David T

Karma: 1,426

David T 28 Feb 2026 22:47 UTC
8 points
5 ∶ 0
in reply to: Aithir’s comment on: Pete Hegseth declares Anthropic a supply-chain risk
In this case Anthropic chose to supply the DoW via a partnership with a company deeply embedded in the administration’s part of the political spectrum, and even pointedly denied any objections to being used to support the administration’s little expedition in Venezuela, and the administration decided that wasn’t enough. There are many criticisms that can be made of Anthropic’s stance on those issues; reluctance to engage with the current US administration isn’t one of them.
If declining to actively support MAGA’s demands they support development of AI with the explicit purpose of being an autonomous killing device is “virtue signalling”, what’s left of “AI alignment” to pursue?

David T 26 Feb 2026 20:19 UTC
4 points
0 ∶ 0
on: How I read studies
A useful post and interesting starting point for further discussion
Few more that spring to mind
- is there an intuitively plausible alternative/complementary causal factor which might explain the results? Has the study made some sort of attempt to control for this or estimate the effect size?
- Does the paper include a large number of hypotheses and find some of them to be statistically significant at the 5 or 10% level.? You would expect that to happen by random chance and it smells of p-hacking, though including them all in the paper is at least more intellectually honest than the alternative approach of testing a lot of hypotheses and only acknowledging the ones that were [incidentally] statistically significant. This is why preregistration is valuable. Note that testing so many hypotheses one of them is bound to be a “finding” is not the same thing as testing whether an association between x and y persists across a large number of alternative regressions including other variables that might plausibly have a relationship; that’s good practice.
- Is there a standard methodology for conducting research of this nature in the field? has it been used, and if not, has a plausible rationale for taking a different approach been provided?
would be interesting to hear some of the more specialized ones that organizations like GiveWell and Rethink Priorities that evaluate a lot of research papers in particular fields use.
n.b. on the SMC vs ITN example, I’m fairly confident the answer is that the ITNs are a baseline directly comparable to “no treatment” as they shouldn’t affect the progression from bite to symptomatic malaria infection targeted by SMC at all; they simply reduce the frequency of bites (but not to zero if the sample size is sufficiently large). Prevalance of malarial bites varies between cohorts in “no treatment studies” already. Access to some level of treatment after the fact (HMM) isn’t a problem of study construction either; it complicates comparing severe malaria or death statistics with papers where sufferers may have had no treatment at all, but if anything would probably reduce the reported effect size for SMC. Medical ethics means the appropriate baseline/comparator for lifesaving treatment usually isn’t “do absolutely nothing”, it’s “do the [next] best alternative”

David T 25 Feb 2026 9:11 UTC
5 points
0 ∶ 0
in reply to: Ethan Roland’s comment on: Exclusive: Hegseth gives Anthropic until Friday to back down on AI safeguards
I suspect that any resolution to this dispute is likely to be a lot less public than the OpenAI one
It’s fairly obvious though that Amodei is signalling the company didn’t object to the use of Claude to support the Venezuela operation, and that the company freely choose to be a defense contractor with a formal partnership with Palantir when they had plenty of other revenue/capital sources...

David T 20 Feb 2026 21:43 UTC
2 points
0 ∶ 0
on: A More Precise Expression For Evaluating Strong Longtermism
It seems like most of the additional coefficients you’ve added are impossible to estimate with any degree of confidence, particularly when it is plausible the impact may be negative. Whether it was the intention or not, that is the main message I get from your formulation
As someone who is not a strong longtermist, I note that an advantage for using non longtermist heuristics to evaluate impact is that identifying whether an action appears robustly positive for aggregate utility [for humans] on earth up to time t is much easier than anticipating the effect on the Virgo supercluster after time t
(A more sophisticated approach might use discounting for temporal and extreme spatial distance rather than time bounding, but amounts to the same thing; attaching zero weight to the estimated impact of my actions on the Virgo supercluster a thousand years from now)

