I’m a researcher at Probably Good and a philosophy PhD student at the University of Edinburgh.
Dylan_Balfour
Karma: 100
Hi John,
Again, I want to thank you for the feedback you’ve given. We really appreciate engagement with the content, including meaningful critical engagement.
Given the volume of comments you had, I’m not going to provide too much in the way of object-level discussion (with one exception in a reply below). However, I think there are two important overall takeaways:
1: Many points you raise give important context and nuance, and in some cases importantly different headline conclusions. With these comments in mind, as well as further research, we’re going to revisit the content and make changes where we think it’s appropriate. It’s highly likely we’ll conclude that some changes are warranted – for instance (but not exhaustively), by clarifying where we’re giving figures conditional on more severe emissions scenarios, and discussing the effect economic growth is likely to have on strengthening adaptation. I think the content will be more accurate as a result of your comments, so thank you!
2: It’s possible someone who has just read the comments on this post could come away with the impression that each of these criticisms reflects an inaccuracy with our content, which we don’t think would be a correct inference. In some cases, we’re either already aware of the nuances you’ve raised, don’t think they give us sufficient reason to update our content, or don’t think they necessarily represent claims we’ve made.
For example, in a comment below you imply that we exaggerate the importance of wildfires. I don’t think this is a fair characterisation; not only do we exclude wildfires from our priorities article, we devote only one word to them in our climate overview. We mention them only to note that they are expected to increase in frequency due to climate change (which is true). Additionally, we believe you substantially underestimate of the mortality impacts of wildfires. I discuss this in a reply below. Here are a few quick smaller points:
You say that climate change will increase food insecurity. True, but studies that incorporate the effects of climate change, economic growth and agricultural progress find reductions in food insecurity on nearly all socioeconomic scenarios. I think it would be worth providing this context.
Agreed—we discuss this in our article on climate change priorities. It might also be worth mentioning in the overview.
In support of the claim that climate change is an important contributor to conflict risk you cite an 80,000 Hours article. The IPCC is highly equivocal on the effects of climate change on civil conflict. It basically says nothing about the effects on interstate conflict and no scholars of great power war think it is an important driver of 21st century potential great power conflicts.
I think the wording of this point implies we have more confidence in this dynamic than we state in the article. Just for reference, our phrasing of this point is ‘On top of this, some are also concerned that the effects of climate change may indirectly exacerbate other large-scale risks, too – such as increasing the chance of international conflict… These considerations increase the importance of climate change beyond its direct effects’. This isn’t intended to (and hopefully doesn’t) convey a high probability of climate change leading to great power war – though I understand that more context and additional caveating would likely be beneficial.
To characterise the trend in flood deaths as anything other than a dramatic downward trend seems clearly wrong.
If you’re referring to my comment, my claim was that ‘annual flood deaths have been fairly flat since the 1970s.’ I think this is a pretty fair extrapolation from the OWID data, given that we’re talking about absolute mortality numbers and not the average lethality of each flood event!
On a general note, as a small organization we have limited capacity and expertise of our own. This means we rely on deference to both respected institutions and mainstream resources more than we otherwise might. I think this deference is typically justified, though there are many obvious caveats. Regardless, pointing us towards specific points (or more general claims) you believe are suspect as well as resources that raise valid disputes helps us to better evaluate contentious claims—so we appreciate that.
Thanks for these comments! Really appreciate the time you taking the time to give such thorough feedback. Needless to say, there’s quite a lot to think about here (both in terms of where updates to the content might be warranted and where we might stand by the original claims). I’ll give my thoughts when I’m able to, likely in the next few days.
Thanks a lot for the response! I was the main researcher for these articles—here are a few thoughts from me and the team:
It’s a little tricky for us to evaluate the consensus on these issues ourselves as we don’t have the capacity or expertise to review all the relevant papers and assess their relative influence. (We’d welcome anyone sharing their views on this, especially if they have expertise in the field!)
Because of this, we typically defer to respected authorities – especially the IPCC – which is likely a decent proxy for expert consensus. While any one institution will inevitably have its shortcomings, the IPCC’s comprehensiveness in reviewing the literature is impressive, and they do a fairly good job of representing their uncertainty based on the literature in their reports.
That being said, there seems to be a lot of uncertainty about expected adaptation efforts, perhaps more so than there is uncertainty about the direct effects of climate change, and the extent to which the IPCC provides estimates accounting for adaptation depends on the problem. In short – we don’t really know, (though we’re open to being corrected on this).
Because of this, when trying to estimate the value of working on adaptation, given high uncertainty about how much of it will take place without our intervention, as well as a reportedly widening adaptation finance gap, it’s still worth thinking about the expected impacts assuming minimal or current-trajectory adaptation. Though, naturally, the more we learn about the impact of expected adaptation efforts, the better.
Here’s a few more specific points on deaths from non-optimal temperatures and natural disasters:
On temperature and mortality:
When discussing the promisingness of an adaptation response to heat stress, we think what matters is the absolute numbers of heat-stress-related deaths, which we can be quite confident will significantly increase, and not whether on net climate will cause or prevent more deaths (taking into account other types of deaths prevented). We included the estimate that increases in fatalities will outweigh reduction in deaths because it provides interesting and relevant context, not because it directly bears on how promising this area of adaptation is.
It’s also worth noting that even Zhao (2021) claims that ‘in the long run, climate change is expected to increase the [temperature-related] mortality burden’, even if it reduces mortality in the short term. They don’t give any concrete projections, so it’s unclear the extent to which the numbers we’ve given align with this claim – but the direction is the same. Nonetheless, this does complicate the discussion around temperature-related mortality, so thanks for bringing it to our attention!
