Error
Unrecognized LW server error:
Field "fmCrosspost" of type "CrosspostOutput" must have a selection of subfields. Did you mean "fmCrosspost { ... }"?Unrecognized LW server error:
Field "fmCrosspost" of type "CrosspostOutput" must have a selection of subfields. Did you mean "fmCrosspost { ... }"?
A piece such as this should engage with the direct cost/benefit calculations that have been done by economists and EAs (e.g. Giving What We Can), which make it seem hard to argue that climate change is competitive with global health as a cause area.
How much it would take to stay under a mostly arbitrary probability of a mostly arbitrary level of temperature change is a less relevant statistic than how much future temperatures would change in response to reduced emissions.
Okay, I’ve just posted an analysis of the four relevant impact/cost-effectiveness estimates that I’m aware of. You can see my conclusions here—https://forum.effectivealtruism.org/posts/ynRG6JBvARS2cHu63/review-of-climate-cost-effectiveness-analyses
I’ll definitely take a look at the cost effectiveness calculations and see if I can work references to these into my draft. In particular, I’m interested to find out what assumptions they are based on.
The other blog post you shared looks to me to have a key flaw—it models emissions as having a sharp spike where they go from growing quickly to declining quickly. This seems very unlikely to me—and the smoother curve as growth slows and turns into decline implies a greater area under the curve and hence a much greater final impact of delay.
Thanks for this. I found it interesting to think about. Here are my main comments.
Mainline and extreme risks
I think it would be better to analyse mainline risks and extreme risks separately.
Depending on whether or not you put substantial weight on future people, one type of risks may be much more important than the other. The extreme risks appear to pose a much larger existential threat than mainline risks, so if you value future generations the extreme risks may be much more important to focus on. The opposite may be true for people who apply high pure time discounting. (Relatedly, on most worldviews, the ‘scale’ factor will be materially different between the two.)
The ideal responses to mainline and extreme risks appear to be different. (Relatedly, the ‘solvability’ of those responses may differ, as might the amount of resources that are already committed to the relevant kinds of responses.)
The methodologies which are useful are different. Efforts to understand and act on mainline risks are amenable to standard economic approaches, while extreme risk analysis requires substantive judgements about empirical and model uncertainty.
Overall, putting the two together seems to make the analysis less clear, not more.
More expensive things are worse
The framework 80k uses is designed to add up to a cost-effectiveness heuristic. Adjusting this by giving more expensive things higher neglectedness scores in effect takes the ‘cost’ out of the ‘cost-effectiveness analysis’. Using a completely different framework would be fine, but making this adjustment alone causes one to depart from any notion of good done per effort put in.
If I put more time into this, I would focus on the solvability part
In your initial comments on solvability, you give a concrete set of interventions which you say would cost approximately a certain amount and achieve approximately a certain amount. If someone were to analyse these in detail, this could be the basis for a cost-effectiveness calculation. Of course, I don’t know enough to say whether the Project Drawdown analysis you reference is accurate. Other people looking into this might want to focus on that since it seems crucial to the bottom line.
Qualified ’need’s
When you just say ‘we need to do Y’, this seems to be sort of assuming the conclusion, adding little to my understanding of which actions produce the most impact. For example:
I found it much more helpful when you said ‘if we want to achieve X, we need to do Y’, improving my understanding of what actions lead to what effects. For example:
This part of my comment might sound like a nitpick, but I think attention to this kind of thing can make for better analysis and better communication.
All personal views only, of course.
I’m curious to know what you think the difference is. Both problems require greenhouse gas emissions to be halted.
The neglectedness guidelines focus on the level of existing funding. I argue that this is an insufficient view—that if you have two problems who require $100 or $200 of total funding to solve completely, if they both have $50 of funding today, they are not equally neglected. The denominator matters—the $200 problem is much further from being solved.
Perhaps I’m proposing a slightly different framework—but it’s definitely not one divorced from the notion of caring about the good done per effort put in. I just don’t believe that climate change is really at saturation point for the level of effort.
Fair point about use of language. I’ll try and address this in a future edit.