David T 20 Feb 2026 20:57 UTC
9 points
2 ∶ 0
on: Effective Altruism Has Very Good Epistemics
EAs were also warning, for a long time, about the importance of health aid sent overseas. In contrast, non-EA leftists were more likely to call these institutions colonialist and call EAs racist for neglecting domestic political issues. But when Trump got elected, we were vindicated in the worst of ways: he destroyed much of USAID, and this was the single act in his presidency that led to the most deaths.
This feels like a strange argument to make, and one which seems to be trying way too hard to find evidence of vindication even in failures, which ironically is the opposite of what people with good epistemics should be doing. EAs were criticised [by critics whose arguments greatly varied in quality] for tending to treat international aid as primarily an optimization problem best addressed by small specialist charities and individuals maximising their donations, and largely ignoring the political dimension.
Then domestic political issues killed government programmes funding traditional Big Aid multinational aid agencies^[1] with the stroke of a pen^[2] and did far more damage than EA philanthropy is able to repair.
Directionally, thats the opposite of a validation of EA orthodoxy on aid.
I don’t think neglecting the politics of whether aid actually gets disbursed is a strong argument against EA either—not least because I don’t think EAs would have been able to dissuade people from voting Trump even if they’d made it their leading cause area, or convince Trump/Musk that foreigners lives mattered—but it’s definitely not one where the “don’t neglect politics” and “actually big programs that aren’t quite as good as AMF are still really good” critics can be said to have lost the argument.
1. ^
  (programs not run by EAs, but admired by some of them for their results)
2. ^
  For added irony, the person who gleefully signed those death warrants was at least superficially EA-adjacent enough to have enthusiastically endorsed MacAskill’s writing and funded a couple of longtermist organizations in the past.

David T 8 Feb 2026 20:10 UTC
6 points
2 ∶ 0
in reply to: SiebeRozendal’s comment on: SiebeRozendal’s Shortform
It would probably be worthwhile to encourage legally binding versions of the Giving Pledge in general.
Donations before death are optimal, but it’s particularly easy to ensure that the pledge is met at that stage with a will which can be updated at the time of signing it. (I presume most of the 64% did have a will, but chose to leave their fortune to others. I guess it’s possible some fortunes inherited by widow[er]s will be donated to pledged causes in the fullness of time).
I don’t think this should replace the Giving Pledge; some people’s intentions and financial situations are too complex to write into a binding contract, but such pledges should be taken more seriously (even though in practice they are still likely to be reversible).

David T 1 Feb 2026 20:33 UTC
9 points
2 ∶ 1
in reply to: MikhailSamin’s comment on: Unfalsifiable stories of doom
Meta is paying billions of dollars to recruit people with proven experience at developing relevant AI models.
Does the set of “people with proven experience in building AI models” overlap with “people who defer to Eliezer on whether AI is safe” at all? I doubt it.
Indeed given that Yudkowsky’s arguments on AI are not universally admired and people who have chosen building the thing he says will make everybody die as their career are particularly likely to be sceptical about his convictions on that issue, an endorsement might even be net negative.

David T 1 Feb 2026 20:15 UTC
0 points
3 ∶ 2
in reply to: Nick K.’s comment on: Unfalsifiable stories of doom
The opportunity cost only exists for those with a high chance of securing comparable level roles in AI companies, or very senior roles at non-AI companies in the near future. Clearly this applies to some people working in AI capabilities research,^[1] but if you wish to imply this applies to everyone working at MIRI and similar AI research organizations, I think the burden of proof actually rests on you. As for Eliezer, I don’t think his motivation for dooming is profit, but it’s beyond dispute that dooming is profitable for him. Could he earn orders of magnitude more money from building benevolent superintelligence based on his decision theory as he once hoped to? Well yes, but it’d have to actually work.^[2]
Anyway, my point was less to question MIRI’s motivations or Thomas’ observation Nate could earn at least as much if he decided to work for a pro-AI organization and more to point out that (i) no, really, those industry norm salaries are very high compared with pretty much any quasi-academic research job not related to treating superintelligence as imminent and especially to roles typically considered “altruistic” and (ii) if we’re worried that money gives AI company founders the wrong incentives, we should worry about the whole EA-AI ecosystem and talent pipeline EA is backing. Especially since that pipeline incubated those founders.
1. ^
  including Nate
2. ^
  and work in a way that didn’t kill everyone, I guess...

