Every once in a while I think about Robert Freitas' 1984 essay Xenopsychology, in particular his Sentience Quotient (SQ) idea: 

It is possible to devise a sliding scale of cosmic sentience universally applicable to any intelligent entity in the cosmos, based on a "figure of merit" which I call the Sentience Quotient. The essential characteristic of all intelligent systems is that they process information using a processor or "brain" made of matter-energy. Generally the more information a brain can process in a shorter length of time, the more intelligent it can be. (Information rate is measured in bits/second, where one bit is the amount of information needed to choose correctly between two equally likely answers to a simple yes/no question.) Also, the lower the brain's mass the less it will be influenced by fundamental limits such as speed of light restrictions on internal propagation, heat dissipation, and the Square-Cube Law.

The most efficient brain will have the highest information-processing rate I, and the lowest mass M, hence the highest ratio I/M. Since very large exponents are involved, for the convenience we define the Sentience Quotient or SQ as the logarithm of I/M, that is, its order of magnitude. Of course, SQ delimits maximum potential intellect–a poorly programmed or poorly designed (or very small) high-SQ brain could still be very stupid. But all else remaining equal larger-SQ entities should be higher-quality thinkers.

The lower end of our cosmic scale is easy to pin down. The very dumbest brain we can imagine would have one neuron with the mass of the universe (10⁵² kg) and require a time equal to the age of the universe (10¹⁸ seconds) to process just one bit, giving a minimum SQ of -70.

Whenever I see the "The difference between genius and stupidity is that genius has its limits" quote (usually apocryphally attributed to Einstein) I imagine Freitas retorting "no, so does stupidity, the limit is SQ -70".

What is the smartest possible brain? Dr. H. Bremermann at the University of California at Berkeley claims there is a fundamental limit to intelligence imposed by the laws of quantum mechanics. The argument is simple but subtle. All information, to he acted upon, must be represented physically and be carried by matter-energy "markers." According to Heisenberg's Uncertainty Principle in quantum mechanics, the lower limit for the accuracy with which energy can be measured–the minimum measurable energy level for a marker carrying one bit–is given by Planck's constant h divided by T, the duration of the measurement. If one energy level is used to represent one bit, then the maximum bit rate of a brain is equal to the total energy available E ( = mc²) for representing information, divided by the minimum measurable energy per bit (h/T) divided by the minimum time required for readout (T). or mc²/h = 10⁵⁰ bits/sec/kg. Hence the smartest possible brain has an SQ of +50.

For a while I wondered what such a superbrain would be like, and then I found Seth Lloyd's paper quantitatively bounding the computational power of a hypothetical "ultimate laptop" of mass 1 kg confined to volume 1L, which derives the same computation limit to within an OOM, concluding that "a typical state of the ultimate laptop’s memory looks like a plasma at a billion degrees Kelvin: the laptop’s memory looks like a thermonuclear explosion or a little piece of the Big Bang!"; its energy throughput would need to be a preposterous 4.04 x 10²⁶ watts, slightly more than the entire sun's output of 3.846 × 10²⁶ watts(!!). 

Where do people fit in? A human neuron has an average mass of about 10^-10 kg and one neuron can process 1000-3000 bits/sec. earning us an SQ rating of +13. 

That 50 - 13 = 37 OOMs of headroom estimate between humans and Freitas' "mini-Big Bang superbrains" has stuck in my mind ever since. The "practical" headroom is definitely much lower, although how much I don't know. 

What is most interesting here is not the obvious fact that there's a great deal of room for improvement (there is!), but rather that all "neuronal sentience" SQs, from insects to mammals, cluster within several points of the human value. From the cosmic point of view, rotifers, honeybees, and humans all have brainpower with roughly equivalent efficiencies. Note that we are still way ahead of the computers, with an Apple II at SQ +5 and even the mighty Cray I only about +9.

As an update on that 40-year old estimate, ChatGPT-5 medium estimates that "the highest value you can plausibly assign to a real, shipping computer “brain” today belongs to Cerebras’s wafer-scale processor (WSE-3) used in the CS-3 system. Using public performance and physical data, its chip-only SQ comes out around +19½. If you insist on a whole-system number (including packaging/cooling/rack), the CS-3-as-appliance is roughly +16; the most compute-dense Nvidia rack (GB200 NVL72) is about +15.9; and the #1 TOP500 supercomputer (El Capitan) is about +14.2." I have a feeling smartphones might beat this, not sure why GPT-5 considered and dismissed assessing them in its reasoning trace.

