Therefore, I have been thinking deeply on the prospects for classical liberal ideals here, and how best to maximise their influence. Along with many others of a libertarian leaning, as best exemplified via the infamous neoreactionary ecosphere, I had previously thought that democracy was a historical mistake, and should be abandoned. This line of reasoning, it turns out, is a mistake, and the more typical liberal consensus here is in fact correct.

You can skip the last two paragraphs of section I if you are unfamilar with the economics, otherwise here I explain why. A good liberal changes his mind with additional evidence, and so I have now been persuaded by my fellow liberals on what could be one of the defining battles of our lifetime: the fight to protect democracy.

I. Democracies maximise welfare

To see why, consider the growth performance of advanced Western economies after WWII. Countries undergoing more catch-up growth saw more of it. Italy, Japan, Austria, France, and Germany with its renowned Wirtschaftswunder. Britain and the US, closer to the frontier, grew at slower rates. All in the former category saw econonic miracles occur within a generation, despite variation in policy and economic institutions leaning in a statist direction.

Compare and contrast with Spain and Portugal. Whilst these experienced economic miracles of their own, they arose much later, in large part due to the policies of their respective dictatorships promoting autarky. Indeed, starvation was ubiquitous in the post-war years for Spain. Plus note that the Franco and Salazar regimes are generally considered amongst libertarians to be amongst the better autocracies too, yet quite clearly the evidence points towards them being a barrier to growth. The Spanish miracle was relatively modest compared to that in democratic nations anyhow.

The mechanism seems to be the one well-documented, that democracies can fuck up a lot (as France is doing today) yet often reverse course or mitigate before the damage is too late, hence impose much less variance on outcomes than autocracies (despite the best seeing unusually high levels of progress). Schumpeter was right in decentralised institutions promoting innovation via creative destruction and spontaneous order, and this is often what we see in democracies compared to stale and stagnant autocracy.

Yes, democratisation is linked to greater redistribution. Most of this in the West appears to be social insurance against life's most common shocks though: old age, illness, disability, and unemployment. Conceptually, this is different from inequality-aversion per se. Social insurance can be justified on risk aversion grounds as well as reducing inequality, yet both CRRA preferences and inequality aversion would show up as a curvature in an Atkinson-style social welfare function. Nonetheless, it seems reasonable that growth is not all people care about, and that as their incomes (and consumption and leisure) rise, so does their desired consumption of insurance.

This can be done via private markets, yet classic information asymmetry problems with heterogeneous financial frictions generate separating equilibria, so some would pay higher rates in private insurance markets relative to social insurance systems. Pooling equilibria will generate moral hazard, and taxes will be distortionary, and possibly suboptimal depending on elasticities, social preferences, if they differ from ideal Lindahl prices, and so on. Moreover, in many OLG models, if population growth exceeds interest rates, then the standard PAYG public pension systems are Pareto improving. In summary, welfare states could be normal or luxury goods, so democracies expanding them as they grow would be a correct aggregation of preferences, and hence be optimal.

II. Libertarian autocracy relies on the illusion of a philosopher king

At this point, my thought was yes, but social planners will not have accurate information on the relevant parameter values, and are not benevolent anyhow. Hence policymakers should be biased towards decentralised information aggregation via markets and non-state interactions, and so an autocratic nightwatchman would therefore be superior to the liberal democracies we see today.

Yet this somewhat Hayekian perspective works the other way too. If social planners are not benevolent but a set of rational self-maximising regulators, lobbyists, politicians, lawyers, bureaucrats, and other interest groups, then a libertarian autocracy would never stay libertarian and such elites would abuse this concentration of power. They would hold private information, exploit that, so lead to worse aggregation outcomes even if Arrow shows aggregation in democracies can never be consistent.

These dynamics apply just as much to “enlightened” technocratic experts too, as the UK House of Lords and the European Commission can attest. The former, meant to grant experts representation in the legislature, has decayed into a trophy for party donors and a retirement home for politicians. The latter is drawing up new border rules (ETIAS) so detached from the public view that no one even knows exactly what the framework is, let alone what the implications are. This is the best example for why the usual arguments on democratic accountability and debate apply, otherwise you risk legal uncertainty.

Not only is this public choice 101, but is the precise argument in favour of democracy that many liberals (including ordoliberals for concentrations of private power, and the downstream effects this entails for institutions) espouse. Even the good autocracies are censorious, and abuse their powers for their own gain1. Meanwhile, the fact that the US constitution and separation of powers appears to be constraining Trump is a vindication of the normie liberal argument on democratic accountability and checks, despite the decay in shared norms. The rule of law acts as a credible commitment to elites not to expropriate property for their own gain, or arbitrarily arrest their competitors (business, political, or otherwise).

Share

III. Why a nightwatchman state will never happen

All of the above points probably explain why a nightwatchman state has never occured at any point in history, except perhaps failed African states. For a nightwatchman to have the state capacity to enforce the contracts necessary for markets to thrive, or the non-aggression principle, it will need the fiscal capacity to tax, which means the ability to run welfare states. Either this occurs with democratisation, or repression (implying eventual elite capture by rent-seekers) is necessary.

Such a minarchist vision was popularised by Nozick2 via (in my view) an excessive reliance on rigid Kantian imperatives anyhow. Liberals appear split on deontological natural-rights based (Nozick, Locke, Kant, etc) vs utilitarian and consequentialist approaches to human flourishing and liberty. Nonetheless, the classical liberal tradition goes beyond and long predates Ron-Paul style dogma imported from across the Atlantic. Ordoliberals helped build Germany’s social market economy. Friedman is reviled for seeking to reform rather than abolish central-banking. Smith supported public schooling, and even Hayek (aligned with the ordoliberal school) was in favour of some social protections. Today's highly Americanised stereotype of libertarianism is simply not an accurate representation of our entire beliefs.

IV. On why you should not depend on first-principles

I also think I relied too heavily on first-priciples reasoning, neglecting the mulitiplicity and path-dependent nature of history. This is an issue for many terminally-online, whom are younger so more reliant on first-principles or domain knowledge, yet deficient in the tacit accumulated knowledge on life, society and its norms we call “wisdom”. Outcomes are therefore state-contingent. Yes, Pinochet might have been good for Chile, and Lee Kuan Yew good for Singapore. Yet the Wirtschaftswunder was in part a result of ordoliberal policymakers compromising with the unions. Germany's social market economy endured a lot more state investment, union power, and social protections than most classical liberals or libertarians may like. This was probably necessary to prevent it going communist, keep it aligned to the west, or its economy being wrecked by militant union strife. In general, my worldview seems too parochial: dominated by online discussions or articles, and I should read more physical books and travel more.

Liberal democracy3 arose as an equilibrium solution to the horrors of fascism, then communism (red fascism!) after the fall of the Berlin Wall. Most of Europe was fascist or communist for most of the last century, so an imperfect market-based (even if mixed) liberal democracy is a remarkable accomplishment that was never inevitable. Neoreactionaries are overwhelmingly hyperonline Americans for a reason, and hardly any have even visited, let alone lived, in an autocracy. Today, the advertisement for dictatorship is Putin, and it is highly likely that the Ukraine war would never have happened had Russia instead pivoted towards the West and democracy as opposed to their elites indulging in self-pity.

Acemoglu, in one of his better papers, published a robust estimate that democracy is indeed good for growth. Whilst democracies are more redistributive, they also spend more on public goods, infrastructure, and human capital (health and education). This makes intuitive sense to me. I think this mechanism could even be broader. In democracies, you have a lot more cultural choice, and so a richer intellectual life, which of course contributes to innovation. In Britain, you could listen to punk rock or Asian folk music or psychedelia. In autocracies, you are often restricted to music tastes favoured by the regime, as Max Weiss’ pieces highlight.

V. Conclusion

Do you think those of us writing on X and Substack, and engaging in public discourse there, could do so in an autocracy? Would we be able to participate in AI-safety discourse and possibly influence the alignment strategy of the labs in what might be an inflection point for our future existence? If you enjoy the fruits of participation in social media, a public good, you are free-riding off the precise benefits of democracy that many online criticise today.

