He does prefer email, at: jason.r.derr@protonmail.com
Jeffrey Heninger
Talking to an LLM is not a substitute for reading the relevant literature or talking to an expert in the field.
I sent this to my friend Jason Derr, who is an actual expert. I did fusion plasma physics—he does space plasma physics focusing on aurora. I can call out egregious errors, but if I start diving into the details, I’m likely to make mistakes.[1] He has been working with these details for years.
The plasma physics in this post is mostly wrong. The figure I criticized earlier has the highest density of errors, but the text is also riddled with them. Other parts of the posts also look suspicious, but we’re not relevant experts for them. A few of the major errors:
Solar storms ≠ geomagnetic storms ≠ substorms ≠ auroras ≠ electrojet intensifications ≠ coronal mass ejections. If you want to be vague, say “space weather”. This post is mostly about geomagnetic storms (in the Earth’s magnetic field), not solar storms (in the sun’s atmosphere).
There are multiple relevant satellites at L1. You can see a list here.
The direction of the magnetic field of the solar wind is not a coin flip. It’s mostly determined by which side of the heliospheric current sheet the Earth is currently on, and there’s finer structure both around the current sheet and coming from the CME. We keep track of the local structure, and can figure out the structure of the CME by looking at it when it leaves the surface of the sun.
If the main point of the article is that solar activity can have major impacts on the grid, that is true—and also on GPS satellites and HF radio communication. Geomagnetic activity’s effects on the economy and infrastructure are relatively unknown by the public. But it is a topic which has substantial research. One neglected area that Jason mentioned is getting better magnetotelluric data.
Jason gave me a “sparse” list of recommendations of the literature if you want to learn about this topic, and is happy to chat with people who are interested:
On the uncertain intensity estimate of the 1859 Carrington storm
Major Space Weather Risks Identified via Coupled Physics-Engineering-Economic Modeling
Global landscape of space weather observations, research and operations
The impact of space weather on the national-scale power grid and the associated economic losses
I like posts like this that dive into something that most people are not an expert in, and have written some similar ones myself. But they need to be researched by reading the actual literature, not by asking an LLM.
- ^
For example, in my comment on electrojets, I described the figure as “normal conditions”, when those currents only rarely look like that. The figure is meant to illustrate the dawn-to-dusk or dusk-to-dawn directions, which is how those current systems are defined.
No, they’re not.
In the left hand picture, it shows electrojets in 3 places: the equatorial electrojets near the equator (correct), the auroral electrojets near the equator (wrong), and the equatorial electrojets in the mid latitudes (wrong).
The “totally reversed” equatorial electrojet in the right hand picture points the same direction as the one in the left hand picture.
The auroral electrojets in the right hand picture are 2 rotating ovals near the pole. Actual auroral electrojets lie along the auroral ring around the pole. Normally, there are two of them, pointing in the nightside direction. During a substorm, one of these strengthens and can become strong enough to reverse the other—forming a complete ring around the pole.
Another figure, which shows what the auroral electrojets actually look like, under normal conditions.
Even if this part of the picture were correct, a scientific diagram that does not contain relevant true information (like the magnetic field of the Earth, or which side of the Earth is day), but does include a bunch of false information (the ionosphere is lower than the atmosphere, the magnetosphere is centered on the sun, reconnection happens inside the atmosphere at the boundary of the gray lumpy ionosphere, etc), is a bad diagram.
I asked Gemini 3.1 Pro Preview to make an XKCD-style diagram of a CME’s effects on electrojets. This was a one-shot!
It’s also completely wrong. The most egregious error is repeatedly portraying the Earth’s magnetic field as a sphere (magnetic fields can never form spheres) - unless it’s not bothering to show the magnetic field at all?
Here’s an actual diagram:
The biggest differences:
Epoch looks at more types of data, including Usage, Compute Spend, and Valuation.
Epoch does not look at projections, either those made by the company itself or by extrapolating the current trend.
Differences in methodology, for the things we both consider:
Epoch uses reports of annualized revenue (the revenue in a month x12), while I focus on annual revenue (the amount of revenue a company received in a calendar year). I discuss this in the post. The results are similar for OpenAI, the only company I have decent revenue data for.