Another kind of sentience, which we may call "hormonal sentience," is exhibited by plants. Time-lapse photography shows the vicious struggles among vines in the tropical rain forests, and vegetative phototaxis (turning toward light) is a well-known phenomenon. All these behaviors are mediated, it is believed, by biochemical plant hormones transmitted through the vascular system. As in the animal kingdom, most of the geniuses are hunters–the carnivorous plants. The Venus flytrap, during a 1- to 20-second sensitivity interval, counts two stimuli before snapping shut on its insect prey, a processing peak of 1 bit/sec. Mass is 10-100 grams, so flytrap SQ is about +1. Plants generally take hours to respond to stimuli, though, so vegetative SQs tend to cluster around -2.

How about intelligences greater than human? Astronomer Robert Jastrow and others have speculated that silicon-based computer brains may represent the next and ultimate stage in our evolution. This is valid, but only in a very limited sense. Superconducting Josephson junction electronic gates weigh 10^-12 kg and can process 10¹¹ bits/sec, so "electronic sentiences" made of these components could have and SQ of +23 – ten orders beyond man. But even such fantastically advanced systems fall short of the maximum of +50. Somewhere in the universe may lurk beings almost incomprehensible to us, who think by manipulating atomic energy levels and are mentally as far beyond our best future computers as those computers will surpass the Venus flytrap.

Just as consciousness is an emergent of neuronal sentience, perhaps some broader mode of thinking–call it communalness–is an emergent of electronic sentience. If this is true, it might help to explain why (noncommunal) human beings have such great difficulty comprehending the intricate workings of the societies, governments, and economies they create, and require the continual and increasing assistance of computers to juggle the thousands of variables needed for successful management and planning. Perhaps future computers with communalness may develop the same intimate awareness of complex organizations as people have consciousness of their own bodies. And how many additional levels of emergent higher awareness might a creature with SQ +50 display?

The possible existence of ultrahuman SQ levels may affect our ability, and the desirability, of communicating with extraterrestrial beings. Sometimes it is rhetorically asked what we could possibly have to say to a dog or to an insect, if such could speak, that would be of interest to both parties? From our perspective of Sentience Quotients, we can see that the problem is actually far, far worse than this, more akin to asking people to discuss Shakespeare with trees or rocks. It may be that there is a minimum SQ "communication gap," an intellectual distance beyond which no two entities can meaningfully converse.

At present, human scientists are attempting to communicate outside our species to primates and cetaceans, and in a limited way to a few other vertebrates. This is inordinately difficult, and yet it represents a gap of at most a few SQ points. The farthest we can reach is our "communication" with vegetation when we plant, water, or fertilize it, but it is evident that messages transmitted across an SQ gap of 10 points or more cannot be very meaningful.

What, then, could an SQ +50 Superbeing possibly have to say to us?

If we replace "SQ +50" (which we know can't work because of Seth Lloyd's analysis above that they'll be mini-Big Bangs so we wouldn't survive their presence) with the more garden-variety ASIs I guess one possible answer is Charlie Stross' Accelerando: "...the narrator is Aineko and Aineko is not a cat. Aineko is an sAI that has figured out that humans are more easily interacted with/manipulated if you look like a toy or a pet than if you look like a Dalek. Aineko is not benevolent..."

Scott strongly encourages using well-crafted concept handles for reasons very similar to what Raemon describes, and thinks Eliezer's writing is really impactful partly because he's good at creating them. And "Offense is about status" doesn't seem to me like it would create the reactions you predicted if people see that you in particular are the author (because of your track record of contributions); I doubt the people who would still round it off to strawman versions would not do so with your boring title anyway, so on the margin seems like a non-issue. 