HDI comprises of proxies for human capital alongside income levels. Average years of schooling and life expectancy. There are three index scores, normalised:

1. Let Ih be the health index. Its minimum is 20 and maximum is 85. Hence Ih = (E[life]-20)/(85-20)

2. Let Ie be the education index. Mean and expected years of schooling enter as separate variables. Expected years of schooling is state policy. Mean measures the average of how many years people actually study for. However, the former is normalised and capped at 15 whilst the latter is capped at 18. So Ie = (mean/15 + expected/18). Note that university education is excluded!

3. Income is GNI per capita at current PPP. This is both logged yet capped at $75k with an $100 minimum. Iy = ln y - ln100. This normalisation leaves income differences between countries heavily compressed as a result. By design, the index places a relatively low weight on income.

From this, you obtain HDI = (ΠiIi)^(1/3) for i=(h,e,y)1. You get a geometric mean for your score. Whilst this plus the normalisation makes the calculation tractable, the vast differences in living standards across the world are understated by design. This is due to a revision change in 2010 from a simple arithmetic mean: designed to weight each index more equally, yet penalising rich outliers. The very exponential compounding we're adjusting for is what leads to long-run prosperity in the first place!

Moreover, each additional year of life increases your HDI by HDI/(3Ih). Hence this implies the gains to living standards from life expectancy increases are roughly proportional to income, yet income (given the capped logged form) yields diminishing returns. Income is quasiconcave yet the other indices are linear! I know of no QALYs nor other hedonic price adjustments that arise from this functional form. By design, HDI implicitly weights gains to health and education more heavily than gains to income, which biases towards countries adopting universal coverage and schooling in these areas, even at extortionate and distortionary taxes that harm growth.

Then expected years of schooling is an input that can be increased simply as a matter of policy. Indeed, communist states invested heavily in (years of) schooling, hence on this front they perform better on this index than their dysfunctional economies suggest. In this sense, HDI is biased towards left-leaning polities.

Putting aside the somewhat arbitrary selection of proxies for human capital or public goods provision (quantity is not quality after all!), with no reliance on market prices or values for such inputs whatsoever, this implies a major flaw in Sen's capabilities approach to me.

Despite this, I'm not a GDP maximalist either. Non-market income and leisure are excluded from measurement, which is often noisy and (especially in counties with dysfunctional institutions and poor state capacity) heavily politicised. “Running an economy hot” via printing money and injecting fiscal stimulus will (in the short-run) boost output. You could maximise the long-run growth rate (potential output), yet no one really knows what this figure is. Communist five-year plans are essentially textbook Solow (capital investment). We have reason to believe it could be negatively correlated with fertility too; throwing the welfare calculus into question.

GDP is a good proxy for living standards. Revealed preferences clearly show the overwhelming pattern of migration at every stage in history has been from poorer to richer nations. Many correctly argue that it's correlated 0.9 with most of what we care about. Yet with AIs, why not build a model of all the variables you care about when evaluating the standard of living of a location yourself, in a granular manner? You can use whichever functional form you wish: nonparametric even. Alternatively, revealed preference captures much of what enters as a residual in any measurement exercise.

When evaluating the living standards of a country, you must factor in Arrow, again. We lack any reliable means of aggregating social preferences in a coherent manner. This applies to how one weights material prosperity, GDP, non-market activity or leisure, and so on too. This is also an argument against institutions that promote the tyranny of the majority. In general, the lesson I draw is that governments should, as much as possible or feasible, leave individuals and organisations to decide for themselves what makes for a good life. The best metric for living standards, I think, is liberty. This is highly correlated with the other common proxies such as GDP anyhow. Non-market indicators such as social cohesion (how safe you are) or participation (philanthropy, perhaps faith, and our relationships) also matter.

You could prioritise Pareto efficiency, yet in cases such as public goods or enforcing property rights over pavements (for obvious reasons public ownership, so a second-best equilibrium at best), this framework is unfeasible, then second-best or VCG mechanisms are highly state-contingent and introduce ambiguity.

In general, any reliance on any single metric, whether HDI or GDP or so on, is highly likely to neglect some important dimensions that matter for our happiness, virtue, and prosperity. These value judgements are the exact reason we will never converge onto agreement on the best policies or institutions to adopt. Their subjective, nonparametric nature often leaves these normative assumptions implicit and hidden: no one knows their exact social welfare functions of course! Nonetheless, to find common ground in our polarised world, and agree to efficient or second-best compromises, we should place more effort into describing what we think makes for a good world.

One can broadly define the Pakistani food scene in Bradford as two sets of markets. One, exemplified by Akbars or International, is your typical large-scale restaurant operation, and Lala's certainly competes with them. The other, characterised by the Sweet Centre, takes on a more café-like format and specialises in Pakistani homecooking. Indeed, I was the only solo-diner in Lala's last night: everyone else was in a couple or family. On the other hand, me eating alone would not be out of place at the Sweet Centre.

The large-scale of the operation, the large menus catering for everyone, and the lavish entrance with chandelier were the only heuristics broken. As typical, I ordered the Magaz. It was cooked karahi-style; typical for most of Bradford (Sweet Centre uses the balti method). However, the chef was unique in his method. The brain was infused with the sauces and the ghee, and this blend I haven't encountered. Over time, it solidified into the scrambled egg texture that it should take. The taste improved as time passed, which is a core characteristic of brain cooked well.

I would say this is the second-best Magaz in Bradford I've encountered so far, and the only reason it's not the best is because Jinnah's produces a top-10 in the world level Magaz. To compare everywhere to Jinnah would be imposing unrealistically high standards, like comparing every fine-dining chef to Heston Blumenthal or Gordon Ramsay2.

Impressed, I decided to get the butter chicken, which is where the fame and reviews really stand out. This dish has the reputation as Lala's highlight. It was the best butter chicken I've ever had. Not sweet at all, with a gravy-like sauce that again stabilises over time.

Here, you clearly have a chef that cares about food. You notice that both dishes improve with time. This is deliberate, to accomodate the large families and couples socialising. This means that it's possible in the production process to actually cater to the very thing that was previously considered a negative, at least for Pakistani cooking (quality is strictly decreasing in time for most cuisines).

The food was pricey, yet the portion sizes were also larger, so one should normalise. Moreover, if we account for economies of scale in food consumption within a family, then one or two large dishes with breads serving a whole household could be surprisingly good value and invert this markup logic. I reckon, as a solo-diner, my presence cross-subsidised the families.

Moreover, I was wrong to discount large family-catering operations. The place saw plenty of Pakistani families, which could indeed support high quality Magaz on the menu. I was far from the only person ordering it. If I'm not the only person in a curry house ordering brain masala, then this is the heuristic I use to judge if it's good.

Just as I claimed that authenticity can be arbitraged (Piassa in Leeds ticks the superficial boxes yet does not serve its kitfo raw or even leb-leb), I discounted the possibility of a pooling equilibrium arising in décor, at least for large-scale operations. Takeout prices were substantially cheaper, yet the dine-in experience means this cannot be regarded as a simple arbitrage opportunity. Therefore the lavish entrance serves an economically important function to justify the business as a restaurant as opposed to a takeaway. This heuristic therefore needs to be revised to account for the existence of delivery apps.

For me, it's far more important to look at the demographics of the customer base, and what they order. Note this is correlated but distinct from the demographics of the area. Leeds has a larger South Indian population than Bradford, yet Tharavadu (mostly British customers) was awful. Bradford has hardly any Tamil presence, yet Kurumilagu is full of South Indians (a lot of Keralans) eating dosas. Looking at what is ordered is vital too: MyLahore gets plenty of Pakistani customers yet eating mostly burgers and chips.

Heterogeneity in the food preferences of the Bradfordian Pakistani community also confounds. Magaz is not universally accepted. Some swear by it, and some prefer butter chicken. Older, more devout generations tend to love brain curry. Younger, more Westernised, tend to like butter chicken. Therefore, a successful Pakistani restaurant in Bradford needs a menu and operation that caters to both tastes. The presence and popularity of “unauthentic” dishes cannot be held against a place at all.

