Updates on AI and employment. Another labour market summary here.
My theory of art as myth (in the Levi-Strauss sense) is holding up.
Fluctuations in LGBT identification. Bisexuality is probably our natural state, given that homosexual practices amongst men were widespread in Ancient Rome, and are still a social norm in New Guinea tribes today. Evolution probably selects for such: bisexuality correlated with greater libido or success in mating? The puzzle is in exclusive homosexuality, which has only really emerged as a phenomenon in the last half-century in the West, as this is hardly adaptive. Given that many gay men adopt highly (dramatic and exaggerated) feminised modes of conversation, I doubt this is natural and exogenous to sociocultural environment or practices. Gender ideology is of course almost entirely a social fad: intersex chromosone pairs (or in some cases more) outside of XX and XY are incredibly rare. Consenting adults should be able to practice whichever nonconformist sexual practices they wish, yet parents and doctors that indulge minors in this fantasy should be criminally prosecuted: even from a libertarian perspective I couldn't see a clearer case of fraud violating NAP.
Is the Singaporean government overrated? They appear to be indulging in the exact same anti data-centre NIMBYism that plagues Western democracies.
Fracking is good.
Automation exposure by occupational class. Only the relational sector and some trades appear safe. Most white-collar jobs, as we know them, will be done by AI agents.
A case for open-source. The discourse has pivoted too much to the risks and underweights the (potential) gains again.
Much foreign aid spending finances corruption, yet some programs are remarkably effective. We should just spend a lot more on the latter. Governments should be spending 1% minimum on aid. Just financing GiveWell or Coefficient Giving projects and charities alone via this will produce large gains.
Maximising elite influence means even relatively niche movements can punch massively above their weight to become some of the most consequential in a century. This is why I don't lose hope for classical liberal ideas despite the disease of populism and leftism dominating most of the West. Liberalism has become an elite ideology, for its benefit and for the better.
Selection effects could drive the observational link between social media use vs MH. Those predisposed to worse MH are more online. This hypothesis makes many of my anecdotal observations here make sense. In general, despite the causal literature on the effects of social media on MH or cognition returning near-zero or null effects, my intuition is still that social media is negative for both. Anecdotally, the world has become much dumber! I observe heterogeneity: modal individual gets dumber (via platforms like TikTok) whilst the right-tail of IQ gains a boost due to networking with other smart folk. Any classic Coasian model of search and matching dynamics with spillovers would also predict this.
With endogenous risk selection, "larger firms and those with lower markups are less volatile and less correlated with aggregate productivity". As such, "distortions such as taxes and markups can make GDP more volatile". This bears into the longtermism debate too. High discounting and risk-seeking can be confounded observationally, like for instance with drinking.
One mechanism for why automation could reduce wages, although I'm not convinced this has been the case thus far. In many factories, there is almost as much maintenance staff as there are operators.
Remote work reduced birth rates it seems.
Should we make people vote explicitly on the tradeoffs instead? Make the implicit binding constraints explicit? Will this resolve our looming fiscal crises?
On the significance of the OpenAI maths results. The intelligence explosion starting?
“Avoiding the EU's AI and data regulations is valuable, and these will probably only get worse.”
Call centre job losses. AI capabilities in art and even writing, tasks involving creativity and some relational sector component, are also impressive and exponentially increasing.
Meta competing with the dupoly on price? Nonetheless the winners on the capabilities front are clear…
To my knowledge, we are the only country in the world that pays for cars for the disabled.
Emotions (and therefore sentience) determine whether moral worth is positive, in a utilitarian and consequentialist framework.
“China shock 2.0” summarised.
“All finite-order Hecke L-functions are zero-free to the right of 0.8749! This improves on 31/32 (= 0.96875), established only a few days ago by Cureton and Thorner.” Renaissance 2.0? This is how the 21st century could go if Putin doesn't spark World War III (which in my view is the greatest existential risk today, as opposed to AI per se).
Heterodox economics is (political or ideological) propaganda. Yes this is obvious to me, and one reason why I don't lean too heavily towards the Austrian school.
Soul conjecture proof (something to do with noncompact Riemannian manifolds) in three pages! No Siegal zeroes in three pages too! Can't help but think since Quasi-Riemann we have made a lot of progress on Hecke-L functions already. No implications of the Hodge conjecture results for cryptography yet, but these math results (particularly for computational time) are a thing to watch: with AI safety implications too.
I will stick to 4% beers or the occasional (usually red) wine with a meal from now on. Probably more incentive-compatible than hard quantity caps. I will avoid cocktails and spirits, as that is where the markups are highest and they generate a disproportionate share of the externalities too.
More measures of AI capabilities: task edition.
"Across 100,000 simulated revenue and discount rate combinations, 85 percent of plausible outcomes produce a negative net present value of state revenues". Losses from lower income tax revenue tends to outweigh the gains from wealth taxation.
The next AI casualty, this time running deep into the relational sector, and the bureaucrats closest placed to argue for protection?
Migration aside, Reform has a surprisingly good platform.
The singularity? Also strengthens the case for the most rapid change (for better or worse) occuring in cybersecurity (risks). How will the regulatory bottlenecks in biotech shift the distribution of expected outcomes? Will autonomous self-replicating AIs be able to circumvent these bottlenecks by this century?
I would like to see how autism or ADHD diagnosis rates shift once disability is reformed (and sooner or later, it will, if we don't want to end up like France or worse). This reminds me of the spike in non-binary identification during Woke 1.0. Many of our identities are endogenous to our social environment.
November 2026 links