For future readers of this post and other writings on heroic responsibility who feel a bit amiss, Miranda Dixon-Luinenburg's The Importance of Sidekicks may be for you (as it was for me). Think Samwise to Frodo or Robin to Batman, or if you know investing Charlie Munger to Warren Buffett, or if you like team sports Scottie Pippen to Michael Jordan. There's probably a gradient from "assistant" to "second-in-command"; I lean more towards the latter. Miranda:

I suspect that the rationality community, with its “hero” focus, drives away many people who are like me in this sense. I’ve thought about walking away from it, for basically that reason. I could stay in Ottawa and be a nurse for forty years; it would fulfil all my most basic emotional needs, and no one would try to change me. Because oh boy, have people tried to do that. It’s really hard to be someone who just wants to please others, and to be told, basically, that you’re not good enough–and that you owe it to the world to turn yourself ambitious, strategic, Slytherin.

Firstly, this is mean regardless. Secondly, it’s not true.

Samwise was important. So was Frodo, of course. But Frodo needed Samwise. Heroes need sidekicks. They can function without them, but function a lot better with them. Maybe it’s true that there aren’t enough heroes trying to save the world. But there sure as hell aren’t enough sidekicks trying to help them. And there especially aren’t enough talented, competent, awesome sidekicks.

Miranda's post clearly struck a chord as it generated 200+ comments back in the days when LW was smaller, including an endorsement and apology from Eliezer, but the one I personally found most memorable was this one because it seemed so counterintuitive:

I am male. I have high testosterone. I love competing and winning. I am ambitious and driven. I like to make a lot of money. I make a lot of money. I prefer the sidekick role.

If someone asks me "King or Prince?" I will respond with Prince every time. Hey, you can still be royalty without the weight of the world on your shoulders. I would still be a hard working Prince, too. If some asks me "Candidate or Campaign Manager?" I'll take Campaign Manager, thank you. If someone asks me "President or Chief of Staff?" well, you know the answer by now.

The more money I make and the more wisdom and experience I acquire, the more people naturally turn to me to lead. And I do it when necessary. I'm even pretty good at it. But, I don't love it. I don't require it. I don't see myself as growing more in that direction.

For my own future reference, here are some "benchmarks" (very broadly construed) I pay attention to as of Nov 2025, a mix of serious and whimsical: 

• the AI village and blog, not really a "benchmark" per se but my richest source of intuitions about current frontier models' capabilities at open-ended long-horizon tasks by far, made me notice stuff like the Claudes being way better than other "benchmark-equiv" frontier models
• certain folks' opinions, e.g. lc on cybersec (like this take on AISLE), Kevin Buzzard and Terry Tao on math, Adam Karvonen on physical tasks relevant to manufacturing, Sarah Constantin on lit reviews, nostalgebraist on blog posts worth reading, Gwern and Jennifer Chen on writing diversity & creativity (but not Sam Altman), Cole Wyeth on novel ideas, Thane Ruthenis (although Thane's milestones are on a totally different capability tier), Steven Byrnes on AGI etc many others
• Chats on read.haus with AI simulations of prominent authors become preferable to reading the latter's real content. Scott Alexander, Sarah Constantin, Spencer Greenberg, Byrne Hobart, Tyler Cowen, Dwarkesh Patel, Andy Matuschak etc are all on there but they never come across quite right to me
• Starburst, fictional theoretical physics. I don't really get their leaderboard though
• a small set of work-related spreadsheet modelling problems I keep thinking current agents should easily do but they keep failing in very irritating ways, Claude Code included. I'm waiting for agents that will finally speed me up not slow me down on these. Possibly skill issue on my part  
• FWIW, Anthropic's members of technical staff estimates of productivity boost: currently 1.15-1.4x with Sonnet 4.5 for most, except that one person at 2x as "their workflow was now mainly focused on managing multiple agents", wonder if it's the same person Sholto Douglas mentioned worked with 9 agents at the same time
• how blind models see the earth, plot the Mandelbrot set, etc
• avoiding spiralling into spiritual bliss attractors (maybe this is just Claude being a hippie)
• just for fun, Gary Marcus' 5 challenges by 2029 (from 2022)
• their slope on the chart below exceeds that of humans (I'm not a fan of the notion of task horizon length, it bakes in perf plateauing that doesn't happen to humans thinking longer, hence slope)

