/etc as opposed to hoarding currency - his theory does not account for this nor can it be modified to do so. Thirdly, he still appears to think he's critiquing modern economics, while being unfamiliar with any literature on choice/decision theory from the past 30 years. Furthermore, his particular critique of EU - that it is incorrect to use in dynamic choice problems where ergodicity is violated - is both wrong and irrelevant to most results in modern economics. Overall, Peters does not actually have a rebuttal.