David T 30 Jan 2026 21:18 UTC
24 points
9 ∶ 2
in reply to: Thomas Kwa🔹’s comment on: Unfalsifiable stories of doom
$235K is not very much money. I made close to Nate’s salary as basically an unproductive intern at MIRI.
I understand the point being made (Nate plausibly could get a pay rise from an accelerationist AI company in Silicon Valley, even if the work involved was pure safetywashing, because those companies have even deeper pockets), but I would stress that these two sentences underline just how lucrative peddling doom has become for MIRI^[1] as well as how uniquely positioned all sides of the AI safety movement are.
There are not many organizations whose messaging has resonated with deep pocketed donors to the extent that they can afford to pay their [unproductive] interns north of $200k pro rata to brainstorm with them.^[2] Or indeed up to $450k to someone with interesting ideas for experiments to test AI threats, communication skills and at least enough knowledge of software to write basic Python data processing scripts. So the financial motivations to believe that AI is really important are there on either side of the debate; the real asymmetry is between the earning potential of having really strong views on AI vs really strong views on the need to eliminate malaria or factory farming.
1. ^
  tbf to Eliezer, he appears to have been prophesizing imminent tech-enabled doom/salvation since he was a teenager on quirky extropian mailing lists, so one thing he cannot be accused of is bandwagon jumping.
2. ^
  Outside the Valley bubble, plenty of people at profitable or well-backed companies with specialist STEM skillsets or leadership roles are not earning that for shipping product under pressure, never mind junior research hires for nonprofits with nominally altruistic missions

David T 21 Jan 2026 19:57 UTC
6 points
1 ∶ 0
in reply to: huw’s comment on: Why Isn’t EA at the Table When $121 Billion Gets Allocated to Biodiversity Every Year?
Easier to persuade commercial entities of the merits of making more money (by incidentally doing the right thing) than persuade a reviewer of multiple competitive funding bids scoped for habitat preservation to fund a study into lab grown meat. At the end of the day, the proposals written by biodiversity enthusiasts with biodiversity rationales and very specific biodiversity metrics are just going to be more plausible,^[1] even if they turn out to be ineffective.
For similar reasons, I don’t expect EA animal welfare funds to award funding to an economic think tank proposing to research how to grow the economy, even if the economic think tank insists its true goal is animal welfare and provides a lot of evidence that investment in meat alternatives and enforcement of animal welfare legislation is linked to overall economic growth.
1. ^
  Biobanks and biodiversity charity effectiveness research might stand a chance, obviously

David T 19 Jan 2026 23:14 UTC
6 points
1 ∶ 0
in reply to: titotal’s comment on: Has EA given up on politics?
Also failures trying to do really outlandish things like bribing Congresspeople to endorse Jim Mattis as a centrist candidate in the 2024 US Presidential Election are likely to backfire in more spectacular ways than (say) providing malaria nets for a region with falling malaria or losing a court case against a factory farming conglomerate. That said, this criticism does apply to some other things EAs are interested in, particularly actions purportedly addressing x-risks.

David T 18 Jan 2026 15:39 UTC
3 points
0 ∶ 0
on: Everything is neglected
Feels like in the real world you describe in which few/no cause areas are actually satiated for funding, neglectedness is of interest mainly in how it interacts with tractability.
If your small amount of effort kickstarts an area of research rather than merely adds some marginal quantity of additional research or funding, you might get some sort of multiplier on your efforts, assuming others find your case persuasive. And certain problems that have being neglected due to the relative obscurity/rarity of who/what they affect might be an indication that more tractable interventions exist (if there is a simple cure for common cancers it is remarkable we have not found it yet; conversely certain obscure diseases have been the subject of comparatively little research). On the other hand, the relationship doesn’t always run that way: some causes like world peace are neglected precisely because however important they might be, there doesn’t appear to be an efficacious solution.

