David T

Karma: 470

David T 26 Mar 2024 19:46 UTC
47 points
17 ∶ 5
in reply to: Michael_PJ’s comment on: Announcement on the future of Wytham Abbey
Transparency has costs, but potentially so does opacity (in terms of both loss of trust and reduced consideration for decisions that don’t need to be justified externally). Arguably both apply here: the decision obviously wasn’t uncontroversial, and the decision making process sounds quite limited for a significant commitment in a novel area. It’s also possible some of that information that wasn’t shared or subsequent metrics (I don’t think anyone is asking for original research here) would actually cast the original decision in a more favourable light
I also think its notable EA organizations and figures have made a lot of noise criticising non-EA philanthropy for lack of transparency and rigour in the past, and a norm of exempting EA organizations from calls for such scrutiny is probably worse than oversharing. In this particular case the how effective a use of funds might it have been? question isn’t arbitrary, it’s actually core to the EA mission, and might actually be useful to the people who think there is future value in EA events space as well as the doubters.

David T 17 Apr 2024 19:03 UTC
27 points
13 ∶ 0
in reply to: Jason’s comment on: harfe’s Quick takes
I also didn’t vote but would be very surprised if that particular paper—a policy proposal for a biosecurity institute in the context of a pandemic—was an example of the sort of thing Oxford would be concerned about affiliating with (I can imagine some academics being more sceptical of some of the FHI’s other research topics). Social science faculty academics write papers making public policy recommendations on a routine basis, many of them far more controversial.
The postmortem doc says “several times we made serious missteps in our communications with other parts of the university because we misunderstood how the message would be received” which suggests it might be internal messaging that lost them friends and alienated people. It’d be interesting if there are any specific lessons to be learned, but it might well boil down to academics being rude to each other, and the FHI seems to want to emphasize it was more about academic politics than anything else.

David T 31 Mar 2024 20:36 UTC
25 points
3 ∶ 0
in reply to: David Mathers’s comment on: David Mathers’s Quick takes
“Fishing,” said the old man “is at least as complicated as any other industry”.
I was sitting in a meeting of representatives of the other end of the fishing industry: fleets of North Sea trawlers turning over >£1million each per year, fishing in probably the world’s most studied at-risk fishing ecosystem. They were fuming because in the view of the scientists studying North Sea fish, cod stocks had reached dangerously low levels and their quotas needed reducing, but in the view of the fishermen actually catching the fish, cod stocks off the east coast of England were at such high levels they hit their month’s cod quota in a day whilst actively trying to avoid catching cod. (I have no reason to believe that either view was uninformed or deceptive). “What they’re probably not factoring in,” he closed on, “is that cod populations in different regions are cyclical”
The point of that waffly anecdote is that factoring in the effects of mosquito nets on local fish ecosystems would actually be really hard, because an RCT in one area over one year really isn’t going to tell you much about the ecosystems in other areas, or in other years. Even more so in isolated African watercourses. (I think we can largely rule out the hypothesis that amateurs with free 2x2m malaria nets instead of proper nets and boats are depleting the seas more than foreign factory ships with nets hundreds of metres long...)
Like you, I don’t find the idea that ~80% of distributed nets are appropriately used (at least for the first year) to have settled the debate, but quite a few things have to be the case for AMF’s distribution programs to be the major factor in ecosystem harm caused by overfishing:
- the level of fishing with relatively small mosquito nets is sufficient to destroy fish stocks in the immediate area
- other areas are unable to replenish the fish stocks depleted in the local area
- in the absence of a continuing supply of free mosquito nets, people wouldn’t nevertheless use mosquito nets—which are also available for sale at a relatively low cost—for fishing
- the most likely alternatives to mosquito nets for local fishermen—potentially including other fabrics which also don’t allow smaller fish to pass through—don’t have the same impact on the ecosystem
for the program to be net negative for humans, you’ve also got to assume
- avoidable future falls in fishing stocks actually kill people. Or at least that the net economic loss from them is so great it outweighs the lives saved and malaria cases averted (and short term positive impact on availability of nutritious fish!). For the overall programme to be net negative we’re looking at ~24k deaths per year...
Plus of course many of the nets are distributed in regions where fishing wasn’t viable in the first place...