Epoch does not use estimates for the number employees made by market research organizations, like LeadiQ or RocketReach. Instead, they focus on media reports of the number of employees. This difference does have significant consequences.
The number of employees from market research organizations is often higher than the media reports used by Epoch.
For Anthropic, the media reports Epoch used have been lower than the data I used, but they are converging.
For OpenAI, the data were similar until the end of 2023 (8 years since the company was founded). After this, the media reports have been about a factor of 2 lower than the market research organizations.
I do not know why this discrepancy exists. The market research organizations do not publish their methodology, but they presumably have incentives to provide accurate information, and so their estimates should be reasonable. Media reports from organizations like The Information also have incentives to be accurate, and try to avoid publishing unverified numbers. For OpenAI in particular, Noam Brown has cautioned against the higher numbers.
If we accept Epoch’s data for number of employees, then OpenAI is growing at a rate of 1.7x per year and Anthropic is growing at a rate of 3x per year. In 2027, OpenAI would have 7400 employees and Anthropic would have 14,000 employees. In 2030, OpenAI would have 37,000 employees and Anthropic would have 360,000 employees. In 2033, OpenAI would have 180,000 employees and Anthropic would have 9.4 million employees. In 2033, Anthropic would be larger than any company today, and larger than the entire US tech industry. This would be dramatic, although it still seems like this would be less of a limitation than receiving all global wealth as capital.
Epoch AI has just released a database with similar data, in addition to some other metrics, here: https://epoch.ai/data/ai-companies
The same day this was posted, Sri Muppidi at The Information shared some new projections for OpenAI’s finances for 2025-2030.[1] The graph includes 3 projections: R&D compute cost, Inference compute cost, and Revenue.
OpenAI is not projecting current growth rates to continue. Comparing to the previous data, these projection show a much lower rate of growth.
R&D compute cost is not just projected to grow slower than an exponential over the next 5 years, it is projected to grow slower than a linear trend. Or, if we treat this as a logistic curve that has had exponential growth in the past, OpenAI is projecting that we will reach the point of steepest growth next year, and that the leveling off will be apparent in 2-3 years.
Compared to the previous projections, these projections suggest that OpenAI will be cash flow positive by 2027, not 2029 - assuming that R&D compute and Inference compute constitute most of their total costs. They also suggest that OpenAI currently has enough capital to last until then.
The revenue projections are not that different from the previous ones ($55B vs $60B for 2027), so it seems as though OpenAI is now projecting that it will spend less over the next few years than it previously projected. I think that the main difference is that compute is not the total costs. If we include the unspecified other costs, the projection might be the same as before.
- ^
To get to “R&D costs amount to ~45% of total revenue in 2030”, there is also around $40B of other research spending—which seems to be mostly “compensation costs.”
- ^
Trends in Economic Inputs to AI
I applied it to an office because the conversation that caused me to write this post involved an AI safety group office that the person I was talking to used to work at, which does function as a community.
It’s plausible to me that these recommendations work better for other kinds of community.
My guess is that standardization has been more important for you than for the typical member of the church. It sounds like you move a lot more than most members, and so you spend a lot more of your time having just moved somewhere new. Standardization is helpful to build community when you’re traveling or just moved somewhere new, but most people aren’t in those situations all that often.
I also claim that standardization by itself does not build community. There is not a particularly strong community in McDonalds or in airports—despite these being very standardized situations. What standardization does is it reminds you of the similar situations you previously have been in. This allows the sense of community to travel with you between wards. But if your home ward does not feel like a community, going to something that looks similar doesn’t make you suddenly feel at one with them.
I didn’t mention narratives about persecution, and maybe I should have. They don’t feel like a strong contributor to feelings of community for me personally—but I might be unusual here.[1] I’m also not sure how to disentangle narratives of persecution from actual experiences of people treating them differently because they are Mormon. Either way, I don’t think that this is something other groups trying to build community should want to copy.
- ^
The fact that I post on LessWrong is some evidence that I’m not near the center of the distribution.
- ^
The term “Gentile” to refer for non-Mormons was mostly a thing in the 1800s, and is basically not a thing today. The last time this usage occurred in General Conference was in 1981, and that was quoting something written in the 1800s. The last time this usage occurred in General Conference not quoting something else was in 1936. (The LDS General Conference corpus is great.) The terminology that is used is “the world”, which puts much less focus on the uncleanness of individuals.
