I do research around longtermism, forecasting and quantification, as well as some programming, at the Quantified Uncertainty Research Institute (QURI). I'm currently in the Bahamas as part of the FTX EA Fellowship

I'm also a fairly good forecaster: I started out on predicting on Good Judgment Open and CSET-Foretell, but now do most of my forecasting through Samotsvety, of which Scott Alexander writes:

Enter Samotsvety Forecasts. This is a team of some of the best superforecasters in the world. They won the CSET-Foretell forecasting competition by an absolutely obscene margin, “around twice as good as the next-best team in terms of the relative Brier score”. If the point of forecasting tournaments is to figure out who you can trust, the science has spoken, and the answer is “these guys”.

I have also been running a Forecasting Newsletter since April 2020, and have written, a search tool which aggregates predictions from many different platforms. I also generally enjoy winning bets against people too confident in their beliefs.

Otherwise, I like to spend my time acquiring deeper models of the world, and generally becoming more formidable. A good fraction of my research is available either on the EA Forum or on I'm particularly proud of my Estimating value series.

I was a Future of Humanity Institute 2020 Summer Research Fellow, and then worked on a grant from the Long Term Future Fund to do "independent research on forecasting and optimal paths to improve the long-term."

Before that, I studied Maths and Philosophy, dropped out in exasperation at the inefficiency, picked up some development economics; helped implement the European Summer Program on Rationality during 2017, 2018 and 2019, and SPARC during 2020; worked as a contractor for various forecasting and programming projects; volunteered for various Effective Altruism organizations, and carried out many independent research projects. In a past life, I also wrote a popular Spanish literature blog, and remain keenly interested in Spanish poetry.

You can share feedback anonymously with me here.


Forecasting Newsletter
Inner and Outer Alignment Failures in current forecasting systems

Wiki Contributions


The framework is AI strategy nearcasting: trying to answer key strategic questions about transformative AI, under the assumption that key events (e.g., the development of transformative AI) will happen in a world that is otherwise relatively similar to today’s.

Usage of "nearcasting" here feels pretty fake. "Nowcasting" is a thing because 538/meteorology/etc. has a track record of success in forecasting and decent feedback loops, and extrapolating those a bit seems neat. 

But as used in this case, feedback loops are poor, and it just feels like a different analytical beast. So the resemblance to "forecasting" seems a bit icky, particularly if you are going to reference "nearcasting" without explanation it in subsequent posts: <>.  

I spent a bit thinking about a replacement term, and I came up with "scenario planning absent radical transformations analysis", or SPARTA for short. Not perfect, though.

See this comment: <>

I am not defending the language of the OP's title, I am defending the content of the post.

You don't have strategic voting with probabilistic results. And the degree of strategic voting can also be mitigated.

Copying my second response from the EA forum:

Like, I feel like with the same type of argument that is made in the post I could write a post saying "there are no voting impossibility theorems" and then go ahead and argue that the Arrow's Impossibility Theorem assumptions are not universally proven, and then accuse everyone who ever talked about voting impossibility theorems that they are making "an error" since "those things are not real theorems". And I think everyone working on voting-adjacent impossibility theorems would be pretty justifiedly annoyed by this.

I think that there is some sense in which the character in your example would be right, since:

  • Arrow's theorem doesn't bind approval voting.
  • Generalizations of Arrow's theorem don't bind probabilistic results, e.g., each candidate is chosen with some probability corresponding to the amount of votes he gets.

Like, if you had someone saying there was "a deep core of electoral process" which means that as they scale to important decisions means that you will necessarily get "highly defective electoral processes", as illustrated in the classic example of the "dangers of the first pass the post system". Well in that case it would be reasonable to wonder whether the assumptions of the theorem bind, or whether there is some system like approval voting which is much less shitty than the theorem provers were expecting, because the assumptions don't hold.

The analogy is imperfect, though, since approval voting is a known decent system, whereas for AI systems we don't have an example friendly AI.

Copying my response from the EA forum:

(if this post is right)

The post does actually seem wrong though. 

Glad that I added the caveat.

Also, the title of "there are no coherence arguments" is just straightforwardly wrong. The theorems cited are of course real theorems, they are relevant to agents acting with a certain kind of coherence, and I don't really understand the semantic argument that is happening where it's trying to say that the cited theorems aren't talking about "coherence", when like, they clearly are.

Well, part of the semantic nuance is that we don't care as much about the coherence theorems that do exist if they will fail to apply to current and future machines

IMO completeness seems quite reasonable to me and the argument here seems very weak (and I would urge the author to create an actual concrete situation that doesn't seem very dumb in which a highly intelligence, powerful and economically useful system has non-complete preferences).