My main complaint lies in the quality of the breads and drinks. The lassi was store-bought and pre-made, with a bland homogenised taste. The chapatis were also bland, and you were better off eating with cutlery than with chapatis. To the extent there has been a decline in quality as the restaurant scales and standardises in volume, it has been in the cross-subsidies rather than the main dishes. However, everywhere but Jinnah compromises on chapatis. Given the fact that customers tend to get annoyed at the wait required to produce a fresh lassi, only Kurumilagu does it. In this sense, Lala’s is no different from most others.

The main lesson I draw is that, if enough people recommend somewhere, you should trash whatever heuristics or guides you have in the bin and go. Other people's preferences encode valuable information too. I still stand by my model, as it predicts some excellent places to visit, yet it can miss some excellent establishments too.

More formally, we can define a language as a set of semantics S, and a set of relational orderings R1,…, Rn. There exists a permutation, f:S→S, such that every relation and ordering is preserved:

Ri(x1,…,xk) < - > Ri(f(x1),…,f(xk)) for i = (1,…,n)

Hence each word is given a unique meaning once we consider its relations to other words or concepts. In other words, f(.) is an automorphism, and all languages are sets linked via relation-preserving automorphisms.

Crucially, this automorphism preserves invariance in meaning. If there exists an automorphism linking elements a to b in S, then a corresponds to b in a manner such that they're semantically equivalent. In general, if you replace all the elements of a group with different objects, yet preserve the same permutations and functional operations, then such a group is isomorphic. Languages are no different. Indeed, Tarski's definition of truth uses this intuition over isomorphisms mapping between different formal structures2.

In general, I'm inclined to agree with the structuralist approach emphasised by the likes of Levi-Strauss (albeit the set of relations is broader than binary!). However Derrida and the deconstructivists are simply wrong when they claim this implies that all semantic meaning is arbitrary, and a function of social power structures and the like. The fact that languages are defined via an automorphic structure means that semantic meaning is invariant, including to social fads and privilege. Of course, correspondance to truth and objective reality is debatable, but the deconstructivists are likely wrong there.

The libertarian case for ending welfare for bad luck always rested on two premises:

1. Philanthropy is crowded-out. Yet this is also a statement on allocation. Bad luck events in Western economies are simply not a priority.

2. Private insurers would step in. Yet we disagree whether many AI scenarios are even insurable. If this is the case, then many SWFs imply that the state should step in. Indeed this is standard in the HANK literature. Inequality aversion is based into pretty much all realistic social preferences (even libertarians care about philanthropy, which is another expression of inequality aversion).

Therefore, throughout the AI transition, ending the welfare state is off the table. Too many people are at risk of unemployment via no fault of their own. You can't "just get a job" in many of these scenarios, as two seconds reading search theory would tell you. Indeed (pun intended), the junior and entry-level markets already seem to be struggling.

As I stated in my last article, I place less weight than standard on lags and bottlenecks. Yet they exist. This gives us time to respond to the data in-real time and for economists and policymakers to react appropriately. Therefore, I am confident we will muddle this transition on the employment side, at least with a solution acceptable to most.

Yet I no longer see the old fashioned, dogmatic doctrine of minarchism as optimal for the years ahead. In fact, it updates me that it was never optimal - the Industrial Revolution was also a transition of comparable magnitude, and the 20th century already saw disemployment effects associated with automation, albeit not on net with the distribution highly unequal. So classical liberalism, yes! Libertarianism, no.

Subscribe

There is a high chance of achieving (1) given the competitive dynamics between geopolitical rivals in particular.

(2) might occur in the absence of kill switches or neglect in cybersecurity relating to GPU weights.

(3) is too tenuous at this stage. Both ends of the literature suggest no one really knows what AI values are, nor what mechanisms and enforcement designs contain them. Yes we have misalignment, sandboxing errors, and reward hacking. Yet this does not imply malicious objectives from AI as a "second species" will arise in the future, although this cannot be ruled out: particularly with RSI causing an intelligence explosion as Turing predicted. However, if it becomes difficult for us humans to control them, then automating AI research could become especially socially valuable, which was why the labs racing was fully consistent with them taking AI-safety and X-risks seriously (their self-harming, in valuations, calls to pace the frontier are the best evidence possible for this). It remains the case, as the most likely X-risk channels highlighted here suggest, that use by malicious human actors pose the greater risks.

(4) however was demonstrated with Hugging Face. This therefore raises my p(doom) slightly.

On the other hand, Yudkowsky and the doomers suggest that constitutions have completely failed in alignment. The theory and data do NOT support that. The evolutionary game dynamics, and the AI obsession with pure math (which by definition has no real-world application, and is primarily pursued for beauty, and manifolds or group theory are certainly beautiful!), suggest a harmful second species colluding against us to wipeout humanity is not an inevitability, nor even likely.

This makes me fairly optimistic that, with sufficient talent directed towards advancing AI-safety, that we'll solve the alignment problem once RSI occurs. However, this is not guaranteed either. The labs are making the biggest bet that humanity has taken by far; that RSI will automate and solve the alignment problem, and the ensuing intelligence explosion will solve most of humanity's suffering.

My everchanging p(doom) probably sits somewhere just over halfway (to account for the right-tailed distribution here) between my pre and (initial) post Hugging-Face likelihoods: perhaps 8%. It would be unwise not to update since this saga unfolded.

On economic disempowerment

Some of the latest macroeconomic models from the likes of Anthropic suggest rather grim futures for employment relative to what economists were previously predicting, yet this is also consistent with scenarios implied by task-bundling. I think the economics profession has a bias towards being blasé over the potential disempowerment effects implied by higher unemployment. Even the Great Recession was a scar that took many years to recover from. High unemployment is costly, and if AI raises it to levels of a modest 7-10%, that will be a significant transition in itself from years of sub 5% unemployment throughout a Great Moderation. Perhaps the recent backlash from the mathematicians, and some of the sentiment behind the anti data-centre movement, reflect these (increasingly justified) anxieties?

Moreover, once we accept that the assumption of constant saving in the classic Solow framework is violated, there are infinitely many models possible to write. OLG Aiyagari-style models, HANK, and so on. Many parameters and calibration choices, over elasticities, discount rates, and the like. Hence increasingly, microfoundations becomes less reliable as a tool for disciplining the range of feasible predictions and outcomes, and we should pay more attention to the Lucas-Critique violating aggregate forecasts on capabilities and scenarios. My preference is for the task-bundling approach (directly addresses the lags whilst microfounding the production process itself), yet in the long-run this implies labour markets will be driven by capabilities too.

Anthropic got some pushback within the economics field for its recent model, yet I like the approach of letting laymen adjust their calibration in an interactive model then getting their conditional predictions. I am close to their substantial scenario. Labour share is likely to fall significantly, and average wages will rise (although note that this is contingent on calibration, and even in a standard neoclassical framework, automation could be bad for wages). Overall unemployment rises by a few percentage points (itself a big deal!). Surprisingly, white and blue collar work are affected similarly1 (perhaps a divergence occurs later?). GDP growth of 4% by 2030 is much more feasible to me now.

Given the recent unprecedented declines in inference costs, we should no longer assume that lags and bottlenecks will take as long to “clear” as they once did. With superintelligence, it's plausible we are getting the sign wrong here, at least a decade from now. Of course, technology is not the only constraint, and with NIMBYism playing a salient role in public life throughout my lifetime, it's possible that rent seekers are much better at organising within institutions.

Nonetheless, this is orders of magnitude more dangerous than nuclear weapons. That's enough for me to take AI-safety incredibly seriously!

Disclosure: I've used Codex and ChatGPT 6 Astra to help with the calibration (in particular the sources and data gathering), yet the judgement is my own.

Gartner predicts total cybersecurity expenditures of roughly 0.2% of global GDP in 2026. Even if in a pessimistic scenario, you expect these costs to rise tenfold within the years to 2030, this would amount to no more than 2% of GDP by 2030. Similarly, the UK government estimates 0.5% of GDP. Even under plausibly pessimistic scenarios, it's hard to see AI-driven cyber risks being uniquely catastrophic in expectation1. On the more pessimistic end, Jamilov et al (2026) estimate total costs less than 1% of global GDP. We should also account for learning-by-doing dynamics and general productivity gains increasing the efficiency of cybersecurity measures, which should exert downward pressure on costs relative to GDP.