David T 27 Mar 2024 19:58 UTC
22 points
2 ∶ 1
in reply to: David Mathers’s comment on: [Linkpost] Leif Wenar’s The Deaths of Effective Altruism
I think his fundamental objections are philosophical (i.e. he’s more annoyed by the rhetoric of GiveWell et al than the inevitable limitations to their research). Most of the details he picks out are weaknesses GiveWell themselves highlighted and others are general foreign aid critiques, some of which apply less to small orgs distributing nets and pills than other types of aid programme and organization.
The wider idea that GiveWell and similar RCT-oriented analysis usually misses second order effects especially when undertaken by people with little experience of the developing world is valid but not novel: a more nuanced critique would note that many of these second order effects absent from GiveWell figures are positive and most of them are comparatively small. Criticising a charity evaluator for not estimating how many future deaths are likely from bandit attacks on charities’ offices is more dramatic than questioning what proportion of nets would otherwise have been sold to families by local shops anyway, but it’s not a more useful illustration of their methodological limitations.
I actually broadly agree with his general argument that EA overestimates the importance of being smart and analytical and underestimates local/sector knowledge, but I’m not sure there’s much in there that’s actionable insight (and you could probably apply half the fully general aid criticisms he applies to EA to the surfer-helping-his-friends-in-Indonesia example which he actually likes too)

David T 30 Dec 2023 16:38 UTC
20 points
3 ∶ 0
on: Public Fundraising has Positive Externalities
Good post. I think you miss the biggest benefit of all though: growing the size of the donor pie is more effective than fundraising from EA donors’ set budgets. If you’re promoting a public health charity to EA funders you’re competing for funding with the likes of AMF. If you’re promoting it to the wider public, you’re mostly competing for funding with other less effective charities or the money being spent on consumer goods.
(corollary: the efforts of a moderately efficient charity to fundraise from the general public via low-cost, viral gimmicks like ice bucket challenges can result in more net DALY gain than a more efficient charity securing funding from an EA fund. Yes, I’m inverting MacAskill’s ‘funding cannibalism’ argument… )
This benefit is likely much bigger than the positive externalities, provided you have a cause which is actually suitable for public fundraising (something like AMF has a much better chance of success than a meta charity or research org), or at least a subset of the public reasonably easily reached via interest groups (e.g “animal lovers” for campaigns against factory farming)
The feedback and “red teaming” aspect might be less useful than you think though, even for organisations that lack enough contact with the non-EA world (unless you’re counting non-EA HNWIs, corporations and funds with the time and inclination towards in-depth engagement as the general public)
Many areas of public fundraising don’t actually involve any real interaction with the public at all, so all you really get to do is A/B test your messaging.
As for actually talking to random members of the public… the feedback is probably going to disappoint, even when people are really nice. If you’re asking for non-trivial commitments, people will be in “say no” mode, which means you’ll hear quite a bit about which other charities they give to, how tight their financial circumstances are and the odd middlebrow dismissal of your pitch or the general area of charity you fit. But you won’t get any criticism of the “I spotted this error in your paper” sort that EAs seem to value most. If you’re shaking a bucket, conversations will be shorter still.
Literally talking to as many people as possible works with well organised teams on low wages and the right logistics (there’s a reason even big name charities use agencies). Even with only five new donor signups a day, low attrition rates can make it cost-effective in the long run and more so than paid media, but talking to the public all day for such a low response isn’t something it makes much sense to ask the core team (or volunteers) to be doing as anything more than an experiment. I’m speaking from experience here, having spent a couple of summers a row knocking on doors for a health-related charity with a strong local angle, and doing well enough to get invited to the meeting where all the metrics were discussed. It’s hard work and won’t work equally well for all causes
Lastly, because of all of the above, an EA central team that helps incubated causes with public fundraising strategy and implementation is probably a good idea if one doesn’t already exist...

