Mid-March 2026 links
“Hayek taught us long ago that it’s nearly impossible to determine what a price should be”. Relevant for the literature on markups and monopsonies.
BMI is a good proxy for obesity, despite what you want to think.
In other words, internet anonymity will soon be a crime in Britain. Although AI makes this largely redundant anyhow, this is signalling more than that…
Institutional investors in housing are in fact a boon for social mobility.
AI reduces the costs of debunking misinformation too, economics edition. Then of course you also have Refine for researchers.
“Nutrition science is the area of science that's suffered the most in the replication crisis”. Admittedly I never realised the replication crisis applied just as strongly to a physical science too. Obviously this imposes doubts on the validity of my entire longevity series, hence why I retracted it (theoretically I could've replicated each and every paper I cited, yet that would've been highly time-consuming). If I'm wrong, I'll never double-down: I admit graciously whilst changing my mind, and everything posted on this Substack is perpetually up for review as my epistemic habits mature or the literatures evolve. Another common criticism of that series was excessive citation of modest effect sizes (often via their reciprocals - implausible dosages required to generate the equivalent effect sizes in humans), so I'll be mindful of this going forward. In general, the reliance on using the model organism equivalent of toddlers in the longevity field plagues the external validity of its claims; notably the calorific restriction results only hold as metabolic defects typical of early childhood clear.
Hmmm. I was suspicious that selection bias would undermine whatever findings came out of Aella's survey (name someone not terminally online whom has heard of her?). Can we take from this that sample size and balance on observables seems to be sufficient for representativeness?
Well said! If you asked me, “is there one person on Earth whom you wish to emulate?”, then Tyler comes close (teetotalism aside, which I wouldn't have even attempted were it not for him). Indeed, the primary goal of this blog is that when he dies and if MR goes along with him, that this Substack becomes a viable and suitable heir.
This is the best evidence I've seen on the causal link between poverty vs crime. RCT. Pre-registered. You don't get much better than this.
Take synthetic control papers with a pinch of salt. Note that this doesn't mean the methodology is entirely junk; it's a research direction that yields promise. Often there's no alternative to stimulating likely counterfactuals.
As Ben Southwood mentioned earlier, the British government has an unhealthy addiction to stealth taxes.
On the ubiquity of insider trading. Also worth mentioning that an underrated benefit of prediction markets is that they harness insider trading for social gain: by incorporating this information explicitly into prices, they allow for more efficient trades. There's no reason not to legalise them in Britain.