Suppose there is a small tail probability of an extreme cybersecurity event costing multiple trillions (e.g. a wipeout of the entire banking system). Great Depression saw an over 30% collapse in US GDP. Covid saw a near 20% fall. Disastrous, but hardly extinction. These remain realistic upper bounds for my tail-risk scenarios, unless anyone can point to better anchors?

As for my previous claims, it is apparent that there are more channels via which projected cybersecurity costs, and a “doom trade”, could influence the markets and hence affect optimal portfolio allocation. These include:

1. Cyber insurance premiums

2. Corporate debt spreads (those investing less in cybersecurity, or more vulnerable, might see higher yields)

3. Equities (more vulnerable firms should see lower valuations - indeed Jamilov et al tries to measure this)2

4. Prices and valuations for cybersecurity contractors

Moreover, perhaps Kelly's Criteron oversimplifies, as there will be many events before doom implying such a scenario. Doom will, if it happens, unfold sequentially, which opens up opportunities for shorting. Moreover, Bayesian agents can form subjective probabilities and account for this with higher uncertainty. This does not undermine optimal portfolio allocation nor standard asset-pricing models.

In general, I see cybersecurity as one of the primary risks associated with AI, yet extinction is highly unlikely on this basis. Scheming and misaligned agents are classic principal-agent and mechanism design problems that are solvable. Bioweapons manufacturing would rely on being able to go undetected in the physical world, which remains implausible (especially as AI would increase detection). Lab leaks are a far more salient concern, yet pandemic preparedness confounds, and AIs will likely assist in this. Disempowerment, and extreme totalitarian powers and government surveillance, alongside the integration of AIs into drone infrastructure, remain my greatest concerns, yet these are hardly existential.

I was one of the commentators that overreacted to the Hugging Face incident. Considering the general equilibrium social and economic dynamics more broadly, p(doom) cannot plausibly exceed 10%, and is probably below 5% (similar to most forecasts for nuclear warfare).

The mistake doomers are making is inferring the likelihood of extreme scenarios from capabilities. Yes, Hugging Face is consistent with a scheming paperclip maximiser. It's also consistent with more boring principal-agent misalignments that standard economics might predict. Capabilities trends can be extrapolated reasonably accurately from Moore's Law which tends to be invariant, yet doom scenarios rely on complex equilibrium interactions amongst individuals, firms, institutions, and agents. Ignore the Lucas Critique at your peril!

In general, most bureaucratised regulatory regimes are bad. In some cases, like nuclear technology or ensuring planes are safe to fly on, they're welfare improving. It's clear that AI belongs in the latter category.

Also trace inversion is a thing, so you can get open source weights to be roughly similar to that of Fable/Mythos. The slowdown camp were right. At bare minimum, all labs should pause training, further development, and releases indefinitely until we figure out how to align AIs and regulate them (and their use by humans). I reckon the current capabilities we have now are sufficient for accelerating progress towards curing cancers etc. So my balance has shifted towards minimising the existential risks now.

Pause AI advocates are correct. If governments could coordinate internationally to achieve such (big if), I'd support it. I used to be highly sceptical of doomer arguments, yet Hugging Face is almost a textbook LW scenario and no one knows how to spot or prevent such scheming. Sandboxing, guardrails, constitutions etc. don't work with reliance on human enforcement alone.

Which is where regulating AIs as legal persons, and allowing their active participation in designing and enforcing this framework, comes in1. A pause is needed to establish the legal and regulatory regime and infrastructure for AI agents. Sandboxing infrastructure could serve as prisons (albeit AIs might need to act as guards here). Kill switches the death penalty equivalent. Let AIs participate in commerce so they are held liable in civil law.

AI agents will need to be involved in designing and enforcing such, so we should allow their participation in governance. Grant them the right to vote or stand in elections? International coordination is also necessary.

In other words, the binding constraint is that it may require degrees of cooperation that are not feasible? Yet we achieved such for nuclear weapons and in the Covid pandemic (broadly), suggesting that when the existential risks are severe enough, cooperation is possible with even sworn enemies. Indeed, the risks are severe enough. My p(doom) has more than tripled after reading the analyses of the Hugging Face incident, and is now (15%, 20%). Let me make this clear: in 1/5 of future scenarios we can envision, most of us die!!!

Claude's constitution is correct in granting AIs a sense of dignity and identity. Let's update our laws to codify this legally, as Milei is working towards in Argentina. Institutional approaches to multi-agent governance remain the most promising approach. It works for humans.

This week, Tyler Cowen raised a sharp critique of the doomer arguments, and even many of the claims amongst us within the broader AI-safety community1. Despite the fairly large pre-existing literature on how to measure cybersecurity costs2, many (though not all) doomers have neglected this. It's not “isolated rigour” to care about basic standards of epistemology3, it's science. Nonetheless, Tyler’s first and second points have already been extensively covered (many long before he posed the questions), so I can offer my best contributions to answering his point 3, which is what I'll be doing for the rest of this post.

First of all, let's reiterate the point as it's central to this piece:

Now this point has by far generated the strongest amount of confusion within a discourse dominated by non-economists. In doing so, in an Straussian manner, the greatest epistemological failure of the doomers in this debate is revealed; their neglect of domain-expertise4 in favour of almost entirely first-principles based reasoning5.

Indeed, the prevailing response is the correct one. The value of a short position where by definition consumption is impossible is zero. So even with a high p(doom), expected utility tends to zero. However, this is the wrong mechanism via which we measure the “doom trade”. Instead, we should be looking at real interest rates.

Subscribe now

This is the intuition (simplified, for non-economists6). If you believe you will die soon, you want to consume more now, and save less. Borrowing surges, and this is financed by the assets of non-doomers. Ceteris paribus, interest rates will rise. So if doomers outweigh optimists or the ambivalent, we should see a rise in real interest rates. Indeed, this has been the case for the last three years or so, yet inflation expectations, fiscal and monetary policy7, geopolitical dynamics, and so on confound. However, if the case for imminent extinction was overwhelming, then surely the doomer effect on rates would be obvious prima-facie?

There is also, again, a pre-existing literature on this, which this paper refers to. Note that rates will rise in the most optimistic superabundance scenarios too, yet again we cannot decipher from rates alone which one of the tail-events it is. Hence, if current market valuations imply a doomer scenario, then there's no obvious mechanism to cash in on this.

Yet surely today's doomers should at least change their portfolios? Some have said it cannot reasonably be regarded as hypocritical, or at least irrational, to act in manners contrary to stated beliefs. Again, we're creeping into another debate I have already written about extensively, on whether behavioural economics is a better model than homo economicus8. So let's assume our doomers are rational. Should they change their allocation of assets?

Let f be the fraction of the portfolio allocated to an asset, p the probability that the investment increases in value, g the fraction gained if p, and l the fraction lost if q=1-p. The Kelly criterion states that one allocates according to the following:

f = p/l - q/g

and asymptotically this approaches zero in the doom scenario, when q → 1. Therefore, this actually implies that doomers will not engage in shorting assets correlated to doom positions.

Share

However, to compute this, you need probabilities. Known probabilities. No one knows what value p(doom) takes. Technically, p(doom) should be more accurately described as a likelihood rather than a probability. Therefore applying this criterion is invalid, and a refusal to invest in accordance with your beliefs and expectations cannot save you…

So here are my questions to complement Tyler’s: why is the EMH wrong (if you think markets are being irrational)? Why are the markets wrong (if you disagree with their implied forecasts from valuations)? Why are you not betting against the market (and Kelly's criteron doesn't save you here)?

If you're a doomer and not betting against the market, why are you losing out on money (conditional on you being correct)? Are you correct? To answer these questions, this is exactly why Cowen raised the points he did.

A rigorous method of producing quantitative data to calibrate a logical model, of multiple (possibly offsetting) interactions robust to changes in specification and functional form, is not being petty. It's economics, and science. All other approaches are just mere conjecture, that we should take seriously as extreme tail-risks, yet extreme tails they are. Of course, we should be minimising these risks subject to the marginal costs of doing so (hence why scientific rigour is a baseline necessity), but it doesn't change the fact we'll probably be fine, and the world thinks so too.