Mormonism, outside of Utah, is not disassociated from society at large. The attitude is instead “in the world, but not of the world”. There are some people for whom most of their friends are in the Church, and marriage within the Church is strongly encouraged. But there aren’t Mormon-specific workplaces or grade schools.[1] We do not think that entertainment or education can only be obtained within the community.
Relating insularity to ethnic purity also seems very wrong here. Missionary work is a major thing ! About 1⁄3 of all current members of the Church in the US were not raised in the Church. The world may be perceived as a hostile thing, but it’s full of people who we might potentially convert.
- ^
Our Church does have some private grade schools in the Pacific Islands, but none in the US. The ward I grew up in had <20 teenagers, and we went to at least 5 different high schools. In my 22 years of full time education, I have had exactly 1 class that had another member of the Church in it.
At the university level, things are more complicated. The Church does own BYU, BYU-Idaho, and BYU-Hawaii. There’s also SVU, although it’s not run by the Church itself.
- ^
I agree that details matter. The system, in principle, equalizes status by passing around callings—and in practice often does. But it doesn’t always work the way it should and you get The Same 10 Families rotating between the higher effort or leadership callings, while other people stay in lower effort callings.
Adequately answering this question would be at least its own blog post, but here’s a gesture at a response:
In the 1600s, there was a debate over whether natural philosophy should be structured like math (‘rationalists’) or whether it should be based on sense & memory (‘empiricists’). The empiricists won one of the most lopsided victories in the history of philosophy and empirical science was born.[1]
Joseph Smith & the Book of Mormon have a wildly more empirical approach than any of the religions at the time.[2] Alma 32′s ‘experiment on the word’. Moroni 10 is even framed in terms of falsifiability, a full century before Karl Popper introduced it to the philosophy of science.
Yes, observations can be theory-laden, but that doesn’t mean that we should abandon the empirical project—in theology any more than in other fields.
This is less relevant for you than for other people on this site, but I should maybe note that I don’t think that the evidence visible from outside the church is sufficient to ‘prove the existence of God’ or something like that. I do think that it is sufficient to justify a serious investigation, and that the empirical evidence builds up over time as you build a personal relationship with God.
- ^
See Shapin & Schafer’s Leviathan and the Air Pump (book review) for more details. Despite being postmodernists, the authors have done substantial historical work.
- ^
Parts of Protestantism have since become more empirical, emphisizing personal experience with the Spirit over systematic theology. I think that this reflects Pentecostal influence, but am not familiar enough with the history to be sure.
- ^
Joseph Lawal on YouTube also has some good epistemology arguments, but I don’t remember which video they were in.
- ^
I think you’d be interested in Tocqueville’s description of how New England towns worked in the early 1800s. My guess is that the system of callings descends from it, and it was substantially more democratic.
Search: “Limits of the township” to find the relevant section.
I was assuming that building strong community is a good thing, because of the post this is responding to. If Scott (or other people) are looking at Mormonism to see what they can learn about building strong communities in a liberal society, it is better if they have an accurate understanding of how the community works.
I think that the things that seem most likely to be worth exporting are ministering and callings. Callings seem harder to export. but have been very important for me feeling part of the community. Ministering seems easier to export, and I wouldn’t be that surprised if rationalists end up doing it better than we do.
Some of the things you’re talking about do not feel like the Church as I have experienced it. I only have my limited view, and for example have never lived in Utah, so maybe we’ve just experienced different things.
I don’t think that any of the leaders I have known could reasonably be described as “covertly power-seeking and who can succeed at this by veneer of niceness”. Including the leaders who I have had significant disagreements with. All of the ones I’ve dealt with are sincerely trying, and would be relieved to have a less effortful calling. I don’t think that most of the niceness you see in the Church is actually a veneer. (N here is maybe 20, if you include bishops, counselors, and elders’ quorum presidents.)
Mechanistically, the Church is a really ineffective route for power seeking. ‘Advancing’ in callings is a slow, highly uncertain process, during which time you are expected to do a ton of service. Maybe there’s some inflection point above stake president where this stops being true (I wouldn’t know), but at least at the levels I’ve been able to see, the incentives point strongly against trying to get more power.[1] If this is a bigger problem in Utah than elsewhere, then I would guess that it would be because having a leadership calling helps you get promotions at work (or something else), and so there are more incentives coming from outside the Church itself.
