Peter_Hurford

Along with my co-founder, Marcus A. Davis, I run Rethink Priorities. Previously, I was a professional data scientist.

Comments

Nathan Young's Shortform

We are definitely planning on doing this kind of research, likely sometime in 2021.

Recommendations for prioritizing political engagement in the 2020 US elections

 What I can say is that I'm confident that the suggestions we've offered a) will be helpful rather than harmful and b) are substantially better than just randomly signing up for a volunteer opportunity.

 

I agree that's totally fair! And I appreciate you doing that work.

As a side note, you seem to have a lot more faith in the efficient markets hypothesis as it applies to organizational behavior than I do! I don't think I have ever encountered an institutional ecosystem that was unburdened by poor strategic thinking, inefficient legacy practices, and failure to coordinate complementary or duplicative efforts. In that regard, I've actually been pleasantly surprised with how high-functioning the progressive organizing space seems to be on the whole.

I'm not sure this is a side note... this might be the main crux of our disagreement!

My prior is that a team of smart non-experts uncovering some large tactical error in large well-funded groups that are highly incentivized to not have errors of that type is certainly possible and actually quite achievable, but probably takes on the order of >4K hours of work. I also think it is easy to think you have found an error that is not in fact an error. I'm not sure how much time you've spent on this?

I do have massive uncertainty about how true the efficient market hypothesis is, for a variety of domains.

Recommendations for prioritizing political engagement in the 2020 US elections

Perhaps you are right, but my outside view would suggest some skepticism that professional campaigns haven't identified these winning techniques that you have. Perhaps they have some reason for doubting them?

Relatedly, do you worry about optimizers curse when identifying top opportunities?

(FWIW my personal experience does suggest campaigns can be shortsighted at times... I recall volunteering for Hillary in 2016 and spending an unfortunate amount of volunteer time being directed to make calls in Iowa)

Recommendations for prioritizing political engagement in the 2020 US elections

I agree it is definitely not a neglected issue but [per this table](https://mediaproject.wesleyan.edu/releases-101520/#table5) there is still definitely room for Dems to increase their advantage... some races (e.g., GA, MI) aren't even 2-to-1 in favor of Dems yet.

Recommendations for prioritizing political engagement in the 2020 US elections

The best volunteer opportunities we've seen in terms of cost-effectiveness cash out to the equivalent of about a $150-200/hr rate if you were going to donate that money to our top recommendations instead

I find that very surprising! Couldn't a campaign hire a person to do these top volunteering opportunities for ~$20/hr? Presumably that would cap the value of volunteering at ~$20/hr?

Getting money out of politics and into charity

matching a pair of opposing campaigns and sending the money to a less-controversial charity instead.

It does assume the two campaigns will use money equally effectively, which may well be false.

Recommendations for prioritizing political engagement in the 2020 US elections

What do you see as the case for political volunteering, as opposed to trying to use time to earn/save money to donate more instead?

Recommendations for prioritizing political engagement in the 2020 US elections

Estimates suggest investing in these techniques will be roughly 8 to 10 times as cost-effective as donating to the Biden campaign, with the potential to net hundreds of thousands of votes across key states.

Can you offer more detail on this? I would probably consider donating to it if I understood it better.

Recommendations for prioritizing political engagement in the 2020 US elections

I'm pretty undecided on how to give to pro-Biden/Dem efforts (as well as undecided on how pro-Biden/Dem efforts compare to other opportunities), but I wanted to give a small amount (as I no longer earn-to-give), so here is my giving:

My top idea right now is this campaign, that put together by a strong EA-aligned progressive-aligned data scientist with strong expertise in helping Democrats win and seems like a better bet than solely focusing on Theresa Greenfield. I haven't vetted it but I put $250 for it based on my intuition. (EDIT: I notice you mentioned this in your post, so I guess we'd either have to agree to disagree or have a longer discussion. My guess is that your analysis may be missing marginal media market costs?)

Worth keeping in mind that (as you probably know) due to FCC rules, donating directly to candidates seems like a stronger bet than donating to any other PAC, because candidates (a) get guaranteed minimum advertising rates and (b) get direct access to the candidate (more persuasive).

An honorable mention is this ad campaign, which feels strong and is also backed by strong message testing by people with a good track record for this sort of thing. I donated $100 to it.

I also think state legislature races are important, so I donated $100 to this impact-focused list that was also put together by people with a strong track record of putting together this sort of thing.

I like the idea of flipping Texas, so I donated to the Swing Left campaign for it. I don't know how that compares to Lone Star Votes or the Princeton Election Consortium. I also don't know the track record of Swing Left. This is the idea I feel least confident about, so I only donated $50.

Of course I also donated $100 to Biden directly following the logic that due to campaign finance and FCC rules, it is good for small donor individuals to donate directly to candidates.

And I gave $10 to ActBlue for facilitating all of the above.

Ideally I would've liked to have done more research and focused more money on one campaign I feel best about (and to have donated more money overall), but I don't think I'll ever get to that point, thus this scattershot giving approach.

(Obvious disclaimer: These are purely personal opinions and doesn't reflect any policy or position of Rethink Priorities.)

Sortition Model of Moral Uncertainty

What about integrating this into a Monte Carlo method?

Load More