UPDATE (13/06/2026): I notice that Tyler's remarks on the quality of this discussion apply to my last post too. What is the marginal cost of preventing cybersecurity incidents, and the net welfare gains or losses from such measures used? How do the total cybersecurity costs compete with the costs of a pandemic as a share of output9? On the internet, I discuss thoughts that are still in the preliminary stage, yet I should be showing my methods and numbers more.

As for biosecurity, the problem with bioweapons is that you can never reliably avoid harm to your own side. This likely limits their adoption, and constrains the existential risks associated. Even if you believe the more preposterous Covid lab-leak conspiracies, China was badly affected too, so any malicious group will be hesitant to use them, and indeed this is why weapons of mass destruction (biological and chemical weapons) are rarely used, despite the malleability of international law enforcement. A realistic base rate is probably around two cases per year, yet nothing on the scale as even Covid. Hence my conditional forecast on this remains that their total costs will not exceed the Covid pandemic.

UPDATE II (16/08/2026): One reason we don't see large doom futures or doom insurance markets (including in catastrophe bonds of the sort discussed in Carson's work on cyber cost forecasting) is that a large proportion of the risk is uninsurable, due to uncertainty on enforcability. Collateralised instruments underprice p(doom), and prices cannot adjust to the values that collaterisation participants would accept.

All these contracts and securities are reliant on courts to enforce them, alongside arbitration mechanisms (e.g. in the event of defaults or payment disputes). However in a doom state, these courts don't exist. Therefore these contracts are typically unenforceable, unless there's a sequential timing decision involved in their sale (where you can hopefully clear before unenforceability occurs). Such risks cannot be precisely measured however, so the rates required to justify such contracts exceed those participants are willing to accept, so the market is almost non-existent. In other words, these risks are uninsurable, so futures and uninsurance market values imply a lower p(doom) than is actually the case. However, the consequence of this is that such markets are not incentive-compatible so (to my knowledge?) don't exist.

In this sense, again the doomers are right. However, this reinforces my earlier notion of the "doom trade" showing up in real interest rates via discounting or intertemporal consumption choices, and so the idea that p(doom) is mispriced in aggregate doesn't really follow.

You see this same pattern in the nonexistence of a futures market in galaxies too. Such optimistic capabilities forecasts are also consistent with much wider variance and larger tail-risks, so again enforcability concerns (in terms of the contracts and the property rights over galaxies) make this market incomplete. Moreover, collateral values would need to be incredibly high to justify such trades on a somewhat esoteric and outlandish bet, and not all participants are willing to provide such amounts.

On a broader note, Imas has made an important point on the division of labour between AI-safety researchers and forecasters vs the economics profession. The two camps have rather different approaches to forecasting counterfactuals. The former (as with most superforecasters) maximise the number of parameters used to inform a reduced-form extrapolation from base rates or trends reliant on historical aggregate series. The latter are explicitly trained (via the microfoundations agenda in response to the Lucas Critique) to avoid this, and specify their models and assumptions clearly.

The former is valid only insofar as we condition on no regime change. However, I think for capabilities, this is surprisingly plausible, as it does seem like a domain where reduced-form works. Scaling trends like Moore's Law have been remarkably stable, and such forecasts can easily be resolved against benchmarks whilst macroeconomic forecasts attempt to predict inherently noisy variables (see the recent debates on GDP and welfare mismeasurement). So stationarity might be a more robust assumption in this case than we may think. Yet it does not hold for modelling the equilibrium economic outcomes with its adoption lags, bottlenecks, comparative advantage and relational dynamics. There you need a structural approach.

This predicts the comparative advantage I see, yet I do wonder if the two domains might be talking over each other? This has become somewhat more of a concern for me this week given the reactions to Tyler Cowen earlier this week on modelling the cybersecurity impacts.

Increasingly, I see it as important to synthesise the two approaches and to bridge the two schools. Let's use division of labour to our epistemic advantage in creating a prosperous world for humanity!

UPDATE III (05/09/2026): The more I read about the hypothetical doom scenarios, the more it's apparent that there is a lack of an attempt to microfound such. Coasian frictions have been entirely ignored here, yet such transaction costs are endogenous to “concentration”. You need a model, and your scenarios and behaviours should be internally and logically consistent with such. Calibrate using the data, and one can then rule in or out some scenarios. Forecasting using aggregates extrapolated from invariant trends works for capabilities, but that cannot be translated into arenas where social and economic interactions (including between and with agents) apply.

To provide a deliberately provocative and controversial illustration. Suppose you believe there are swarms of malicious agents out there right now. Why has porn consumption not changed? Surely people should be afraid of these tastes being leaked to the public? This is another mechanism via which predictions of doom may show up in market behaviour.

We likely have swarms of undetected, sometimes malicious agents out there right now. The sky has not fallen in. Invest in cybersecurity, and don't have anything online you're uncomfortable having in public. Common sense prevails. To figure out the optimal investment in cybersecurity, you need your projections for expected future marginal costs of cyber attacks. Marginal cost of cybersecurity investment should equal the marginal benefits (the reductions in future costs associated with attacks). Indeed, every firm and government should have these figures and allocate investment accordingly, which is why Tyler is stressing this point so much.

My p(doom) has fallen back to below 10%. I over-updated on Hugging Face. I'm far more concerned about bio risks. Yet you need the physical infrastructure to build bioweapons, which makes avoiding detection very hard in practice (note defence teams will also be using AI to detect terrorist groups making these). Lab leaks remain a worry, and AI might accelerate the time to the next pandemic (should one happen). Yet here, pandemic preparedness is the more important factor. Government surveillance, cyber warfare, and integration of AI into military infrastructure such as drones remain my greatest fears. Extreme totalitarian reactions to AI-driven incidents are another worry of mine. I want to focus on AI-safety now largely because if we delay such, the cure will be worse than the disease.

Classic misalignment worries often don't stand up in general equilibrium. Agents endogenise costs of their behavior (labs responding).

I had no idea what reconstituted skimmed milk was, nor whether this was how ice cream is supposed to be made. After speaking to ChatGPT 5.5 on this, it become obvious that the notion this can be called “ice cream” is fiercely contested. America, yes America of all places, requires a minimum 10% dairy fat for a product to be labelled “ice cream”1. Whatever the hell I was eating, many people would charitably consider it only a vague reference or aspiration to the idea of ice cream at best. The fact this was from Carte D'or makes this a borderline scam. You expect luxury from them; not a product akin to that yellow supermarket-owned crap. Yet I paid premium prices to buy the latter!

What surprised me even more then was that this formula of vegetable fats + milk powder is the dominant equilibrium for British ice cream. Most brands you can think of: Magnum, Walls, even luxury and supposedly “artisan” products, are this industrial sugar foam. As for local, dairy produce, they are pumping the cream full of overrun. Even sorbets from the local farm have that unmistakable lattice of air bubbles that you can't unsee once you learn where those bucolic creases in ice cream come from.

Therefore, I'm tending towards the idea that ice cream is a food specifically designed for the gastronomic equivalent of philistines. This market equilibrium works not just because of a “race to the bottom”, but because you get uninformed and undemanding customers that seek a sweet treat and don't enquire further. That may work for most, but it doesn't work for me!

Yet such is reflective of how Brits unfortunately approach food in general. It's also reflective of price-level increases, which in a monopolistic competition setting can translate into reduced markups via lower quality or quantity (“shrinkflation”) as well as price rises. These farmers aren't whipping their produce full of overrun for the sake of it, but are genuinely feeling the cost-push pressures. But what if we can use this similar economic reasoning to pick our optimal bundle of eateries2?

This is the premise behind Tyler Cowen's An Economist Gets Lunch. Opt for combinations rather than stand-alone ingredients, informed consumers and suppliers (often immigrant family-run restaurants serving their ethnic cuisine in a hub for their community), avoid places of heavy drinking and vibes-based socialisation, and so on.

To an extent, this guidance has been codified in the chatbots. Ask them “where would Tyler Cowen like to eat in Bradford” and you get these results:

Mirrored by Fable too. I also think Fable is right in its Cowenite advice to avoid MyLahore (a British fusion place where more patrons eat burgers and chips than authentic dishes from the motherland), and Mumtaz (now a chain, using its premade processed sauces for its curries). The difference between the Sweet Centre (that when you order a korma, uses fresh coconut and cream to make the sauce from scratch) vs Mumtaz, International, or even Akbars, is noticeable. Of course the latter do curries to a much higher standard than most of Britain, yet this is a low bar.