Here are some scenarios:

  • Our highly intelligent system notices that to have complete preferences over all trades would be too computationally expensive, and thus is willing to accept some, even a large degree of incompleteness. 
  • The highly intelligent system learns to mimic the values of human, which end up having non-complete preferences, which the agent mimics
  • You train a powerful system to do some stuff, but also to detect when it is out of distribution and in that case do nothing. Assuming you can do that, their preference is incomplete, since when offered tradeoffs they always take the default option when out of distribution. 

The whole section at the end feels very confused to me. The author asserts that there is "an error" where people assert that "there are coherence theorems", but man, that just seems like such a weird thing to argue for. Of course there are theorems that are relevant to the question of agent coherence, all of these seem really quite relevant. They might not prove the things in-practice, as many theorems tend to do. 

Mmh, then it would be good to differentiate between:

  • There are coherence theorems that talk about some agents with some properties
  • There are coherence theorems that prove that AI systems as will soon exist in the future will be optimizing utility functions

You could also say a third thing, which would be: there are coherence theorems that strongly hint that AI systems as will soon exist in the future will be optimizing utility functions. They don't prove it, but they make it highly probable because of such and such. In which case having more detail on the such and such would deflate most of the arguments in this post, for me.

For instance:

Coherence arguments’ mean that if you don’t maximize ‘expected utility’ (EU)—that is, if you don’t make every choice in accordance with what gets the highest average score, given consistent preferability scores that you assign to all outcomes—then you will make strictly worse choices by your own lights than if you followed some alternate EU-maximizing strategy (at least in some situations, though they may not arise). For instance, you’ll be vulnerable to ‘money-pumping’—being predictably parted from your money for nothing.

This is just false, because it is not taking into account the cost of doing expected value maximization, since giving consistent preferability scores is just very expensive and hard to do reliably. Like, when I poll people for their preferability scores, they give inconsistent estimates. Instead, they could be doing some expected utility maximization, but the evaluation steps are so expensive that I now basically don't bother to do some more hardcore approximation of expected value for individuals, but for large projects and organizations.  And even then, I'm still taking shortcuts and monkey-patches, and not doing pure expected value maximization.

“This post gets somewhat technical and mathematical, but the point can be summarised as:

  • You are vulnerable to money pumps only to the extent to which you deviate from the von Neumann-Morgenstern axioms of expected utility.

In other words, using alternate decision theories is bad for your wealth.”

The "in other words" doesn't follow, since EV maximization can be more expensive than the shortcuts.

Then there are other parts that give the strong impression that this expected value maximization will be binding in practice:

“Rephrasing again: we have a wide variety of mathematical theorems all spotlighting, from different angles, the fact that a plan lacking in clumsiness, is possessing of coherence.”


“The overall message here is that there is a set of qualitative behaviors and as long you do not engage in these qualitatively destructive behaviors, you will be behaving as if you have a utility function.”


  “The view that utility maximizers are inevitable is supported by a number of coherence theories developed early on in game theory which show that any agent without a consistent utility function is exploitable in some sense.”


Here are some words I wrote that don't quite sit right but which I thought I'd still share: Like, part of the MIRI beat as I understand it is to hold that there is some shining guiding light, some deep nature of intelligence that models will instantiate and make them highly dangerous. But it's not clear to me whether you will in fact get models that instantiate that shining light. Like, you could imagine an alternative view of intelligence where it's just useful monkey patches all the way down, and as we train more powerful models, they get more of the monkey patches, but without the fundamentals. The view in between would be that there are some monkey patches, and there are some deep generalizations, but then I want to know whether the coherence systems will bind to those kinds of agents.

No need to respond/deeply engage, but I'd appreciate if you let me know if the above comments were too nitpicky.

I am also curious about the extent to which you are taking the Hoffman scaling laws as an assumption, rather than as something you can assign uncertainty over.

I thought this was great, cheers. 


Next, we estimate a sufficient horizon length, which I'll call the k-horizon, over which we expect the most complex reasoning to emerge during the transformative task. For the case of scientific research, we might reasonably take the k-horizon to roughly be the length of an average scientific paper, which is likely between 3,000 and 10,000 words. However, we can also explicitly model our uncertainty about the right choice for this parameter.

It's unclear whether the final paper would be the needed horizon length.

For analogous reasoning, consider a model trained to produce equations which faithfully describe reality. These equations tend to be quite short. But I imagine that the horizon length needed to produce them is larger, because you have to keep many things in mind when doing so. Unclear if I'm anthropomorphizing here.

But I think it is >30% likely you can compensate for past over or under estimations.

I'd bet against that at 1:5, i.e., against the proposition that the optimal forecast is not subject to your previous history

This is true in the abstract, but the physical word seems to be such that difficult computations are done for free in the physical substrate (e.g,. when you throw a ball, this seems to happen instantaneously, rather than having to wait for a lengthy derivation of the path it traces). This suggests a correct bias in favor of low-complexity theories regardless of their computational cost, at least in physics.

Load More