David T 16 Mar 2024 16:57 UTC
19 points
3 ∶ 0
in reply to: Kyle Smith’s comment on: The Lack of EA in US Private Foundations
It seems like both EA and non-EA private foundations’ grantmaking is only really accountable to their funders, so I guess that also means the question has to be asked one level up: why haven’t EAs convinced more major funders of the merits of their approach?
I suspect Ian is right that EA’s approach hasn’t always been geared towards making friends. And there’s never going to be much common ground between EA and an arts endowment foundation. But I’d question whether the foundations in general felt they were being disrupted: it was mostly the major multinational charities being accused of wasting money and misleading with their marketing to the general public, whilst the grant-makers in foundations funding health/development overseas would typically be familiar with RCTs and cost benefit analysis, and possibly wishing they had the internal resource to be as thorough and as public in their evaluations as GiveWell...
(I’m also curious what sort of conversations might have taken place between the likes of GiveWell and Rethink Priorities and broadly aligned foundations at various times, if anyone can shed any light on that?)
I also suspect that (beyond direct differences with Charity Navigator) the perception of EA being more “elitist philosophy” than a rigorous approach has been shaped more by later developments: after being the disruptive upstarts that criticised incumbent orgs even in “high impact” fields for spending big on stuff unsupported by evidence like capacity building initiatives, EA orgs pivoted towards running capacity building initiatives of their own, only more internally-focused and sometimes more lavish, and the general advice for improving your chances of EA-funding became “network”. Most of the more recent criticism of EA has been of the movement rather than the methods, and it certainly hasn’t been based on the idea it’s too rigorous...
I’d be certainly be interested in more posts on what methodology other foundations use(d). I suspect a lot of them use(d) something not dissimilar to the ITN framework and/or cost/benefit analysis without the EA gloss (and sometimes within different constraints). I think there’s a lot of merit in assessing where there’s common ground and points of difference. Would be interesting to hear what motivates the transition to TBP as well, especially if it’s foundations taking a decision to reduce the amount of impact calculation they do themselves.

David T 1 Feb 2024 14:59 UTC
18 points
5 ∶ 0
on: UPDATE: Critical Failures in the World Happiness Report’s Model of National Satisfaction
This is very interesting analysis. Just in case you’re not already aware, EA has a Happier Lives Institute that’s specifically focused on research questions involving happiness and Oxford University has the Wellbeing Research Centre. I know Andrew Oswald’s been researching socioeconomic factors behind national happiness measures for at least a couple of decades.
By way of critique, I agree education is a particularly surprising omission from the Gallup survey (particularly when they chose to include a measure of individual generosity) and suspect you’re right that they over-weight GDP per capita (and GNI is more appropriate), though I don’t think they claimed their factors were definitive. It appears your variables actually overlap quite a bit with theirs (e.g. the binary question “Are you satisfied or dissatisfied with your freedom to choose what you do with your life?” to a significant extent encompasses LGBT rights and female empowerment and to a lesser extent general political freedoms), but there’s definitely benefit to disaggregating GDP/GNI and its positive side effects like being able to afford clean water, safe cooking etc. and some very logical additions like unemployment. I’m also curious whether your “social support” variable was similar to Gallup’s (e.g. answers to a question about personal networks specifically) or whether it encompassed other stuff (e.g. welfare state)?
The challenge with throwing up to 926 noisy and often highly-correlated variables at a relatively small panel data set is that some of the improvements in fit are likely to be spurious, particularly when it leads to odd conclusions like only HIV infections in one particular demographic affecting overall wellbeing. I’m not sure the best performing variables in statistical testing are actually those with the strongest causal relationship.
I’d be particularly cautious of drawing policy conclusions like introducing public campaign financing could be a cheap happiness win from it because I’m not sure there’s any causal relationship with that one at all. I thought it might be a simple case of most countries scoring zero and a small number of particularly well-functioning democracies having positive scores (which I’d still say was likely to be correlation not causation) but actually looking at the original data for the latest year it’s a continuous series of weighted scores that.… make very little sense (this surprised me since V-DEM is a meticulously researched and respected dataset, but they don’t use this index in any of their composite indices...). Sure, there are a few well-functioning democracies at the top and an absolute monarchy at the bottom, but in between it’s just bizarre: Switzerland has one of the worst scores, North Korea is in the middle (I suppose the Dear Leader does depend on state funding for his political power...), Cuba does pretty well compared with the UK etc. Why this very noisy series fits the regression better than dozens of other presumably more representative ones I don’t know, but I don’t think it’s anything to do with political races in North Korea being fairer!

