That seems true for most things EAs fund apart from direct service delivery interventions such as distributing malaria nets.
I.e. it is a valid consideration but it is not a justification to work on surgical instead of systemic interventions in areas where all interventions are operating uncertainly over multi-year indirect theories of change (the majority of what EAs do outside GiveWell-style GHD work).
Yes, I saw this and was happy for it to exist.
What I am trying to say is that this being one of the longest treatments on this to exist feels like a failure / blind spot of the community.
We’re in the midst of very severe systemic changes, domestic and international, and—ideally—there’d be lots of thorough analysis on the forum and elsewhere.
Thanks! I don’t think that hard-to-measure explanation is quite right—lots of other similarly speculative / hard-to-measure interventions that EAs have been traditionally very excited about.
I think it has more to do with priors of low neglectedness and low tractability and a certain aversion to act in ways that could be seen as political.
That said, my goal here is not to re-litigate the whole “surgical v systemic change” debate, but rather to say that current changes seem to suggest that systemic work should be relatively more important and it’s something that seems (vastly) under-discussed and not systemically explored.
Great piece!
You can always replace chicken meat with plant-based foods for health reasons, or if you are very concerned about GHG emissions. I am not
I think signaling that you don’t think GHG emissions are important does not help your message here / makes this less convincing that it would otherwise be!
I think we are relatively close and at the risk of misunderstanding.
I am not saying psychology isn’t part of this and that this work isn’t extremely valuable, I am a big fan of what you and Stefan are doing.
I would just say it is a fairly small part of the question of collective decision making / societal outcomes, e.g. if one wanted to start a program on understanding decision making in key GCR areas better then what I would expect in the next sentence would be something like “we are assembling a team of historians, political scientists, economists, social psychologists, etc.” not “here is a research agenda focused on psychology and behavioral science.” Maybe psychology and behavioral science were 5-20% of such an effort.
The reason I react strongly here is because I think EA has a tendency to underappreciate social sciences outside economics and we do so at our own peril, e.g. it seems likely that having more people trained in policy and social sciences would have avoided the blindspot of being late on AI governance, for example.
Critical decisions about advanced technologies, including artificial intelligence, pandemic preparedness, or nuclear conflict, as well as policies shaping safety, leadership, and long-term wellbeing, depend on human psychology.
I am surprised by this. Ultimately, almost all of these decisions primarily happen in social and institutional contexts where most of the variance in outcomes is, arguably, not the result of individual psychology but of differences in institutional structures, culture, politics, economics, etc.
E.g. if one wanted to understand the context of these decisions better (which I think is critical!) shouldn’t this primarily motivate a social science research agenda focused on questions such as, for example, “how do get decisions about advanced technologies made?”, “what are the best leverage points?” etc.
Put somewhat differently, insofar as it a key insight of the social sciences (including economics) that societal outcomes cannot be reduced to individual-level psychology because they emerge from the (strategic) interaction and complex dynamics of billions of actors, I am surprised about this focus, at least insofar as the motivation is better understanding collective decision-making and actions taken in key GCR-areas.
Worth pointing out that FP staff who could reply on this is on Thanksgiving break, so will probably take until next week.
So, he did some bad things but it was around expectation and nothing yet in the tails and thus I shouldn’t update in the direction of totalitarianism.
No one speaks of totalitarianism here, but a risk of authoritarian drift.
Over the past two weeks, the President-Elect has indicated he wants to appoint extreme loyalists without substantive qualifications to positions most relevant for democracy working well (or not) (DOJ, DOD, etc.). He is also trying to weaken the power of the Republican Senate Majority, both via the threat of recess appointments plus generally by pushing the Senate to confirm unqualified candidates.
I don’t think anyone knows what will happen, but I think being confident that he is not doing anything in the tails seems overconfident, what he is doing now is exactly what one would be doing if one wanted to move towards more authoritarianism.
I was convinced of this for the election and for election integrity efforts around the election. I am less convinced this holds for the work now that is comparatively longer, less flashy, and costlier.
The worry that people have with Gabbard is that her sympathies with Russia and Syria would essentially make it hard for her own and allied intelligence services to trust her severely undermining the function she is to serve.
This strikes me as too optimistic/not taking the evidence from the last two elections seriously enough.
In both of them a leading contender for the Presidency did not commit to a peaceful transfer of power. In one of them he incited an insurrection. In both of them, more severe outcomes were prevented by contingent factors.
Thanks!
I think if we optimize for short-term impact we would want a list that focuses on short-lived pollutants (e.g. methane) or short-term adaptation measures.
I think the weakness of the article for EA prioritization is that it optimizes for something—domestic certain reductions within countries—that is not related to any globally relevant target metric.