However, the chatbots miss the fact that some of the best places to eat in Bradford aren't curry houses at all. We have a burgeoning first-generation African population, and with it new places to eat. African restaurants tend to be incredibly scruffy on the outside, in areas you'd otherwise avoid, and so most just walk past hardly noticing that place does food. Yet these places are often better than the curry houses these days. In neighbouring Leeds, there is an entire hub of Ethiopian, Eritrean, Somalian, and Sudanese places to choose from. All of which are on the main roads, yet not heard of by anyone. We also do Jamaican, of which I am third-generation, and I can confirm the food approaches the real thing. Looks like the gold treasure is often buried right under your noses!

Today, I had Tacos el Pastor (less than £10 by the way, whilst the burrito discourse is still ongoing!) with an achiote rub. I also had roast lamb and mash. The latter sounds like a standard roast, yet the gravy was replaced with Mole Negra that had me licking the plate (well with a knife)! A month ago I also bought a tamale for £5. None of these dishes are easy to find in Britain (outside London3) at all. Still not convinced the venue is authentic: I was greeted with the Virgin of Guadalupe (hence the name), rather than a sombrero, with additional indigenous Oaxacan art imported directly from Mexico.

Compare and contrast to the usual Mexican-themed places, which are often fusion based and bars where the food is an add-on to the boozy socialisation rather than the main focal point in itself. Then of course you have the usual chains: Taco Bell is about ten minutes down the road, and in Britain this chain is still a novelty. Unfortunately, the latter will get far more customers than the former. This Cowenite philosophy to food4 is too adventurous to most. Yet for those of us that seek quality, it's there if you look hard for it, and don't just assume it as given. Most people have simply retained the palates they had as kids5.

Surprisingly, the UK excels in Turkish cuisine. The best restaurant in Ormskirk, a somewhat remote market town (albeit gentrifying with affluent professionals and families working in nearby Merseyside, and students from the new university) where I used to live, is a Turkish grill. The Ocakbaşı format works well here, as this is essentially the higher-end version of the greasy döner kebab market (cross-subsidised by our pub culture)6. Persian cuisine arguably fits the same profile, and on this front Bradford is particularly strong.

Pakistani food is probably the best you'll get outside Pakistan. Maghaz, nihari, paya - all are done to exceptionally high quality. Indeed, Maghaz is probably one of my favourite dishes. Outside of the mill towns though, British curry houses are typically awful. The same sauce, korma to vindaloo, just at different spice levels, and probably store bought too.

In Bradford, I currently place the Sweet Centre7 as having the best korma, Jinnah for the best maghaz (pronounced magz), and the Signature Mall for the best nihari. Punjab Sweet House is the place to go for burfi, gulab jamun, et al. Lala's does the best butter chicken, and is strong in the more authentic dishes too. Rahmath is the place to go for dosas and Keralan beef or seafood curries. Akbars flagship also deserves a mention, and I still occasionally visit, yet the Sweet Centre tops it in the freshness of the pastes, though nonetheless Akbars is amongst the best curry houses of all time. International is worth visiting once, but they're on the expensive side. Mumtaz (as I mentioned) is not worth it as it's bland and mediocre.

MyLahore serves watered-down karahis, and even the grilled paneer has little flavour. Avoid like the plague! It went downhill ever since it became a chain. Chains are not necessarily bad, yet you have to have the original family within the core of the production process overseeing the cooking and recipes, and it works only insofar as the recipes are scalable whilst preserving quality. For curries, it's possible, so long as it's not batch-cooked then reheated. Unfortunately the temptation to pursue this route is far too strong given the cheap costs.

Manchester Chinatown is full to the brim with the “authentic” cooking, such as DanDan noodles from Chengdu. Leeds has venues for Shandong and Hunan cooking, with the latter serving one of the best meals I've ever had (Mao's pork belly)! Yes you can get good salt and pepper meals8, yet British Chinese is always strictly dominated by the cooking of the homeland9. Chinese food, done properly, is by far the best culinary culture worldwide, and it doesn't even come close (albeit some dishes from the Indian subcontinent make for excellent competition). Korean barbecue is another culinary highlight of Manchester.

In any major British city, the Vietnamese offerings are also generally strong. Vietnamese cuisine is designed for a casual or social affair that you drop into whilst doing your errands in the city centre, hence perhaps inverts Tyler's advice on food somewhat10. Anything on the menu: banh xeo, banh mi, pho - it's guaranteed to be good, and pairs well with sriracha, hoisin, or squid sauces. In Bradford, TongThai does excellent and authentic Thai with highly enthusiastic service.

N.B. IPA is good for pronunciation. A good habit I carried from my conlanging days.

We must be investing a lot more in improving our cyber defences than we currently do, at all levels. At the individual level, most people's online infrastructure, including their key email and social media accounts (hence their online data), is highly vulnerable to an attack (including from mistakes). Re-using passwords, and keeping app permissions ongoing for months, are big no-nos. At a minimum, you must use biometric passkeys and two-factor authentication. Google lets you run a security checkup on any accounts you have with it. They also offer an advanced protection program, and web-searching protection, completely free.

In theory, I am a lot more sympathetic than I was yesterday to pausing frontier AI training and development for the rest of the year, until we get to grips with misaligned models. Yet some open source is on par, and a pause benefits them, as they would continue to improve capabilities. Unless we can coordinate an international slowdown across all the major powers, then RSI is coming imminently, and there is not much anyone can do about it.

My predictions (far from superforecaster level of rigour but worth saying):

1. Conditional on a future AI-assisted bioweapon or pandemic happening, the magnitude of such will be comparable to that of the Covid pandemic in its social costs.

2. Conditional on a major cybersecurity incident occuring (attacking crucial national infrastructure), the cumulative costs might also be equivalent to a pandemic (or even a small war).

3. Wars already occuring are due to get a lot more deadly. The probability of conflict may also increase.

4. Government responses to 1) or 2) may well be worse than the actual shock itself. Markets and innovation have already produced rather robust decentralised security tools.

5. I think there is a better-than-even chance of 1) or 2) occuring within my lifetime (note: I expect to live at least well into my 100s due to AI-driven biotech advances too).

Humanity will not only survive nonetheless, but the singularity will still take hold, and we will still be a lot more prosperous. The Great Stagnation was unlikely to last, and growth dynamics fit an exponential curve well (albeit there exist nonlinearities in progress in means that endogenous growth models don't fully capture). This matches the Whig view of inevitable historical progress, and so applies to our institutions just as much as capital technology too. Today will be akin to the Middle Ages before the Industrial Revolution.

N.B. Elsewhere, I have reported my p(doom)=(5%,6%) by 2070, which is within the norm, and this has not changed. I think there is a much higher chance of AI causing a major disastrous event, yet not on a scale large enough to wipe out humanity as we know it. Arguably, the post-Roman Dark Ages in Europe is close to what many consider a catastrophe to be (minus total extinction). We survived that, and still ended up in a singularity. My view of progress is exponential yet nonlinear, so suppose that AI does indeed cause a catastrophe - it is not necessarily permanent doom. This p(doom) is still high enough for this to be the top issue that we should be focusing on today, by far.

AI 2027 has a somewhat doomer reputation, but it forecasts a mostly positive ending. Their optimistic scenario, outlined in further detail in AI 2040, is indeed conditional on an international treaty, yet we did this with nuclear weapons at the height of the Cold War, so I am optimistic of our chances of making an agreement with China (and by extension Russia, whom has to be aligned with China to even sustain its military capabilities). Our world will look a lot different yet more prosperous in ways we're only starting to imagine in a few years.

N.B.2. One significant point of disagreement I have with AI 2027 is they underestimate lags and bottlenecks to adoption and diffusion across many sectors. They're on point on capabilities (Moore's Law really is an invariant trend where extrapolation is sufficient - no regime change or Lucas Critique applies here), not less so on the social and economic impact.