David T 18 Jan 2026 15:22 UTC
2 points
0 ∶ 0
on: Ancient precedents of longtermist thinking
This stands in notable contrast to most other religious and philosophical traditions, which tend to focus on timescales of centuries or millennia at most, or alternatively posit an imminent end-times scenario.
Feels like the time of perils hypothesis (and associated imperatives to act and magnitude of reward scenario) popular with longtermists maps rather more closely to the imminent end times scenario common to many eras and cultures than Buddhist views of short and long cycles and an eventual^[1] Maitreya Buddha...
1. ^
  there have also been Buddhists acting on the belief that the Maitreya was imminent or the claim that they were the Maitreya...

David T 16 Jan 2026 16:17 UTC
4 points
0 ∶ 0
on: Is rushing friendly AI our best hope of preventing “grey goo” nanotech doom?
I thought Altman and the Amodeis had already altruistically devoted their lives to saving us from grey goo. Since they’re going to do this before 2027 you may already be too late
Peter Thiel wants to know if your AI can be unfriendly enough to make a weapon out of it.

David T 9 Jan 2026 0:25 UTC
4 points
1 ∶ 0
in reply to: Lydia Nottingham’s comment on: Vitae per person (VPP): a new metric of civilizational resource
It’s a little different, but I’m not sure indexing to the consumption preferences of a certain class of US citizen in 2025 represents a better index, or one particularly close to Rawls concept of primary goods. The “climate controlled space” in particular feels particularly oddly specific (both because much of the world doesn’t need full climate control, and because 35m^2 is not a particularly “elite” apportionment of space )
To the extent the VPP concept is useful I’d say it’s mostly in indicating that no matter how much it bumps GDP per capita, AI isn’t going to automagically reduce costs of land and buildings, and is currently driving the amount of compute+bandwidth an “US coastal elite” person directly or indirectly consumes up very rapidly...

David T 8 Jan 2026 22:28 UTC
40 points
10 ∶ 2
on: Public intellectuals need to say what they actually believe
I don’t have a global audience, but if I did I wouldn’t have share this view I expressed to individuals back when COVID was first reported:
probably this isn’t going to become a global pandemic or affect us at all; but the WHO overreacting to unknown new diseases is what prevents pandemics from happening
That take illustrates two things: firstly that there are actual lifesaving reasons for communicating messages slightly differently to your personal level of concern about an issue, and secondly hunches about what is going to happen next can be very wrong.
In fact, semi-informed contrarian hunches were shared frequently and often by public intellectuals throughout the pandemic, often with [sincere] high confidence. They predicted it would cease to be a thing as soon as the weather got warmer, were far too clever to wear masks because they knew that protective effects which might be statistically significant at population level had negligible impact upon them, they didn’t have to worry about infection any more because they were using Invermectin as a prophylactic and they were keen to express their concern about vaccines^[1] Piper’s hunch is probably unusual in being directionally correct. Of all the possible cases for public intellectuals sharing everything they think about an issue, COVID is probably the worst example. Many did, and many readers and listeners died.
Being a domain expert relative to ones audience doesn’t seem nearly enough to be contradicting actual experts with speculation on health in other contexts either.^[2]
Similarly, I’m unfamiliar with Ball, but if he is “probably way above replacement level for “Trump admin-approved intellectual” he should probably try to stay in post. There are many principled reasons to cease to become a White House adviser, but to pursue a particular cause by placing less emphasis on arguments they’re might be receptive to and more on others isn’t really one of them. It’s not like theories that Open Source AI might be valuable as an alternative to an oligarchy dominated by US Americans who orbit the White House struggle to get aired in other environments. Political lobbying is the ur-case for emphasizing the bits the audience cares about, and I’m really struggling to imagine any benefit to Ball giving the same message to 80k Hours and the Trump administration, unless the intention is for both audiences to ignore him.
I haven’t read either of MacAskill’s full length books so I’m less sure on this one, but my understanding is that one focuses on various approaches to addressing poverty and one focuses on the long term, in much the same way as Famine, Affluence and Morality has nothing to say on liberating animals and Animal Liberation having little to say on duties to save human lives.^[3] I don’t think there’s anything deceptive in editorial focus, and I think if readers are concluding from reading one of those texts that Singer doesn’t care about animals or that MacAskill doesn’t care about the future, the problem with jumping to inaccurate conclusions is all theirs. MacAskill has written about other priorities since; I don’t think he owes his audience apologies for not covering everything he cares about in the same book.
I do have an issue with “bait and switch” like using events nominally about the best way to address global poverty to segueing into “actually all these calculations to save children’s lives we agonised earlier are moot; turns out the most important thing is to support these AI research organizations we’re affiliated with”^[4] but I consider that fundamentally different to focusing
1. ^
  These are just the good-faith beliefs held by intellectuals with more than a passing interest in the subject. Needless to say not all the people amplifying them had such good intentions or any relevant knowledge at all...
2. ^
  At least with COVID, public health authorities were also acting on low information. The same is not true in other cases, where on the one hand there is a mountain of evidence-based medicine and on the other, a smart more influential person idly speculating otherwise.
3. ^
  Even though Singer has had absolutely no issue with writing about some seriously unpopular positions he holds, he still doesn’t emphasize everything important in everything he writes…
4. ^
  Apart from the general ickiness, I’m not even convinced it’s a particularly good way to recruit the most promising AI researchers...