• FrontierMath Tier 4 because I like math x AI, plus commentary like Kevin Buzzard's "I was amused this week to have been sent data on what happens if you ask lots of agents to try and solve these problems and you mark the question as being solved if at least one agent gets the answer correct at least once"
• vibe-proving math theorems in Lean except it doesn't take a week and isn't "extremely annoying" (despite Adam Mastroianni's argument that what a dream job really feels like is to be perpetually annoyed). The main issue is in verifying that the human proof-to-Lean code translation is faithful, which doesn't seem automatable 

• Epoch's Capabilities Index because it's general (composite metric of most of the high-profile benchmarks out there) stitched together using a methodology that seems intuitively correct (item response theory), although admittedly as someone who started out believing anything is measurable if you try hard enough I've gradually grown disillusioned enough to down-weight even ostensibly good composite benchmarks like ECI a fair bit. Also CAIS's definition of AGI
• Scale's Remote Labor Index because I work remotely. 230 projects from Upwork freelancers "excluding projects requiring physical labor, long-term evaluation, or direct client interaction", mean and median human completion time 29 and 11.5 hours respectively, mean and median project value $630 and $200. Manus at 2.50% tops the leaderboard, then Sonnet 4.5 > GPT-5 > ChatGPT Agent > Gemini 2.5 Pro last at 0.83%, which matches my impression of their relative "IRL competence" in the AI Village

My guess is you write them anyway because they are pretty easy to write in your default style and maybe just mostly fun-for-their-own-sake.

This was essentially what he told Dwarkesh:

Dwarkesh Patel 3:43:54

Okay, so this feels like a good place to pause the AI conversation, and there’s many other things to ask you about given your decades of writing and millions of words. I think what some people might not know is the millions and millions and millions of words of science fiction and fan fiction that you’ve written. I want to understand when, in your view, is it better to explain something through fiction than nonfiction?

Eliezer Yudkowsky 3:44:17

When you’re trying to convey experience rather than knowledge, or when it’s just much easier to write fiction and you can produce 100,000 words of fiction with the same effort it would take you to produce 10,000 words of nonfiction? Those are both pretty good reasons.

Dwarkesh Patel 3:44:30

On the second point, it seems like when you’re writing this fiction, not only are you covering the same heady topics that you include in your nonfiction, but there’s also the added complication of plot and characters. It’s surprising to me that that’s easier than just verbalizing the sort of the topics themselves.

Eliezer Yudkowsky 3:44:51

Well, partially because it’s more fun. That is an actual factor, ain’t going to lie. And sometimes it’s something like, a bunch of what you get in the fiction is just the lecture that the character would deliver in that situation, the thoughts the character would have in that situation. There’s only one piece of fiction of mine where there’s literally a character giving lectures because he arrived on another planet and now has to lecture about science to them. That one is Project lawful. You know about Project Lawful?

Dwarkesh Patel 3:45:28

I know about it. I have not read it yet.

Eliezer Yudkowsky 3:45:30

Most of my fiction is not about somebody arriving on another planet who has to deliver lectures. There I was being a bit deliberately like, — “Yeah, I’m going to just do it with Project Lawful. I’m going to just do it. They say nobody should ever do it, and I don’t care. I’m doing it ever ways. I’m going to have my character actually launch into the lectures.” The lectures aren’t really the parts I’m proud about. It’s like where you have the life or death, deathnote style battle of wits that is centering around a series of Bayesian updates and making that actually work because it’s where I’m like — “Yeah, I think I actually pulled that off. And I’m not sure a single other writer on the face of this planet could have made that work as a plot device.” But that said, the nonfiction is like, I’m explaining this thing, I’m explaining the prerequisite, I’m explaining the prerequisites to the prerequisites. And then in fiction, it’s more just, well, this character happens to think of this thing and the character happens to think of that thing, but you got to actually see the character using it. So it’s less organized. It’s less organized as knowledge. And that’s why it’s easier to write.

Echoing MichaelDickens' question on the EA Forum:

Indeed, I think it’s possible that there will, in fact, come a time when Anthropic should basically just unilaterally drop out of the race – pivoting, for example, entirely to a focus on advocacy and/or doing alignment research that it then makes publicly available.

Do you have a picture of what conditions would make it a good idea for Anthropic to drop out of the race?