David T 17 Apr 2024 19:44 UTC
17 points
2 ∶ 2
in reply to: Wei Dai’s comment on: What should the EA community learn from the FTX / SBF disaster? An in-depth discussion with Will MacAskill on the Clearer Thinking podcast
I don’t think the “3% credence in utilitarianism” is particularly meaningful; doubting the merits of a particular philosophical framework someone uses isn’t an obvious reason to be suspicious of them. Particularly not when Sam ostensibly reached similar conclusions to Will about global priorities, and MacAskill himself has obviously been profoundly influenced by utilitarian philosophers in his goals too.
But I do think there’s one specific area where SBF’s public philosophical statements were extremely alarming even at the time, and he was doing so whilst in “explain EA” mode. That’s when Sam made it quite clear that if he had a 51% chance of doubling world happiness vs a 49% of ending it, he’d accept the bet.… a train to crazytown not many utilitarians would jump on and also one which sounds a lot like how he actually approached everything.
Then again, SBF isn’t a professional philosopher and never claimed to be, other people have said equally dumb stuff and not gambled away billions of other people’s money, and I’m not sure MacAskill himself would even have read or heard Sam utter those words.

David T 22 Jan 2024 15:47 UTC
17 points
5 ∶ 1
in reply to: Habryka’s comment on: Report on the Desirability of Science Given New Biotech Risks
I’m not sure that choice is a real one humanity actually faced though. It seems unlikely that humans alive 10,000 years ago actually had the capability to commit omnicide, still less the ability to avert future omnicide for the cost of lunch. It’s not a strong reductio ad absurdum because it implies a level of epistemic certainty that didn’t and doesn’t exist.
The closest ancient-world analogue is humans presented with entirely false choices to sacrifice their lunch to long-forgotten deities to preserve the future of humanity. Factoring in the possible existence of billions of humans 10,000 years into the future wouldn’t have allowed them to make decisions that better ensured our survival, so I have absolutely no qualms with those who discounted the value of our survival low enough to decline to proffer their lunch.
Even if humanity 10000 years ago had been acting on good information (perhaps a time traveller from this century warned them that cultivating grasses would set them on a path towards civilization capable of omnicide) rather than avoiding a Pascal-mugging, it’s far from clear that humanity deciding to go hungry to prevent the evils of civilization from harming billions of future humans would [i] not have ended up discovering the scientific method and founding civilizations capable of splitting atoms and engineering pathogens a bit later on anyway [ii] have ended up with as many happy humans if their cultural taboos against civilization had somehow persisted. So I’m unconvinced of a moral imperative to change course even with that foreknowledge. We don’t have comparable foreknowledge of any course the next 10000y could take, and our knowledge of actual and potential existential threats gives us more reason to discount the potential big expansive future even if we act now, especially if the proposed risk-mitigation is as untenable and unsustainable as “end science”.
If humanity ever reached the stage where we could meaningfully trade inconsequential things for cataclysms that only affect people in the far future [with high certainty], that might be time to revisit the discount rate, but it’s supposed to reflect our current epistemic uncertainty.

David T 10 Jan 2024 23:38 UTC
15 points
1 ∶ 0
on: Copenhagen Consensus Center’s best investment papers for the sustainable development goals
Interesting list! I had a quick look at the papers for three areas not covered by major EA funders: my thoughts below
e-Government procurement—IT systems to manage government tenders/purchases—making it easy for private sector companies to sell to governments, reducing prices through increased competition and reducing opportunities for corruption.
Looks very attractive because of a high benefit to cost ratio and relatively tractable problem,
- benefit-to-cost ratio could be as low as 8 for small, low income countries like Liberia and over 300 for a mid size, lower middle income country—a pretty wide range!
- The paper estimates that in the long-term implementing across half of all government procurement saves an average of 6.75% on prices paid by government for goods and services, providing some evidence from real world programs.
- Their estimated implementation costs look very low for complex projects (but on the other hand they also don’t estimate costs of manual processes replaced, which might balance out somewhat)
The problem isn’t as tractable as it first looks because you have to convince (many levels of) the governments to change their ways. Theoretical ways to tackle:
- advocacy promoting move to e-procurement
  - but IT firms capable of delivering already commercially incentivised to do this
  - and governments with pervasive corruption incentivised to ignore!
- fund development of “best practice” IT procurement solutions
  - but in practice the IT solution isn’t the hard or expensive bit; it’s the integrating with local laws and language, and getting govt departments to adopt it
- capacity building and upfront funding for bodies trying to deliver.
  - if you believe the paper’s numbers, could be funded for as little as $22m over 12 years (and may even repayable in later years based on KPIs)
  - Not a traditional EA opp, but paying a government to implement an IT system to save themselves money is the sort of thing that’s so counterintuitive maybe only an EA org would consider it!
The financial side stacks up on paper, but the net impact in terms of things EAs care about like utility is less obvious. Reduced government procurement costs can allow them to do more for less, but it also means local firms earning less. Not all cost savings will be due to corruption and not all of the money saved will be spent well.
Land tenure security—creating a registry of land ownership—making it easier for new landowners to sell or borrow against the value of their land, and enabling for them to spend money improving it without the risk of being dispossessed.
- The actual mechanisms by which householders and farmers benefit varies between studies and countries, with some finding no effect
  - The benefits take some time to accrue (the paper assumes typically 6 to have any effect and estimates the NPV of benefits over 30 years) and meta analysis suggests timing may explain some of the “no effect” results
- EA organisations could promote the idea, or even subsidise it, but any actual action is contingent on government action
  - This idea is well known to development economists and has been heavily promoted by de Soto (I’m amazed the paper didn’t namecheck or cite him) in Peru and various other places, so I’m not sure how much EA advocacy would add.
  - Organisations like the World Bank have supported land registration projects financially and there is some scope for a privately-funded NGO like De Soto’s ILD to actually do the hard work of identifying informal landownerss with the local government’s blessing
  - The paper assumes a surprisingly high cost per site to maintain the land registry, so it’s not a one-time intervention
Trade—reducing tariffs and trade barriers
- Free(r) trade can only be implemented by national governments. The cost/benefit analysis is looking at the costs to negatively affected industries vs the benefits to everyone else across multiples countries, so it’s not comparable to GiveWell’s estimates
- The idea has been promoted/debated by economists since the 1700s and is promoted by multilateral bodies like the WTO/World Bank/IMF, so it’s not really clear what EA advocacy would add.