David T 28 Dec 2025 13:14 UTC
25 points
3 ∶ 0
in reply to: Richard Y Chappell🔸’s comment on: Epistemics (Part 10: The nontrivial probability gambit) | Reflective Altruism
Seems like you and the other David T are talking past each other tbh.
Above you reasonably argue the [facetious] “time of carols” hypothesis is not remotely as credible as the time of perils hypothesis. But you also don’t assign a specific credence to it, or provide an argument that the “time of carols” is impossible or even <1%^[1]
I don’t think it would be fair to conclude from this that you don’t understand how probability works, and I also don’t think that it is reasonable to assume that the probability of the ‘time of carols’ should be assumed sufficiently nontrivial to warrant action in the absence of any specific credence attached to it. Indeed if someone responded to you indirectly with an example which assigned a prior of “just 1%”, to the “time of carols”, you might feel justified in assuming it was them misunderstanding probability...
The rest of Thorstad’s post which doesn’t seem to be specifically targeted at you explicitly argues that in practice, specific claims involving navigating a ‘time of perils’ also fall into the “trivial” category,^[2] in the absence of robust argument as to why of all the possible futures this one is less trivial than others. He’s not arguing for “many gods” which invert the stakes so much as “many gods/pantheons means the possibility of any specific god is trivial, in the absence of compelling evidence of said god’s relative likelihood”. He also doesn’t bring any evidence to the table (other than arguing that time of perils hypothesis involves claims about x-risk in different centuries which might be best understood as independent claims ^[3]) but his position is that this shouldn’t be the sceptic’s job...
(Personally I’m not sure what said evidence would even look like, but for related reasons I’m not writing papers on longtermism and am happy applying a very high discount rate to the far future)
1. ^
  I think everyone would agree that it is absurd (that’s a problem with facetious examples)^[4] but if the default is that logical possibilities are considered nontrivial until proven otherwise...
2. ^
  he doesn’t state a personal threshold, but does imply many longtermist propositions dip below Monton’s 5 * 10^-16 once you start adding up the claims....
3. ^
  a more significant claim he fails to emphasize is that the relevant criteria for longtermist interventions isn’t so much that the baseline hypothesis about peril distribution is [incidentally] true but the impact of a specific intervention at the margin has a sustained positive influence on it.
4. ^
  I tend to dislike facetious examples, but hey, this is a literature in which people talk about paperclip maximisers and try to understand AI moral reasoning capacity by asking LLMs variations on trolley problems...