Would also be interested to know how your thoughts compare with those of Holden's to a related question:

Rob Wiblin: I solicited questions for you on Twitter, and the most upvoted by a wide margin was: “Does Holden have guesses about under what observed capability thresholds Anthropic would halt development of AGI and call for other labs to do the same?” ...

Holden Karnofsky: Yeah. I will definitely not speak for Anthropic, and what I say is going to make no attempt to be consistent with the Responsible Scaling Policy. I’m just going to talk about what I would do if I were running an AI company that were in this kind of situation.

I think my main answer is just that it’s not a capability threshold; it’s other factors that would determine whether I would pause. First off, one question is: what are our mitigations and what is the alignment situation? We could have an arbitrarily capable AI, but if we believe we have a strong enough case that the AI is not trying to take over the world, and is going to be more helpful than harmful, then there’s not a good reason to pause.

On the other hand, if you have an AI that you believe could cause unlimited harm if it wanted to, and you’re seeing concrete signs that it’s malign — that it’s trying to do harm or that it wants to take over the world — I think that combination, speaking personally, would be enough to make me say, “I don’t want to be a part of this. Find something else to do. We’re going to do some safety research.”

Now, what about the grey area? What about if you have an AI that you think might be able to take over the world if it wanted to, and might want to, but you just don’t know and you aren’t sure? In that grey area, that’s where I think the really big question is: what can you accomplish by pausing? And this is just an inherently difficult political judgement.

I would ask my policy team. I would also ask people who know people at other companies, is there a path here? What happens if we announce to the world that we think this is not safe and we are stopping? Does this cause the world to stand up and say, “Oh my god, this is really serious! Anthropic’s being really credible here. We are going to create political will for serious regulation, or other companies are going to stop too.” Or does this just result in, “Those crazy safety doomers, those hypesters! That’s just ridiculous. This is insane. Ha ha. Let’s laugh at them and continue the race.” I think that would be the determining thing. I don’t think I can draw a line in the sand and say when our AI passes this eval.

So that’s my own personal opinion. Again, no attempt to speak for the company. I’m not speaking for it, and no attempt to be consistent with any policies that are written down.

From the Sonnet 4.5 system card, felt relevant:

9.3.5 Internal model evaluation and use survey

Details
A small number of members of technical staff spent over 2 hours deliberately evaluating Claude Sonnet 4.5’s ability to do their own AI R&D tasks. They took notes and kept transcripts on strengths and weaknesses, and then generated productivity uplift estimates. They were directly asked if this model could completely automate a junior ML researcher. ...

Claude Sonnet 4.5 results
When asked about their experience with using early snapshots of Claude Sonnet 4.5 in the weeks leading up to deployment, 0/7 researchers believed that the model could completely automate the work of a junior ML researcher. One participant estimated an overall productivity boost of ~100%, and indicated that their workflow was now mainly focused on managing multiple agents. Other researcher acceleration estimates were 15%, 20%, 20%, 30%, 40%, with one report of qualitative-only feedback. Four of 7 participants indicated that most of the productivity boost was attributable to Claude Code, and not to the capabilities delta between Claude Opus 4.1 and (early) Claude Sonnet 4.5.

(Found this via Ryan Greenblatt's recent post)

Yeah I remember watching this YouTube video about Puyi and thinking, huh, we do have a real historical example of Ajeya Cotra's young businessperson analogy from Holden's blog awhile back:

Imagine you are an eight-year-old whose parents left you a $1 trillion company and no trusted adult to serve as your guide to the world. You must hire a smart adult to run your company as CEO, handle your life the way that a parent would (e.g. decide your school, where you’ll live, when you need to go to the dentist), and administer your vast wealth (e.g. decide where you’ll invest your money).

You have to hire these grownups based on a work trial or interview you come up with -- you don't get to see any resumes, don't get to do reference checks, etc. Because you're so rich, tons of people apply for all sorts of reasons.

Your candidate pool includes:

• Saints -- people who genuinely just want to help you manage your estate well and look out for your long-term interests.
• Sycophants -- people who just want to do whatever it takes to make you short-term happy or satisfy the letter of your instructions regardless of long-term consequences.
• Schemers -- people with their own agendas who want to get access to your company and all its wealth and power so they can use it however they want.
   