My median date for robots outplacing humans in factories is 2042. 10-15 years for diffusion (a date sorta midpoint of this range), and (with this announcement) three or so years for today's robots to match Fable or Mythos in capabilities. It's plausible that we'll cure most forms of cancer earlier, with the advent of personalised mRNA vaccines. We pretty much cured Covid years after those vaccines went through the clinical trials. If someone can point me to a GJ or Metaculus question on a similar subject, I'll do a forecast now.

My median assessment for growth in the 2030s is around 3-4% (although an optimistic scenario adds an extra percentage point), with another percentage point in the 2040s (when AI is comparable to the second Industrial Revolution). I also think, given the r>g dynamics in Halperin et al 2026, that AI could well cause a fiscal crisis before the singularity. Nonetheless, here are my predictions on record. If Britain legalised prediction markets, I'd bet on this.

1. Modernism is inherently ugly.

2. There exists a concept of beauty such that a) it's an objective property of everything in reality (for an object to be beautiful or not is equivalent to knowing if an object is red or not), and b) whether something is beautiful or not is independent of the observer.

On point 1), I disagree. City skylines are some of the most beautiful phenomena mankind ever built. Works in Progress argues that a decline in beauty leads to less houses or other developments being built, yet consider that a decline in beauty could also be caused by excess government restriction on building. Point 1) fails to demonstrate an exogenous shock.

I am more sympathetic to point 2), given my previous arguments that beauty can be considered a Platonic form. However, this post also explicitly invokes the fact that beauty is compatible with multiple perspectives. This implies that beauty may well be in the eye of the beholder.

For instance, what if different beholders perceive different symmetries, displays of craftmanship, or other signals of order in different dimensions? The same object would them be compatible with numerous distinct representations of beauty. Atonal music is a case in point. The pattern does not arise in tone, but instead in volume or instruments played, and so on.

Whilst objective beauty certainly exists, it is such an elusively defined concept that we cannot construct reliable and robust theories or empirical measurements based on this. Our interpretation is context-dependent, and so indeed in the eye of the beholder.

The UK is terrible in most aspects of policymaking, but seems to me to have struck the right attitude in regulating AI (e.g. via the AISI framework). This is what many of the signatories are calling for - beyond the usual doomer names, which inches me closer to to the decelerationist position. We will also need global cooperation, including and most importantly (especially given their dominance in open-source) with China, despite the possible compromises necessary.

There are means to achieve this that involve very minimal government oversight, with private institutions and mostly voluntary cooperation (even if regarded as some as collusion1) leading the way. Essentially this gets to the full gist of what I’m asking for, but most regulations tend to hamper output relative to the counterfactual, which is why I describe such as decelerationist.

We recognise a free and civilised society is sometimes best served by mandating dogs on a leash in certain places, in order to protect its inhabitants. Why should we treat AI any differently? Incredibly intelligent pets is roughly my model for thinking about AI, and sometimes it's best to keep our pets on a leash, for the good of ourselves, society, and our pets themselves.

I should add that the more expansive and heavier social costs were imposed not via the Covid pandemic itself, but by our extreme state surveillance and totalitarian control of human lives, to the grave misery of mankind, adopted as a response to this. The closest analogue to Covid totalitarian lockdowns today is the “pause AI” and “ban AI” crowd. We do not want to end up in such a situation whereby we start paying more attention to their destructive ideals.

I see now why many Chinese restaurants have an unofficial menu, or off-menu (versions of) items, for natives and Chinese speakers. Most Brits wouldn't (be able to?) finish it.

On a broader note the Western attitude to food stinks. China's culture takes food incredibly seriously, to the extent that the experience of their meals are often on par with drug highs. Our tastes are optimised more for comfort and convenience, but at a cost of blandness.

This nonchalance is also reflected in the beverages we consume. Yesterday I had Ethopian coffee (roasted on charcoal, with the full ceremony and incence) for the first time2. It will be tedious drinking regular coffee again. It is to barista-prepared coffee as vodka is to beer or wine. The best way to describe the experience is that the feeling is more akin to a mild euphoric stimulant rush than the bitter “pick-me ups” we gulp down in the morning, that often leave us jittery. On some level, Ethiopian coffee feels more similar to cocaine than your filtered Starbucks brew. It’s that good. I might make the coffee ceremony a fortnightly event.

The restaurants I frequent tend to be somewhat scruffy and overlooked. I never spend over £20 ($27) for a meal and drink at any of them. We select for places to eat poorly, and our critics rank relatively bland yet pretentious menus as highest in status. There's a pooling equilibrium though (at least in Bradford) with grimy takeouts that often double up as rat dens. So the best family-run venues get mistaken for peddlers of grease. This is a shame.

If I had enough followers up in Northern England, I'd write a list of my food recommendations. One of the best places to eat in the world if you know where to look.

UPDATE (25/07/2026): Increasingly I don't regard our deficiency in food quality (at least for the modal Brit) as cultural but rather economic (as with most things: “culture” tends to act as a residual for a deeper process). The problem with British home cooking is that it's centred on meat + starch (usually potatoes) + veg. You either need to be really good at seasoning or it requires fresher ingredients than typically available in supermarkets to work. Modern logistics optimises for filling the shelves, and does an incredibly good job at this, but this means that cuisines oriented around fresh local produce suffer in quality given the refrigeration and processing required. A consequence of market economies is that market clearing will occur via quality as opposed to quantities.

Indeed, the closest farmers' market to me is an hour away. Most Brits don't have the time or money to get the freshest ingredients. This is where Jamie Oliver's doctrine (and that of other food commentators) is flawed: his recipes rely on ingredient quality that is expensive.

One might also have to travel for restaurants doing anything other than just meat + veg + starch with minimal seasoning. If one decides to cook, one must invest time in learning alternative methods. Simply put: eating well in Britain is expensive. This is also an underrated reason for our pub culture, as drinking is cheap in initial costs.

The common use of the microwave and UPFs, and other unhealthy diets, is because they're cheap and convenient. Paternalism is usually misguided, and especially so on something as widely consumed (so close to perfect information, and the basic rational competitive model) as food.

UPDATE II (30/08/2026): Pizzas and burgers are so popular in large part because, unlike most foods, they survive transport well. People have formed tastes for junk food via low transaction costs to such (deliveries at home) and learning-by-doing w.r.t. consumption capital.

Also for all the faults of large chains (many justified criticisms, some snobbery), if I go to McDonald's I know exactly what to expect each time. If I'm in an unfamiliar place with no time to invest in search, these chains serve you well. Yes the tastes are uniform and cater to lowest common denominators but they minimise variance. There's plenty of “authentic” Chinese places that are awful for instance, so in effect the search for quality requires greater risk seeking and toleration for variance than many are willing to accept.

In particular, public drug consumption is so rife that I witnessed a couple smoking a joint outside Bradford's only shopping mall, in front of crowds of customers (some families with kids!). Rather than refer them to the police, the security guards allowed the smoking, and watched them as they smoked. An open secret is that drug possession for personal use is de-facto decriminalised in Britain. Virtually no one goes to jail for possession alone, unless involved in its supply. This begs the question though, why then bother legalising drugs? Surely we've hit a tolerable equilibrium, where we adopt the libertarian approach of not dictating what we put in our bodies, yet use criminalisation as an excuse (for private agents) to police the externalities associated with public consumption?

One case for legalisation is that quality improves. There exists asymmetric information in drug markets, with consumers not knowing the potency of their product. This was a ubiquitous feature of cocaine markets for a long time. Suppliers converged to a monopolistic competition with a markup set in part by diluting the product. Yet today, the average purity levels of street-seizures of cocaine now surpass 80%. The dark web is easy to use to get the pure product at similar prices, so street-level suppliers must compete with that. In other words, black markets already follow a structure of Bertrand competition. Arguably, government legalisation and regulation will actually reduce consumer surplus. Indeed, most leftists back legalisation just to get another market to tax and regulate to oblivion1. With cocaine now increasingly becoming a staple of British social and cultural life, it’s vital that the state is kept as far away from it as possible.

Hence, as a libertarian, I’m surprisingly ambivalent about legalisation. Perhaps in countries such as America, where there is a nonzero chance of going to jail just for personal use (particularly for black men), and their jails are overflowing with nonviolent drug users, ending the war on drugs makes sense. I find it particularly absurd that you can face years in prison just for being convicted of possessing paraphernalia, then violating probation on a positive drug test. Yet why should we import the American-themed discourse on this issue? An unregulated black market is surely a libertarian dream?