David T 23 Dec 2025 20:13 UTC
6 points
2 ∶ 0
on: Offering higher salaries than social-sector peers
I think this is generally true, but not sure that GiveWell is the best example of competition in social sector peers, since they seem mainly focused on hiring Western-based number crunchers who might otherwise be at a university or analysing financial data in the commercial sector, rather than poaching a small pool of educated staff from local NGOs or the best-networked, highest-performing grant fundraisers in the West.

David T 18 Dec 2025 13:50 UTC
20 points
3 ∶ 0
on: Why haven’t we seen a promising longtermist intervention yet?
I don’t think longtermism necessarily needs new priorities to be valuable if it offers a better perspective on existing ones (although I don’t think it does this well either).
Understanding what the far future might need is very difficult. If you’d asked someone 1000 years what they should focus on to benefit us, you’d get answers largely irrelevant to our needs today.^[1] If you asked someone a little over 100 years ago their ideas might seem more intelligible and one guy was even perceptive enough to imagine nuclear weapons, although his optimism about what became known as mutually assured destruction setting the world free looks very wrong now, and people 100 years ago that did boring things focused on the current world did more for us than people dreaming of post-work utopias.
To that extent, the focus on x-risk seems quite reasonable: still existing is something we actually can reasonably believe will be valued by humans in a million years time^[2] Of course, there are also over 8 billion reasons to try to avoid human extinction alive today (and most non-longtermists consider at least as far as their children) but longtermism makes arguments for it being more important than we think. This logically leads to willingness to allocate more money to x-risk causes, and consider more unconventional and highly unlikely approaches x-risk. This is a consideration, but in practice I’m not sure that it leads to better outcomes: some of the approaches to x-risk seeking funding make directionally different assumptions about whether more or less AGI is crucial to survival: they can’t both be right and the ‘very long shot’ proposals that only start to make sense if we introduce fantastically large numbers of humans to the benefit side of the equation look suspiciously like Pascal’s muggings.^[3]
Plus people making longtermist arguments typically seem to attach fairly high probabilities to stuff like AGI that they’re working on in their own estimations, which if true would make their work entirely justifiable even focusing only on humans living today.
(A moot point but I’d have also thought that although the word ‘longtermist’ wasn’t coined until much later, Bostrom and to a lesser extent Parfit fit in with the description of longtermist philosophy. Of course they also weren’t the first people to write about x-risk)
1. ^
  I suspect the main answers would be to do with religious prophecies or strengthening their no-longer-extant empire/state
2. ^
  Notwithstanding fringe possibilities like the possibility humans in a million years might be better off not existing, or for impartial total utilitarians humanity might be displacing something capable of experiencing much higher aggregate welfare.
3. ^
  Not just superficially in that someone is asking to suspend scepticism by invoking huge reward, but also that the huge rewards themselves make sense only if you believe in very specific claims about x-risk over the long term future being highly concentrated in the present (very large numbers of future humans in expectation or x-risk being nontrivial for any extended period of time might seem superficially uncontroversial possibilities but they’re actually strongly in conflict with each other).

David T 17 Dec 2025 19:26 UTC
4 points
1 ∶ 0
in reply to: Jason’s comment on: EA orgs overhire comms and underhire growth
I think it’s also more fundamental in the sense a number of EA orgs are inherently “comms-focused” because they’re lobbying for some sort of cause to some sort of decision maker (convince politicians to endorse challenge trials or ban datacentres and lead paint,, or maybe persuade fish farmers or maternal care workers in LEDCs to adopt a different approach). Or if they’re not directly lobbying they might be trying to communicate research to a relatively small group of people like computer scientists or people who want to do inter-species utility loss comparisons.
Also, with some notable exceptions I think a lot of EA is quite insular: orgs want to convey that they’re doing important work to OpenPhil funders, a pipeline of talent coming from EA groups, “aligned” organizations to collaborate with or the sort of small donor that’s already thinking about long shot solutions to x-risks or making donations to improve the welfare of unfashionable creatures. That’s a short list to a/b test, a hard group to target with paid media, and also an audience which has exacting expectations about how things are communicated, so the digital marketing to wider audience approach may not work so well. The down side is that competing for the same attention is going to usually be net less impactful than finding interest from the wider public...