Because you're eight, you'll probably be terrible at designing the right kind of work tests... Whatever you could easily come up with seems like it could easily end up with you hiring, and giving all functional control to, a Sycophant or a Schemer. By the time you're an adult and realize your error, there's a good chance you're penniless and powerless to reverse that.

Buck's comment upthread has a guess: 

Yeah, I agree that there's no one who Pareto dominates Eliezer at his top four most exceptional traits. (Which I guess I'd say are: taking important weird ideas seriously, writing compelling/moving/insightful fiction (for a certain audience), writing compelling/evocative/inspiring stuff about how humans should relate to rationality (for a certain audience), being broadly knowledgeable and having clever insights about many different fields.)

This also sounds sort of like how I'd describe what Scott Alexander is among the Pareto-best in the world at, just that Scott is high-verbal while Eliezer is high-flat (to use the SMPY's categorisation). But Scott's style seems more different from Eliezer's than would be explained by verbal vs flat. 

Phil Trammell on the bizarreness of real GDP as a proxy for tracking full automation and explosive economic growth in this recent podcast interview with Epoch After Hours:

Phil

... one thing that I think definitely is in this “Aha, here’s a theoretical curiosity” point is that real GDP is such a bizarre chimera of a variable that you could have full automation and really explosive growth in every intuitive sense of the term and yet real GDP growth could go down. 

An example of why it might at least not go up that much, which I think it probably won’t all work out this way but I don’t think this is crazy, is that you get this effect where there’s this common pattern you find where new goods, just as they’re introduced, have a really small GDP share. Because they have zero GDP share before they’re introduced. At first they’re really expensive—we’re not very productive at making them. As the price comes down, as we get more productive, the price falls but the quantity rises faster. The elasticity of demand is greater than one. Every time the price falls a little bit, the quantity rises a lot. So the dollar value of the good rises. So the share is rising. After a while it goes the other way, once the goods are really abundant, at least relative to everything else. 

Every time we have the price go up, the quantity only rises a little bit because we’re basically satiated in it. So you get this hump: new goods - small share; goods that have been around for a medium length of time that we’re mediumly productive at - high share, they dominate GDP; old goods like food - small share. So we’re continually going through this hump.

Everyone’s familiar with Baumol’s cost disease. But the way it’s usually presented is that AI might have less of an effect on growth than you might have thought, because we’ll be bottlenecked by the few things that have not yet been automated that you still need people for. And actually, you can have Baumol after full automation. Because, remember the hump, right? Real GDP growth at a given time is the weighted average of the growth rates of all the goods where the weightings are the GDP shares. The GDP shares will be dominated by the goods that we’re intermediately productive at in this view.

So let’s say for every good you have its own specific technology growth rate. Like how quickly it can be produced is some arbitrary function of its current technology level. It can be hyperbolic. You can have A dot equals A squared or something. So for every good, there is some finite date by which we’ll be able to produce infinite quantities of it in finite time.

So it’ll be free. So GDP share will be zero. And we just go through these ever higher index goods, ever more complex goods over time. And at any given time, all of GDP are the goods that have a productivity level of five or whatever happens to be in the middle as far as GDP shares go. So some effect like that can produce something like a Baumol effect even after full automation.

I think it would be pretty weird if that kept the absolute number low. Like anything as low as the current number indefinitely. But the idea that maybe it causes measured real GDP growth to not be that high for a while when the world is starting to look remarkably different doesn’t seem crazy to me. And maybe it’s worth knowing and having as a scenario in your back pocket in case things start looking weird and anyone says “What are you talking about? I don’t see the numbers.” I’m trying to be cautious, but that’s an example of destructive economic theory.

Anson

Do we have any quantitative sense of what the hump looks like?

Phil

That’s a good question. There’s that Besson paper and you could just do a bunch of case studies by good. I should look into that more quantitatively.

and then a bit further down, on the chain-weighting in calculating real GDP growth making it a totally path-dependent measure:

Phil

... I mean, digging into the theory of what chain-weighting is has made me pretty viscerally feel like real GDP is a much slipperier concept than I ever used to think.