David T 19 Mar 2024 20:10 UTC
14 points
4 ∶ 0
in reply to: Rohin Shah’s comment on: EA “Worldviews” Need Rethinking
I also think it misses the worldview bucket that’s the main reason why many people fund global health and (some aspects of) development: intrinsic value attached to saving [human] lives. Potential positive flowthrough effects are a bonus on top of that, in most cases.
From an EA-ish hedonic utilitarianism perspective this dates right back to Singer’s essay about saving a drowning child. Taking that thought experiment in a different direction, I don’t think many people—EA or otherwise—would conclude that the decision on whether to save the child or not should primarily be a tradeoff between the future capacity of the child and the amount of aquatic suffering a corpse to feed upon could alleviate.
I think they’d say the imperative to save the child’s life wasn’t in danger of being swamped by the welfare impact on a very large number of aquatic animals or contingent on that child’s future impact, and I suspect as prominent an anti-speciesist as Singer would agree.
(Placing a significantly lower or zero weight on the estimated suffering experienced by a battery chicken or farmed shrimp is a sufficient but not necessary condition to favour lifesaving over animal suffering reduction campaigns. Though personally I do, and actually think the more compelling ethical arguments for prioritising farm animal welfare are deontological ones about human obligations to stop causing suffering)

David T 22 Apr 2024 12:30 UTC
13 points
4 ∶ 0
in reply to: MathiasKB’s comment on: Should we break up Google DeepMind?
I also think that it’s far from given that the option which would minimise consumer harm from monopoly would also minimise pressure to race.
An AI research institute spun off by the regulator under pressure to generate business models to stay viable is plausibly a lot more inclined to ‘race’, than an AI research institute swimming in ad money which can earn its keep by incrementally improving search, ads and phone UX and generating good PR with its more abstract research along the way. Monopolies are often complacent about exploiting their research findings, and Google’s corporate culture has historically not been particularly compatible with launching sort of military or enterprise tooling that represents the most obviously risky use of ‘AI’.
There are of course arguments the other way (Google has a lot more money and data than putative spinouts) but people need to predict what a divested DeepMind would do before concluding breaking up Google is a safety win.

David T 27 Mar 2024 22:10 UTC
13 points
6 ∶ 0
in reply to: Rebecca’s comment on: [Linkpost] Leif Wenar’s The Deaths of Effective Altruism
The “deaths caused” example picked was pretty tendentious. I don’t think it’s reasonable to consider an attack at a facility by a violent criminal in a region with high baseline violent crime “deaths caused by the charity” or to extrapolate that into the assumption that two more people will be shot dead for every $100,000 donated. (For the record, if you did factor that into their spreadsheet estimate, it would mean saving a life via that program now cost $4776 rather than $4559)
I would expect the lives saved from the vaccines to be netted out against deaths from extremely rare vaccine side effects (and the same with analysis of riskier medical interventions), but I suspect the net size of that effect is 0 to several significant figures and already factored into the source data.

