Myself and Zachary Jacobi did some research for a post that we were going to call "Second-Order Effects Make Climate Change an Existential Threat” back in April 2019. At this point, it's unlikely that our notes will be converted into a post, so I'm going to link a document of our rough notes.

The tl;dr of the doc:

Epistemic status: conjecture stated strongly to open debate.

It seems like there is a robust link between heat and crime (at least 1%/ºC). We should be concerned that increased temperatures due to climate change will lead to increases in conflict that represent an existential threat.

• We assumed that:
  • Climate change is real and happening (Claim 0).
  • Conflict between humans is a major source of existential risk (Claim 1).
• Tessa researched whether increased atmospheric CO2 concentrations would make people worse at thinking (Claim 2).
  • She concluded that there is only mixed evidence that CO2 concentrations affect cognition, and only at very high (i.e. indoor) concentrations.
  • If you are concerned about the CO2 → poor cognition → impulsivity/conflict link, worry about funding HVAC systems, not climate change.
• Zach researched whether heat makes people more violent (Claim 3).
  • They concluded that "This seems to be solidly borne out by a variety of research and relatively uncontroversial, although there is quibbling about which confounders (alcohol, nicer weather) play a role. On the whole, we’re looking at at least 1%/ºC increase in crime. The exact mechanism remains unknown and everything I’ve read seems to have at least one counter-argument against it."
  • The quality of the studies supporting this claim surprised both of us.
• We did not get around to researching the intersection of food scarcity, climate change, and conflict .
  • This has been discussed in another comment thread on this post.

The rough notes represent maybe 4 person-hours of research and discussion; it's a shallow investigation.

Some additional notes/clarification/sources for each of the above points:

1. Experts seem (understandably, but rather frustratingly) leery of giving exact death tolls, but here are some examples:

https://www.who.int/news-room/fact-sheets/detail/climate-change-and-health

http://www.impactlab.org/news-insights/valuing-climate-change-mortality/

https://www.youtube.com/watch?v=yXqnKzZiuaE&feature=youtu.be

Note that a lot of expected deaths from those estimates come from exacerbating current neglected tropical diseases like malaria and diarrhea, rather than "direct" climate effects like overheating or droughts.

Note also that 200k-2M puts us in the range of "normal" global health problems like malaria, traffic accidents, etc., rather than making it a uniquely terrifying problem.

2. See prior EA writings by Ozy Brennan:

https://forum.effectivealtruism.org/posts/eJPjSZKyT4tcSGfFk/climate-change-is-in-general-not-an-existential-risk

and John Halstead:

https://docs.google.com/document/d/1qmHh-cshTCMT8LX0Y5wSQm8FMBhaxhQ8OlOeRLkXIF0/edit#

This article has quotes that seem representative of what experts believe:

https://climatefeedback.org/claimreview/earth-is-not-at-risk-of-becoming-a-hothouse-like-venus-as-stephen-hawking-claimed-bbc/

EDIT 2020/1/15: Niel Bowerman estimates the direct risk as real but less than 1/10,000 in the next few centuries:

https://forum.effectivealtruism.org/posts/NLJpMEST6pJhyq99S/notes-could-climate-change-make-earth-uninhabitable-for#2pgNMBikjYTkrGuec

3. Assuming a total, welfarist view about animals, to figure out whether climate change is good or bad for animal well-being, you literally need to have some reasonable estimates of *each* of the following questions:

- Whether climate change will increase or decrease the total biomass of animals in the wild.

- Whether climate change will increase or decrease the proportion of “moral beings with valence” per unit of biomass.

- Whether animals in the wild have net positive lives right now.

- How climate change will affect the average valence of animals in the wild.

People who talk about climate change's impact on wild animal welfare focus on the sharp disequilibria, but I expect it to be relatively small on even a short timescale compared to the (basically unknown) level effects.

4.

https://forum.effectivealtruism.org/posts/GEM7iJnLeMkTMRAaf/updated-global-development-interventions-are-generally-more

5. In addition to the citations above, a) this is my impression from informal discussions with people who I believe know a lot more about this topic than I do, and b) There is the meta-level evidence that EAs who think a lot about cause prioritization usually don't focus on climate change.

6. For example, most (smart, educated) people I talk to are surprised at the balance of increased NTDs as the predominant cause of deaths from climate change, also I learned recently that temperature increases is proportional to log(ppm) rather than linearly, which is really obvious in retrospect but I didn't think about, and I'm willing to bet that 80%+ of STEM college grads wouldn't know.

7. This is a very high-level case. I don't know your life, etc, and if you have an unusually good opportunity to make impact within climate change or if you have detailed models of how climate change affects the world that's very different from my own, you should probably act on your own viewpoints.

In general, I feel like the burden of proof needed to make life decisions based primarily on some stranger on the internet is quite high, and I don't think I have met it.