I agree that both mainline and extreme scenarios are helped by reducing greenhouse gas emissions, but there are other things one can do about climate change, and the most effective actions might turn out to be things which are specific to either mainline or extreme risks. To take examples from that link:
Developing drought-resistant crops could mitigate some of the worst effects of mainline scenarios, but might help little in extreme scenarios.
Attempting to artificially reverse climate change may be a last resort for extreme scenarios, but may be too risky to be worthwhile for mainline scenarios.
For the avoidance of doubt, I think that my point about mainline and extreme risks appealing to different worldviews is sufficient reason to separate the analyses even if the interventions ended up looking similar.
Yep, you could use the word ‘neglected’ that way, but I stand by my comment that if you do that without also modifying your definition of ‘scale’ or ‘solvability’, the three factors no longer add up to a cost-effectiveness heuristic. i.e. if you formalise what you mean by neglectedness and insert it into the formula here without changing anything else, the formula will no longer cancel out to ‘good done / extra person or $’.
Many props for doing this. This is exactly the sort of careful critique I’d like to see more people generating.
Did you approach 80,000 Hours about your post before putting it up? If you didn’t, it seems quite plausible that they’d have integrated some of these changes after speaking with you directly. The benefits of approaching them first are that it appears less adversarial and (less certain) they are more likely to notice. However, I think there are also arguments for publishing publicly and independently, too.
I notified info@80000hours.org about this public post. I thought it would be better to solicit public feedback rather than attempting to work privately with 80K Hours.
Hi mchr3k — thanks for writing this. I’m completely slammed with other work at 80,000 Hours just now (I’m recording 7 podcast interviews this month), so I won’t be able to respond right away.
For what it’s worth I agree with just posting this and emailing it to us, rather than letting us hold you up. Many people are going to be interested in what you’re saying here and might have useful comments to add, not just 80,000 Hours. It’s also an area where reasonable people can disagree so it’s useful to have a range of views represented publicly.
Possibly letting us comment on a Google Doc first might have been helpful but I don’t think people should treat it as a necessary step!
Just commenting here to say that, as a Forum moderator, I love everything about this series of interactions! Props to Roxanne for asking, mchr3k for having notified 80K, and Rob for posting his thoughts on the situation.
In addition to direct cost-effectiveness calculations for the present generation as StevenKaas recommends, I would recommend direct cost-effectiveness calculations for the long-term future. Here is an example where I compare AI and alternative foods to address agricultural catastrophes. It would not take very much work to use that framework for conventional emissions reductions for climate change. However, as others have pointed out, because emissions reductions are so expensive, they are unlikely to be competitive cost-effectiveness. Solar radiation management (SRM) (as opposed to non SRM geoengineering techniques such as CO2 air capture) has the potential of being much more cost effective, but it has its own risks, such as double catastrophe.
Seconding StevenKaas. In particular, there’s this recent post: https://forum.effectivealtruism.org/posts/GEM7iJnLeMkTMRAaf/updated-global-development-interventions-are-generally-more
Global health is a 21 on 80,000 Hours’ list, and you’re recommending updating climate change to 26. This suggests it’s several orders of magnitude better to spend resources to fight climate change than to donate those resources to GiveDirectly, say. This seems to contradict the conclusion of the linked EA forum post.
You also mention donating to charities like Cool Earth. It would be good to know how they compare to GiveWell’s recommended charities.
Thanks! Cool perspective. I’ll just make a quick comment on the neglectedness point:
Much of diminishing returns in an area come not from crowdedness or inefficient economies of scale but because the low hanging fruit are gone. Most EA top reccommended cause areas are so neglected that the majority of the work is still focused on basic/applied research and basic policy advocacy, strategy etc. Work that can have huge ripples, steering global markets and governments where they otherwise would have done nothing for potentially many years.
Another way of evaluating neglectedness is to ask ‘what’s the chance I’ll have an outsized counterfsftual impact by doing X’ and the base rates are just going to be lower for that now given the sheer number of people working on it.
It’s also worth noting that the impact of direct work in climate is inherently limited due to the huge scale of emissions today, so I think it’s very fair to say it has a low neglectedness score.
I think the argument for climate does have to be focused on the scale of the bad outcomes that can be mitigated through additional work, and it still looks very good under that lens. We don’t have to work on neglected problems, it’s just a heuristic for affecting larger speed ups.