David T 27 Mar 2024 21:29 UTC
12 points
3 ∶ 0
in reply to: David Mathers’s comment on: [Linkpost] Leif Wenar’s The Deaths of Effective Altruism
I think his argument is mainly “aid is waaayyyy more unpredictable and difficult to measure than your neat little tables crediting yourselves with how efficient you are at saving lives suggest”, with GiveWell ironically getting the biggest bashing because of how explicit they are about highlighting limitations in their small print. Virtually all the negative side effects and recommendation retractions he’s highlighted come straight from their presentations of their evidence on their website. He’s also insistent they need to balance the positives of lifesaving against harm from nets being redeployed for fishing, but ironically the only people I’ve seen agree with him on that point are EAs.
It’s less an argument they’re not properly accounting for stuff and more that the summaries with the donation button below sound a lot more confident about impact than the summaries with the details for people that actually want to read them. I’m reminded of Holden’s outspoken criticism of big NGOs simplifying their message to “do x for $y per month” being “donor illusion” back in the day…

David T 29 Jan 2024 19:49 UTC
12 points
2 ∶ 0
in reply to: Joseph_Chu’s comment on: Joseph_Chu’s Shortform
As you’ve pointed out, the questions are very different. The Gallup Poll asks people to rank their current position in life from “the best possible” to the “worst possible” on a ten point scale which implies that unequal opportunities and outcomes matter a lot.

The IPSOS poll avoids any sort of implicit comparison with how much better things could otherwise have been or actually is for others, and simply asks whether they would describe themselves as (very) happy or not (at all) on a simpler 4 point scale which is collapsed to a yes/no answer for the ranking

So Chinese and Indian people aren’t being asked whether they’re conscious of the many things they lack which could make their life better like in the Gallup poll, they’re being asked whether they feel so bad about their life they wish to describe themselves as unhappy (or, for various other questions “unsatisfied”). People tend to be biased towards saying they’re happy and there’s likely to be a cultural component to how willing people are to say they’re not too

And to add to the complications, the samples are non-random and not necessarily equivalent. IPSOS acknowledge their developing country samples are significantly more affluent, urban and educated than the population, which might explain why even when it comes to their personal finances they’re often more “satisfied” than inhabitants of countries with much higher median incomes. Gallup doesn’t admit that sampling bias, but even if it’s present to exactly the same extent (it’s bound to be present to some extent; poor, rural illiterate people are hard to randomly survey) it probably doesn’t have the same effect. Indian professionals can simultaneously be “happy” with their secure-by-local standards position in life and aware that their life outcomes could have been a whole lot better.

Think the stark differences are a good illustration of the limits to subjective wellbeing data, but arguably neither survey captures SWB particularly well anyway, the former because it asks people to make a comparison of [mainly objective] outcomes and the latter because the scale is too simple to capture hedonic utility.

David T 10 May 2024 12:01 UTC
11 points
0 ∶ 0
in reply to: Hauke Hillebrandt’s comment on: Probably Good launched a new job board!
Whilst I sympathise with the desire to see more of this kind of information particularly given EA jobs being notoriously competitive, I’d be concerned that the signals sent out by raw numbers might be misleading and deter suitable applicants
The classic example is LinkedIn, which does display applicant numbers. Having seen the other side of LinkedIn job ads, I’m well aware that a job with 30+ applicants probably has about 25 who one-click apply to everything vaguely related to their field even when lacking basic qualifying criteria such as visa status. If I hadn’t seen that side of things, I’d probably be deterred from applying based on not meeting a bullet point or two where actually I’d probably be in the top 10% of qualified candidates.
What I think would be valuable to some people is those organizations with relatively complex processes involving exercises and application forms choosing to indicate roughly how many people completed exercises for similar jobs in the past (as a marginal candidate I’m a lot more likely to fancy my chances of standing out if it’s 10 than if it’s 70), but that’s a helpful thing before people devote considerable time to a process rather than something to search for.