E.g. irrespective of whether one optimizes for the short-term or long-term in neither scenario will the focus on national target achievement be relevant directly (it might matter somewhat indirectly via signaling). Obviously, it is a good article for national policy makers that want to achieve national targets.
Thanks!
I agree that policies can be combined but that doesn’t really change the fundamental problem that evaluating policies for their certain local short-term impact will move us away from more impactful policies.
This is good raw data, but I don’t think one can learn very much from this for what actions to prioritize for a couple of reasons:
(1) The focus is on national targets, while the most transformative policies in the past have had massive global impacts long-term with minimal local effects in the short-term (cf. early solar and wind policies, many of which were not even motivated as climate policies). Indeed, there is often a negative correlation between short-term domestic and long-term global impacts (I explore this a bit here).
(2) What matters for taking action is decarbonization returns on additional effort (activism, philanthropy, policy momentum), which could be quite different from absolute emissions reductions if implemented (similar to what Habryka is commenting on).
(3) In particular, this biases against relatively modest looking interventions that can be transformative (often have been in the past) such as innovation-support for nascent and early-stage technologies, where carbon pricing does ~nothing, but R&D expenditure, creating market demand, public co-financing of demonstration projects etc. make a large difference.
(4) Somewhat less importantly, the authors have a well-known focus on advancing carbon pricing, so this is something to keep in mind.
It turned out to be a very easily solvable problem (heavily concentrated industry, very few production plants, technological substitution being very easy).
I think the Montreal Protocol is often misunderstood to be a great success story / positively indicative reference class for much harder problems like climate, AI, etc., biosecurity, when it really is an example of an exceptionally easy problem to solve.
Apologies for my delay here.
There is indeed no contradiction—solar got cheap through massive public support, first mostly R&D and later deployment subsidies / market creation policies in the hundreds of billions.
So the lesson from solar is definitely that public innovation support massively matters, it is more that different forms of support are most critical at different times (that is something the Kavlak paper emphasizes, how at early TRL R&D dominates and then later induced demand becames the major source of cost reduction) and that the R&D money cited is a small contributor to the cost reductions observed then.
If APs were like solar, I think we should expect things to take a lot longer and require a lot more support and maybe the current plateau would be like the 1980s for solar. (But I think there are good reasons to be more optimistic).
Apologies for the delay, Remmelt.
What I meant to say in my original comment is that when someone who has a bad reputation for being truth seeking / has a reputation for being a conspiracy theorist says something on an area outside their expertise we should not give a lot of credence to it.
The burden of this not being a credible scenario should—in my view—not lie with commenters and, at the time I was commenting, most comments seemed incremental instead of pointing to a strong skepticism on Weinstein’s take to start as if it were a plausible situation that a guy on a podcast without specific relevant expertise has uncovered a mechanism that 50 years of anti-nuclear activism have not produced (or, more nuanced, something that is so implausible that even typical anti-nuclear advocates wouldn’t lead with it because they know it won’t stand).
To the specific question—we do have a long section on nuclear in our 2018 report and I’ve obviously also spent a lot of time on it though I haven’t written it up, but it took me several years to turn from German anti-nuclear environmentalist to the positions I hold now.
Here is an article on why a Chernobyl-style event is not possible with modern reactors. And the diesel generators would not be refueled for a year, given in the scenario there would be a failure of the entire power grid you would just use them to have enough electricity to shut things down safely.
I’ll address the rest later (hopefully this weekend), but just to clarify that I did not mean to call you or your summary conspiratorial, Remmelt!
I only meant the substantive claim and coming from Bret Weinstein, but I’ll address this with more time and the rest of your comment later.
Great stuff!
I am not sure how crucial this is, but I do think this piece gets some of the renewable energy analogies importantly wrong. In particular:
1. The primary drivers of making solar cheap where policies passed in the early 2000s, in particular massive deployment subsidies, not primarily R&D. So, the 30B R&D number for 2004-2012 are unlikely to be a major driver of the dramatic cost reductions in solar, at the same time countries like Germany spent 100s of B on subsidizing solar deployment (see e.g. Kavlak et al 2018, also Nemet’s book)
2. By 2010, while solar was still marginal, the cost reduction dynamic was largely triggered through those policies and the virtuous cycle of investment, economies of scale, and China becoming interested in becoming a renewables manufacturing powerhouse. It is not driven by 2004-2012 R&D and it is not really driven by things that started after 2010, more a self-reinforcing dynamic playing out.
Hopefully and plausibly, APs are an easier technology to get cheap than solar panels!
(3. Nitpick: Solar is 4% of electricity, not energy, electricity is only ~1/3 of energy.)
I’ve now tried to clarify what I mean in my post, Nick.
I agree with you that concrete suggestions are lacking, my claim is that this is—at least partially—due to too little effort on this angle and that this seems worth re-examining in a change of rapid and profound system-level changes.