On natural disasters:
First, although fatalities from natural disasters have decreased, most of this reduction has been in high-income countries (if you look at the OWID page and filter by low-income countries, the downward trend is much less clear).
Additionally, looking at the decadal average of all natural disasters can hide how sparse the data points driving this impression are. If we look at the year-by-year view of natural disaster deaths, just a handful of terrible events are responsible for the vast majority of deaths throughout the 20th century, which means we’re wary about inferring long-term trends from the decadal view – especially for events that fit a power-law distribution.
On top of this, looking at the specific kinds of natural disasters we’re discussing, it’s not clear they’re subject to the same downward trend, even on the decadal view. For instance, the OWID chart on natural disasters shows deaths from extreme temperatures have sharply increased in recent decades, and annual flood deaths have been fairly flat since the 1970s.
And just as a general point on the importance of climate-related extreme weather events, only counting fatalities may also give us a misleading picture, since it excludes other significant effects like damage to infrastructure, worsened mental health, and cumulatively huge economic losses. Again referring to the OWID page, the economic damage inflicted by natural disasters is much higher now than it was in the 1960s (as a share of GDP), despite them causing fewer deaths.
Our current view is that looking at the century-long average as an estimate of expected future fatalities may be an overestimate (primarily due to reductions in high-income countries), but it’s unclear that existing adaptation efforts will similarly continue to reduce harms going forward (or prevent potential rises in disasters where such are expected due to climate change).
The bottom line is we think there’s a lot of complexities in the analysis of the quantitative estimates, especially when adaptation and human response is a critical factor. But we believe that the areas we’ve identified are nonetheless promising areas to have a positive impact given what we know.
Always great to hear more about the Incubation Program—and would highly recommend people apply if they’re interested!
A quick plug: at Probably Good, we’ve written a career profile on nonprofit entrepreneurship (heavily inspired by CE’s approach) that covers other considerations around personal fit—it might be of interest to people who found this post useful.
This is a great post! It’s really good to see some specific advice for people in LMICs—something sorely needed in EA.
Speaking as part of Probably Good, we’re trying to fill some of this gap (to the extent we’re able). On top of our profile on civil service careers in LMICs , we also recently released a profile on monitoring and evaluation careers, a path we note could be particularly promising for people based in LMICs. We’d be really keen to hear your thoughts on how we might do this more/better!
I just wanted to echo your sentiments in the last part of your comment re: Beckstead’s quote about the value of saving lives in the developed world. Having briefly looked at where this quote is situated in Beckstead’s PhD thesis (which, judging by the parts I’ve previously read, is excellent), the context doesn’t significantly alter how this quote ought to be construed.
I think this is at the very least an eyebrow-raising claim, and I don’t think Torres is too far off the mark to think that the label of white supremacism, at least in the “scholarly” sense of the term, could apply here. Though it’s vital to note that this is in no way to insinuate that Beckstead is a white supremacist, i.e., someone psychologically motivated by white supremacist ideas. If Torres has insinuated this elsewhere, then that’s another matter.
It also needs noting that, contra Torres, longtermism simpliciter is not committed to the view espoused in the Beckstead quote. This view falls out of some particular commitments which give rise to longtermism (e.g. total utilitarianism). The OP does a good job of pointing out that there are other “routes” to longtermism, which Ord articulates, and I think these approaches could plausibly avoid the implication that we ought to prioritise members of the developed world over the contemporaneous global poor.
I’m oblivious to Torres’ history with various EAs, so I’m anxious about stepping into what seems like quite a charged debate here (especially with my first forum post), but I think it’s worth noting that, were various longtermist ideas to enter mainstream discourse, this is exactly the kind of critique they’d receive (unfairly or not!) - so it’s worth considering how plausible these charges are, and how longtermists might respond. The OP develops some promising initial responses, but I also think a longer discussion would be beneficial.
I again genuinely appreciate this feedback, and what I said above also applies here: we’ll revisit the content with your comments in mind and will likely make some changes. Because I’ve already given some object-level discussion in this thread, I wanted to provide a response to this claim:
Whilst you’re correct that direct deaths from wildfires are low, these are not what primarily drive deaths caused by wildfires; the indirect effects from wildfires (via air pollution) are much larger.
For instance, Ye, et al (2022) attributes over 130,000 deaths to wildfire-related PM2.5 exposure in Brazil from 2000-2016 (>8,000 per year), and Chen et al (2021) seemingly attributes around 30,000 deaths per year from 2000-2016 to wildfire-caused pollution, globally.
It is difficult for me to verify every assumption in these papers, but even if they’re significant overestimates, it’s clear that indirect deaths from wildfires vastly outnumber direct deaths. Several other papers that directionally support this point (with a fairly wide range between estimates):
Kollanus et al (2016): >1,000 deaths in Europe in both 2005 and 2008
Neumann et al (2021): 720 deaths per year in Western US, 1996-2005
Pan et al (2023): 4,000 deaths per year in US, 2012-2014
Johnston et.al (2015): 339,000 global deaths per year, 1997-2006 (from landscape fires—a broader category than wildfires)
Roberts & Wooster (2021): 677,745 global deaths per year, 2016-2019 (also from landscape fires)
Of course, the extent to which climate change drives wildfires, both now and into the future as warming increases, is an important crux in how many of these deaths we should attribute to climate change. But, when evaluating the importance of wildfires themselves (e.g. when thinking about adaptation), looking just at the direct harms will lead us to significantly underestimate their impact.