I agree that there is too much deference without review. I would prefer deference with review. For example, the first time I was called to be ward mission leader, I told my bishop why I thought I was not a particularly good choice for the role, and suggested that we both go and think & pray about it and talk again next week. We did, and I ended up accepting the calling. The leaders I’ve had have reacted well to this—at least some of them seem to prefer it to either deference without review or outright refusal. This is a direction I am trying to push Church culture in.
The set of lessons is not fixed. The lesson topics are suggested by the Church. This only results in the same lesson if the teacher is putting in a minimal amount of effort. Even in this case, someone in the class can dramatically improve it by asking an interesting[2] question, at least if there are some other people who are willing to engage. If no one there is willing to put in anything more than a minimal amount of effort, then the lessons will be repetitive and boring—and no amount of institutional design will fix it.
I endorse proselytizing. If you think that your believes are true and good, then it is good to offer them to the rest of the world. Even if it is through Harry Potter fanfiction instead of only rigorous argument.
I don’t know what treatment you’ve received when leaving the Church. My impression is that people’s friendships in the Church gradually fade away because it’s much harder to maintain a friendship when you don’t have a built in plan to see each other at least once a week, and as people move away and you don’t meet the new people. This can mean that you feel isolated if you come back to visit, but this doesn’t seem like an avoidable problem. If you’ve been treated worse than this, then I’m sorry.
- ^
Then why do people do it? Because of a sense of duty—there are norms against refusing callings. My guess is that if the norms around not refusing callings significantly weaken, bishop would be one of the harder callings to get anyone to agree to do.
- ^
‘Interesting’ here does not mean ‘controversial’. ‘Interesting’ means ‘something that other people will have nontrivial responses to’. Flagrantly controversial questions are often not interesting, if they result in predictable responses. Crafting interesting questions for Sunday school is an art that I’ve practiced, and I think it’s worthwhile for other people in the Church to practice too.
- ^
Why Latter-day Saints Have Strong Communities
I would also make the same prediction for Q > 10. Or when CFS first sells electricity to the grid. These will be farther into the future, but I do not think that this culture will have changed by then.
I think that I predict the opposite (conditional on what exactly is being predicted).
What exactly would count as a GPT-3 moment for fusion? How about an experiment demonstrating reactor-like conditions? This is roughly equivalent to what I referred to as ‘getting fusion’ in my book review.
My prediction is that, after Commonwealth Fusion Systems gets Q > 5 on SPARC, they will continue to supply or plan to supply HTS tape to at least 3 other fusion startups.
I agree that this is plausibly a real important difference, but I do not think that it is obvious.
The most recent augmentative technological change was the industrial revolution. It has reshaped virtually every every activity. It allowed for the majority of the population to not work in agriculture for the first time since the agricultural revolution.
The industrial revolution centered on energy. Having much cheaper, much more abundant energy allowed humans to use that energy for all sorts of things.
If fusion ends up being similar in cost to existing electricity production, it will be a substitutional technology. This is the thing that we are working on now (well, also making it work at all). People who work in fusion focus on this because it is the reasonable near/medium term projection. If fusion ends up being substantially cheaper, it will be an augmentative technology. It is not at all clear that this will happen, because we can’t know how the costs will change between the first and thousandth fusion power plant.
Notably, we don’t know if foom is going to be a thing either.
The narrative around the technology is at least as important as what has happened in the technology itself. The fusion community could frequently talk about how incredible the industrial revolution was, and how it powered Britain to global dominance for two centuries. A new source of energy might do the same thing ! But this is more hype than we feel we ought to offer, and the community’s goal is not to create a dominant superpower.
Even if foom is going to happen, things would look very different if the leaders credibly committed to helping others foom if they are first. I don’t know if this would be better or worse from a existential risk perspective, but it would change the nature of the race a lot.
The question, “Do you want to live forever?” has seen several academic surveys. Here’s a few:
PEW: 7% want to live to 120+
Barnett & Helphrey: 34% yes, 40% no, 26% unsure
Donner et al.: If guaranteed physical & mental health, 53% yes
These are very different results, and I’m not sure what going on.