Here’s a fun fact. This is crazy. So real GDP and lots of real variables like inflation-adjusted variables, real capital or whatever, let’s say real GDP, is not a quantity. What do I mean? It’s not. Here’s what I mean. Imagine a timeline of some economy. So, the US from 1950 to 2025, 75 years. And imagine an alternative timeline with an alternative economy living it out that’s exactly the same as the US in 1950, at the beginning, in its own 1950, and exactly like the US in 2025, at the end in year 75. But in the middle things happened in a different order. So the microwave was invented in 2006, and the iPhone came out in 1971. And the distribution of wealth changed hands, evolved in a different way. But at the end, it’s exactly the same. Everyone’s got the same preferences. Exchanges the same goods and services for the same dollar bills. Atom for atom. Everything unfolds exactly the same in 2025 and in the 1950 on both timelines. Timeline A, timeline B.

Unless people have homothetic preferences, meaning that the fraction of their income they spend on each good is constant, no matter how rich they are. So no luxuries or inferior goods, which is completely wrong. You don’t spend the same fraction on food when you’re starving as when you’re richer. But unless people have homothetic preferences that are the exact same preferences across the population and totally stable over time—unless those three conditions are met, there is a timeline B on which real GDP growth chain-weighted across the years with perfect measurement is any number. 

Anson

Okay.

Phil

Isn’t that crazy? I mean, even the fact that there could be any variation means that, to my mind, real GDP is not a quantity. Because it’s baking in the history. You see what I’m saying? A yardstick shouldn’t matter—the order in which you measure things. It should order things in the same way. But the order in which things happen can change what share of GDP a given good was while it was growing quickly.

So let’s say there’s two of us and one of us is going to be rich one year, and the other one is going to be rich the other year. And the stuff that I like more, I’m going to bid up the price. I’ve got a lot of clones that have my preferences and you’ve got a lot of clones. We bid up the price more of the things we like when we’re rich. The way things happen is that the things we like are growing quickly in absolute units while we happen to have the money. So our preferences are mostly determining what GDP is. And the things you like are growing quickly when you and your clones have the money. Real GDP is going to be higher across the two years than if it’s the other way, where the things I like grow when I’m poor and vice versa.

And it’s that kind of effect that can mean that you can scramble things up so that as long as people depart from perfect homotheticity, constant preferences, same across population, then real GDP can be any number. So maybe I’ve overinternalized this. But given that I’ve overinternalized this, I sort of feel like I can’t separate the theory from the overall opinion I think.

Phil's point isn't new, John Wentworth brought it up awhile back:

I sometimes hear arguments invoke the “god of straight lines”: historical real GDP growth has been incredibly smooth, for a long time, despite multiple huge shifts in technology and society. That’s pretty strong evidence that something is making that line very straight, and we should expect it to continue. In particular, I hear this given as an argument around AI takeoff - i.e. we should expect smooth/continuous progress rather than a sudden jump.

Personally, my inside view says a relatively sudden jump is much more likely, but I did consider this sort of outside-view argument to be a pretty strong piece of evidence in the other direction. Now, I think the smoothness of real GDP growth tells us basically-nothing about the smoothness of AI takeoff. Even after a hypothetical massive jump in AI, real GDP would still look smooth, because it would be calculated based on post-jump prices, and it seems pretty likely that there will be something which isn’t revolutionized by AI. ...

More generally, the smoothness of real GDP curves does not actually mean that technology progresses smoothly. It just means that we’re constantly updating the calculations, in hindsight, to focus on whatever goods were not revolutionized. On the other hand, smooth real GDP curves do tell us something interesting: even after correcting for population growth, there’s been slow-but-steady growth in production of the goods which haven’t been revolutionized.

There's a bunch of Metaculus questions on explosive economic growth showing up in GDP (e.g. this, this, this, this etc) which I think are just looking at the wrong thing because the askers and most forecasters don't get this proxy decoupling. I've brought up John's post before and elsewhere too because it just seemed odd to me that this wasn't being internalised, e.g. I don't know if Open Phil still thinks in terms of explosive growth as >30% p.a. GWP like they used to but my impression is they still do. It would be silly if explosive growth was underway yet consensus couldn't be formed to coordinate and guide large-scale decision-making because everyone was anchoring to real GDP or anything calculated remotely like it.