That said, some random brainstorming:

Care about helping poor people not die from malaria due to climate change -> work on making sure poor people don't die from malaria, period

Care about the long-term future -> explore other long-termist stuff like AI Safety, biorisk, moral circle expansion etc.

care about animal welfare -> factory farming and look into research on wild animal stuff. (Addendum: I think it'd be surprising but not crazy if climate change work is better for animal welfare than work on reducing factory farming, but in worlds where this is true, your priority in 2019 should probably be to study wild animal welfare rather than to assume the connection).

Less obvious stuff:

Generically care about the environment -> Look into indoor and outdoor air pollution. (I'm less confident about this suggestion than the previous 3)

8. There are also likely similar values, like caring about the Global South, animals, and future people. It’s very difficult to communicate to someone that you think their life’s work is not necessarily the best thing to do with limited resources (and most people are less used to this criticism than EAs), and extreme prudence is recommended.

A secondary point is nuance. I think it's bad from both an epistemic and PR perspective if the message will be distorted from “our best understanding of the situation is that mainline climate change mitigation is unlikely to be the marginal best thing to work on for most EA people with flexible career capital” to something more catchy but much less accurate.

Thanks for this. I think it's valuable when well-informed EAs make easily interpretable claims about difficult questions (another such question is AI risk). This post (including the "appendices" in the comments) strikes a good balance; it is epistemically responsible, yet has clear conclusions.

Here for easy disagreement:

Climate change by itself should not be considered a global catastrophic risk (>10% chance of causing >10% of human mortality)

I'm not sure if any natural class of events could be considered global catastrophic risks under this definition, except possibly all kinds of wars and AI. It seems pretty weird to not classify e.g. asteroids or nuclear war as global catastrophic risks, just because they're relatively unlikely. Or is the 10% supposed to mean that there's a 10% probability of >10% of humans dying conditioned on some event in the event class happening? If so, this seems unfair to climate change, since it's so much more likely than the other risks (indeed, it's already happening). Under this definition, I think we could call extreme climate change a global catastrophic risk, for some non-ridiculous definition of extreme.

Hey there, interesting article! In this talk from the most recent EA Global, Niel Bowerman (climate physics PhD and now AI specialist at 80,000 Hours) gives some thoughts on the relationship between climate change and existential risk. Essentially I think that there's some evidence about point 2 on your list.

• In his talk, Niel argues that climate change could cause human extinction in itself, under some scenarios. These are quite unlikely, but have non-zero probabilities. When we consider that emissions are likely to increase well beyond 2100, beware the 2100 fallacy of cutting shorts impact analyses at an arbitrary point in time.
• The larger contributions very roughly are probably from climate change contributing to social collapse and conflict, which themselves lead to existential risks. Toby Ord has called this an 'existential risk factor'. I think the question isn't "Is climate an existential risk?" but "Does climate change contribute to existential risk?" in which case, it seems that the sign might be yes. Or perhaps "Is climate change important in the long-term?" in which case, if we're thinking across multiple centuries, even with lots of technological development, if we're looking at >6C in 2300 (to pick an example), then I think the answer is yes.
• All of this being said, I still think it's a fair to argue that AI, bio, and nuclear are more neglected and tractable relative to climate change.

What do you think of Niel's talk and this framing?

On Facebook, a couple of people have asked me on the existential/global catastrophic risk posed by climate change causing or exacerbating widespread (nuclear) war.

Here's what I wrote.

(Note that this is what I personally believe, rather than something I'm confident experts on international relations will agree on)

I think climate change as an indirect GCR is less crazy than climate change as a direct GCR. But I still don't find it very compelling.

I think if you look at the entire history of human armed conflict, deaths attributable to wars are substantially lower than the GCR definition (population adjusted). Also most deaths historically have been incidental/civilian casualties, which have gone down over time for various reasons, including better medical sanitation and general wealth (so less likely that, eg. war -> pillage -> mass starvation).

So to argue that conflict from climate change leads to a GCR, you need a strong reason that "this time is different." One possible reason is access to nuclear weapons, but for this to be true, you need a compelling reason for people to differentially use nuclear weapons as a result of climate change at a fairly high probability. (There are reasons that also point in the other direction).

For food/water wars specifically, I'm reasonably convinced by Amartya Sen's research that pretty much all famines in recorded history are a result of distribution rather than production (his exact claim is that "no democracy has ever had a serious famine," which I think is too strong, but I think the general trend is correct).

In addition, the evidence for wars over food/resource contention is a lot weaker than I would have naively guessed before looking into it briefly. For example, here's an account from a science journalist who was commissioned to write a book about "water wars" and finding that there isn't enough credible evidence to write one:

https://www.nature.com/articles/458282a

So to recap,
1) I think it's relatively implausible that the relatively small food shortages from climate change will result in mass famines.

2) The famine -> war connection is also quite tenuous.

3) The war -> nuclear war connection isn't too strong either.