David T 13 Mar 2024 11:56 UTC
10 points
1 ∶ 0
on: The RCT Agenda—Bryan Caplan
I think it’s much more true that non-market solutions—or indeed anything that doesn’t look like unequivocal advocacy for markets—aren’t given enough consideration by Bryan Caplan!
One of his own examples of RCTs is the most market-based hypothesis imaginable: testing why poor Kenyans won’t buy a commercial product that saves them money and concluding they’re much more likely to if they have access to credit! The implicit conclusion of an intervention failing a test of cost-effectiveness is in fact, for the government (or private funders) to spend less money on it. And many RCTs—particularly under the auspices of EA—are funded by private individuals looking to allocate their own market-generated money according to their own goals. Some of them, like GiveDirectly’s, are based on hypotheses that market participation absolutely does offer people in poverty more choice about how to escape than centrally planned interventions.
There are valid criticisms of RCTs as reductionist and missing bigger pictures, but the better ones don’t argue microeconomists do empirical microeconomics research out of hidden anticapitalist agendas!

David T 11 May 2024 17:30 UTC
9 points
3 ∶ 0
in reply to: DavidNash’s comment on: My Lament to EA
I think it’s inaccurate that only people at top universities are likely to have outsized influence, or to dismiss everyone else as “poorer students” that it “doesn’t make as much sense” to encourage to engage in altruistic activity. The university Sorting Hat really isn’t that good.
And more specifically from a movement building perspective it usually makes sense to prioritise reaching more people than to ensure a small group of [already advantaged] people have access to particularly lavish allowances. Students at elite universities’ ability to achieve outsized impact later in life probably isn’t particularly closely linked to the size of the stipend the current organizer of their well-established EA group is able to claim from central funding bodies, whereas actually having some outreach at other universities is going to have more impact, even if fewer of those students’ impact will be outsized and the median earn to give amounts might be a little lower.
Edit: not really sure what’s so controversial here, though I’ve amended the quote just in case it’s because my representation of DavidNash’s original comment was considered uncharitable.

David T 10 Apr 2024 11:49 UTC
9 points
3 ∶ 0
in reply to: Ebenezer Dukakis’s comment on: David Mathers’s Quick takes
Even the Hanania article you linked to entitled “Diversity Is Our Strength” contains as one of its core arguments the suggestion that Hispanic immigrants might be won over to his support for “war with civil rights law” by “comparing them favorably to genderfluid liberals and urban blacks”.
The next sentence links to one of his own tweets about how “selling immigrants on hating liberals would be the easiest thing in the world”, featuring a video of Muslims protesting in favour of LGBT book bans.
Perhaps you don’t find this style of politics repugnant, perhaps it even represents a marginal improvement on his prior beliefs, but I don’t think it’s one EA should be endorsing.

David T 1 Jan 2024 18:14 UTC
9 points
1 ∶ 3
in reply to: Karthik Tadepalli’s comment on: Stop talking about p(doom)
Or to put more bluntly, the p(doom) estimates tended to rise after the “10^35 future humans so even if p(doom) is really low...” arguments were widely dismissed.
Obviously other stuff happened in the world of AI, and AGI researchers are justified in arguing they simply updated their priors in the light of the rise of emergent behaviour exhibited by LLMs[1] (others obviously always had high and near-term expectations of doom anyway). But Bayes’ Theorem also justifies sceptics updating their p(blackmail) estimates[2]
Priors for regular events like sports, market movements and insurable events are easily converted into money so bold predictions are easily put to the test whether they claim to be based on a robust frequentist model of similar events or inside information or pure powers of observation. But doom in most of the outlined scenarios is a one-off and the incentive structure actually works the opposite way round: people arguing we’re seriously underestimating p(doom) don’t expect to be around if they’re right and are asking for resources now to reduce it. I don’t think it’s an isolated demand for rigour to suggest that a probabilistic claim of this nature bears very little resemblance to a probabilistic claim made where there’s some evidence of some base rate and strong incentive not to be overconfident.
So yeah, I agree, p(doom) isn’t persuasive and I’m not sure decomposing it into p(doom | agi) and p(agi) or equivalents for other x-risk fields puts it on a stronger footing. Understanding how researchers believe a development increases or reduces a source of x-risk is much more convincing argument about their value than incrementing or decrementing an arbitrary-seeming doom number. The “doomsday clock” was an effective rhetorical tool because everyone understood it as asking politicians to reverse course, not because it was accepted as a valid representation of an underlying probability distribution.
[1]though they could also have been justified in updating the other way; [notionally] safety-conscious organisations getting commercially valuable, near-human level outputs from a text transformation matrix arguably gives less reasons to believe anyone would deem giving machines agency worth the effort.
[2]also in either direction.