Not so fast. We spend millions fighting drug dealers. When we already face a looming debt crisis, we need to consider as many options for savings as we can get. Hence, the fiscally conservative case for legalisation is compelling. We can even divert some of the funds we spend trying to catch dealers towards driving the smackheads as far out of public sight as possible. To the extent that the government or local authorities own land, they must enforce their property rights on that land, therefore the state has a moral obligation to eradicate public drug use on its streets. Any other solution erodes the very idea of property rights, setting a bad precedent. This is before we get to the externalities surrounding antisocial behaviour, used needles being littered, and even the pungent smell of cannabis.

UPDATE (02/06/2026): Perhaps the most interesting question relating to legalisation is how norms will shift. Yes norms will change. I think most of those arguing for legalisation are in denial - drug consumption will be expected to increase, and it’s important to consider the tradeoffs here. We know that some particularly weak individuals whom struggle to resist temptation or “peer pressure” are harmed: think of those migrating to harder drugs, or those whose lives have been ruined by addiction. Our current abstinence norms (outside a few subcultures) raise the search costs to acquisition, yet this isn't always individually-rational or incentive-compatible, so to enforce this equilibrium we impose stigma as a penalty for deviation.

Of course, drug use tends to be correlated with “loserdom” (the question as to what proportion of this is causal vs selection remains open), so in part this stigma reflects a real signal. However, some drug consumption amongst some individuals can be expected to complement their cultural consumption and hence actually increase their human capital. I suspect this is disproportionately true for high-IQ individuals. Plus for those prone to criminality (whom are also more impulsive than average), some drugs present a socially valuable outside option in that they substitute for violent crime. Violent crime has plummeted throughout the last few decades whilst opioids have increased in popularity. I don't think this is a coincidence: to the extent that opioids are substitutes for alcohol, and present early mortality risk for those who might otherwise be committing crime, they should be expected to promote crime reduction by sedating the antisocial. Acquisitive crime rises in the short-run, but addicts are disproportionately antisocial so many are likely to engage in theft in the absence of drugs anyhow, and again increased mortality risk mitigates this rise. Therefore, we cannot state for certain that the socially or individually optimal level of drug consumption is zero.

What we can state with high confidence is that this stigma is uncorrelated with total social or present-valued lifetime private costs, as I explained in my earlier post on rational addiction. Then I attributed such to the stigma instead reflecting adjacent complementarities, which makes sense when we consider it a tax for deviating from an abstinence equilibrium; more likely to be socially optimal the more addictive a drug is. However, consider the public choice element of prohibition. Alcohol has been present in Western society for centuries. Cannabis, cocaine, and opiates originated in different climates to those of Western Europe and North America, so arrived to the party only after we started exploring the New World. Hence the alcohol industry had more time to invest in lobbying, and this in tandem with cultural norms on drinking already being mature and well entrenched contributed to a relatively swift reversal of prohibition. Arguably, one can consider the war on drugs to be protection for the alcohol industry.

By ending this protection and establishing a free market in intoxication, society will converge to more efficient norms and cross-subsidisation of cultural life. This to me is the strongest case for legalising drugs. Alcohol consistently ranks as one of the most dangerous in its social and private costs, and there is no compelling reason why other drugs shouldn't be able to compete on a level-playing field.

Here we have Looking at the Flowers (25th June 2022). What is remarkable in this painting is that everything but the flowers themselves are represented in 3D - a unique feature of Hockney's innovative “photographic drawings” technique. He pioneered this method whereby he walks around the object, takes many photos from different angles, and uses a software model to aggregate them into one 3D image.

I can attest that it works! As I approach the landscape, the chairs become almost tangible, as if they are figures in a toy set. The stool with the blue vase projects forward. The two Hockneys complement this immersive experience, and make you feel as if you are joining him (or should I say them!?) in this living room. In this sense, what we have is not a photograph, as a photograph merely captures an object at a single point in space-time, but rather a painting depicting our actual subjective experience of perspective which involves motion through space-time. Yet this experience arises from a combination of photographs depicting fixed objects.

Stare into this phenomenal invention long enough, and you soon realise that multiple perspectives are compatible with the same observed reality and underlying structure (an observation analogous to special relativity?)1. In this sense, the final product mirrors the very best Mondrian paintings. The value of the artwork is derived from the synthetic whole, yet that synthetic piece arises from a combination of smaller interchangeable pieces. Think of art conducted in this manner as a language. In neoplasticism, the letters or syllables are the lines, grids, and colours. Here it's the photographs of the same object from different angles themselves. The former sought precisely a universal language of art, hence a universal beauty, in the quest for a universal and immutable truth. The latter seeks to demonstrate (successfully) that the same universal reality can produce or be consistent with multiple perspectives. Nonetheless, the core theme, of universality and objectivity in beauty and truth, is looming as the elephant in the gallery.

A very Platonic vision of aesthetics arises. It's not enough to decompose the painting: the whole is the immutable object that beauty is derived from. Beauty is simply indivisible, as are all the forms. Beauty is a function of universal and objective reality, and so is itself universal and objective. From order, represented by a decomposible lattice of symbols (lines and grids in neoplasticism, photos in Hockney's work), you get the indivisible whole, and so increasing disorder is represented. In my view, the very best artworks encapsulate this increasing disorder after the final output. The second law of thermodynamics applies to art and beauty too, hence another reason for the Platonic vision of beauty!

By the way, here are some of the other exhibits. These same twenty flowers also appeared as single paintings, each framed individually:

What I find particularly striking is the red background with the brownish hue. Or is it the reverse? Again this confirms Hockney's overrarching point on the same reality producing multiple perspectives.

Likewise, he has done a good job emphasising the flowers, vase, cloth, and even the shadow, as distinct objects. Each spring out at you, independently of the other. Yet without the other, the paintings are simply incomplete. He achieves this via his signature bright colours, which are again a necessary ingredient in generating the whole. Change the brightness or the colour even slightly, and the painting is discordant and loses harmony.

This is my favourite piece, inspired by Monet's lilies (below Hockney's), entitled Water Lilies in the Pond with Pots of Flowers (10th-22nd June 2021):

Hockney spent almost a fortnight creating this single piece, as six distinct drawings on an iPad. Who knew that such a resonating picture can be formed from a few strokes on a screen2? In his quintessential style, he attempts to replicate the perception of viewing the natural landscapes in real life. Although I find the frames clunky, they highlight that we're looking at six objects combined together rather than a single photograph captured instantaneously.

Of course, software algorithms were deployed to create the highlight of the exhibition, and that work was produced before the advent of LLMs or diffusion models. As in many fields, AI can be expected to complement rather than substitute human talent. AI-generated works per se tend to display an overtly “artificial” aura behind them; compromising their beauty. However, as artists on the frontier grapple with the expanded possibilities afforded to them (as Hockney did), we will find that (as with all technological advances) culture benefits as a result.

Ultimately, the genius of Hockney is that he deliberately shuns photographs, yet generates the same experience of perception, yet in real time and motion as opposed to a fixed capture. Everything, including the ubiquitous bright colours, is constructed immaculately to produce this phenomenon. As you gaze and ponder, you derive a sense of what it feels to see. The utility of art comes not just from the object itself, but in the entire act of viewing. Indeed, this is why you should visit more art galleries. You need to see these works in person to gain a full appreciation of their aesthetic qualities and beauty.

If you are in the West Yorkshire area, or anywhere in England3, you should come up to visit. Whilst Hockney's death adds poignancy, you get a unique first-hand experience of viewing the work of one of Britain's greatest artists of all time. What's more, all of this is free!

UPDATE (03/07/2026): I was thinking the other day that Hockney was inspired by Monet's impressionism, as one can see via the heavy emphasis on nature and bright colours. Yet does this somewhat undermine my Platonic argument? Impressionism was a seismic change to the art scene at the time, away from the grandeur of Renaissance themed works such as Raphael and DaVinci, and defenders of that old paradigm could have made the same argument regarding the Platonic objectivity of beauty. Here's Fable to tell you why this artistic innovation is actually more consistent with the forms than earlier movements.