Update: See CSS6, in comments below.
Report: https://1drv.ms/b/s!At2KcPiXB5rkvgaVnrs4N3zyN6o6
Modeling: https://1drv.ms/x/s!At2KcPiXB5rkvX-xJkfAYi4xHO6R
Preface
CSS5 marks the first time that a significant revision has been made over our previous recommendations. In CSS4 and earlier reports, Andrew Yang was rated positive but substantially below the recommended Democratic candidate Cory Booker. In this report, we have an approximate tie between Yang and Booker, and both are provided as recommended candidates. Two major changes caused this shift. First, we revised the weight of long-run animal issues to use a more direct estimate comparing animal population sizes and welfare to the global human population, as opposed to the previous method where we had a heuristic comparing animal farming to human issues based on the ratio of farm animals to humans living in the US. This heuristic neglected the direct impacts of many human-oriented policies on non-Americans and this caused animal issues to get too high of a weight by comparison. The second major change was a decrease in the weight given to work experience. This happened when my subjective assessments on the importance of qualifications were updated on the basis of a small survey of other EAs, who generally thought that work experience is less important than I believed it was.
To be clear, Yang was rated positively before just as now, only the relative priority has changed. With your input, the report can continue to grow more accurate. The topic weights are the most uncertain and sensitive parts of this report, and warrant continued critical examination. Due to our increased awareness of this issue we have also reduced the weight of the long-run category from 5 to 4.
Hi Siebe,
I'm the author of those Guesstimate models — glad to hear you found them interesting! I wanted to flag that the models are still evolving, and there's a chance the results will change a bit. I'll be publishing full write-ups of my findings and possible implications for x-risk over the next few months.
I plan to go into detail about this in the reports, but the short answer is that the amount of smoke generated by nuclear detonations depends on a lot on the kinds of targets being hit. Really big, densely populated cities have a lot of flammable material, so bombing a country like China produces much more smoke than bombing the US or Russia. I do think this might have interesting implications for which conflict scenarios should trouble us, but there are a few more factors I’m still building into the models that may end up being more important.
Thanks for the answer. I look forward to seeing your write-up and how the models evolve!
One thing that seems missing in the current model is how smoke maps to famine (does the location of the smoke matter?), but perhaps assuming a linear relationship between amount of smoke and amount of casualties from famine is a good approximation.
I made a significant mistake in this. I calculated the sensitivity analysis for Buttigieg's electability incorrectly and this underestimated the amount by which his position in the nomination scoring can climb. With recent boosts in his popularity and prediction market expectations since CSS5 was published, Buttigieg could be a viable recommendation. I will review this stuff and put out CSS6 fairly quickly.
Here is CSS6: https://1drv.ms/b/s!At2KcPiXB5rkvg3MH6izzguUYpuR
I realize this is a bit late, but all of the 1drive links say the item "might not exist or is no longer available", then ask to sign in to a Microsoft account.
Sorry for my own lateness. I have removed all old versions. The most recent PDF report can be found here (I am keeping this as a permanent link): https://1drv.ms/b/s!At2KcPiXB5rkyABaEsATaMrRDxwj It contains permanent links to the excel model, public draft etc. Now when I make a new version, I save over the previous version while keeping the same link.
Also on the topic of nuclear war: I found it very surprising to see that, according to the report, the expected deaths of a US-Russia exchange is lower (3 billion) than US-China or (5.5 billion) or Russia-China (7 billion). This implies that the existential risk from nuclear war is higher from a Russia-China exchange than from a US-Russia exchange. How should we translate these expected deaths to existential risk?
I found the Guesstimate models interesting, and figured out that this is largely because a nuclear bombing of China is expected to produce the most smoke. But there's a factor in TSG1 ("total smoke generated by nuclear weapons detonated in China (Tg)") that I don't understand, but that factor drives much of the differences in total smoke generated. In the US-China model it's 1.58, and the Russia-China model it's 2.04, while a Russian bombing of the US has only a factor of 0.664. Could you explain what this factor is and where it's derived from?
Also: the Russia-China model has some broken links due to an extra "]" in MNO2.
About Nuclear Nonproliferation you write:
What do you mean by 'being good at threatening'? It seems counterproductive to threaten North Korea, as this could cause international conflict, and backfire by triggering action from North Korea against South Korea and possibly Japan. As argued in this article, it seems that the only options are to persuade China to increase economic pressure, or to accept a nuclear North Korea. North Korea has always been consistent and, given their goal, rational: they aim for nuclear weapons for deterrence and know that using nukes will mean assured destruction.
being good at handling {[nuclear proliferation],[threatening states such as North Korea]}
^ This, will clear up the language for the next version.
I found a possible error in CSS6. On page 91 you say:
Buttigieg supports increased taxes on the wealthy, which presumably would include millionaires andbillionaires, and did not mention equalizing the capital gains tax with the income tax. We give him 1 point.
We’ve found no information to judge Buttigieg, so we give him 0.3 points.
Thank you. Just forgot to delete the 2nd line. The Excel file had the updated 1-point score so the scores were correct. Fixing the report now for next edition.