Hide table of contents

The EA forum has a high density of research talent, and a high density of good ideas for write-ups that would be valuable. Still, lots of good ideas are left unexplored. I think one factor here is that it’s unclear how useful most EA forum posts actually end up being, and the incentive to write them isn’t as strong as it could be. Also, there's a coordination problem between people with important questions and people with the time and talent to answer them.

One way to solve this would be a bounty board on this website for research writeups. In my mind, it could take the form of donation-directing — e.g. “I will donate $2000 to a nonprofit of the authors’ choice for anyone who can provide a post answering 'How do political campaigns affordably identify undecided voters?' or 'Which common cockroach control methods will cause the least suffering?'"

I imagine this as a long bulletin board full of requested forum posts and their associated bounties. Any user who wants to research and produce a writeup on any of these questions is offered some guaranteed impact in the form of getting to direct a donation. The person offering the bounty would get to have their research question answered while also paying in the form of a donation to a highly-effective charity. The bounty could have some minimum standards that need to be met before being paid out, or the payment could be based entirely on the bounty issuer's judgment.

Here's a short Q&A I made regarding this idea. I haven't thought through it too much, so I'd be curious to continue the conversation in the comments.

Q: Why use donation-directing bounties as opposed to normal cash bounties?

A: Normal cash bounties could be useful here, but I think donation-directing (1) fits the spirit of the EA movement more, (2) may eliminate some incentives for people to game the system out of personal interest, and (3) reduces the perceived cost for the person offering the bounty, since they know the payment will probably go to a high-impact charity and thus have positive effects in and of itself, even if the research doesn't prove useful. I also suspect there would be greater legal implications for the EA Forum in the case of unconditional cash bounties, since it becomes a sort of marketplace. I'm not sure though. In some small way, the EA forum also already does this via the bounty tag.

Q: Is the EA forum the best venue for this sort of thing?

A: Probably not. But it's the best I can think of right now.

Q: Let's say the issuer was going to donate a certain amount to effective charities no matter what, and they create a set of bounties for questions they'd like answered, where the bounty total represents the total amount they planned to donate regardless. In this case, what's the incentive to actually pursue a bounty, if it's not causally responsible for a donation to a highly-effective charity?

A: Lots of people on the forum have different priorities about where money can be best used. Bounty pursuers can still find value in directing the donation to the specific cause and org that they find most promising, even if the money would have been donated anyway. This also rewards causes and orgs that are supported by people with the research skills / general thoughtfulness needed to win a bounty — people who may also have better judgment when it comes to charitable giving. Of course, this represents an equivalent cost for the bounty issuer, as they lose control of which cause and org will receive their donation. But some bounty issuers are probably fairly excited about various cause areas and feel this cost less strongly than the pursuer feels the benefit.

Also, some bounties might have a real causal role in increasing the issuer's total charitable giving. I'd be excited to pay for answers to key questions I have above and beyond the amount I plan to donate, and I imagine the same is true for lots of people here.

Q: Should bounties be competitions that reward the best entry or open solicitations that reward the first entry which meets the minimum requirements? If the latter, what if someone completes a bounty first but their contribution is still low-effort, flawed, or misses the central question?

A: Competitions could be good, and this is most similar to what groups like Open Philanthropy have done in the past, but it also might disincentivize people from pursuing the bounty (as they see a lower likelihood of collecting it) or lead to wasted, duplicative effort as multiple bounty pursuers look into the same question and potentially come to the same answer. However, if multiple bounty pursuers find different answers, that would also be useful and help clarify key uncertainties or identify more room for research.

If bounties aren't a contest but instead are awarded to the first person meeting the minimum requirements, I'm not sure how to handle determining what the minimum requirements are. Bounty issuers could have full discretion to determine whether bounties pay out or not based on the perceived quality of the bounty collecter's contribution, but the uncertainty might discourage people from pursuing the bounty, and once the research is complete, the issuer has little incentive to follow through and award the donation (especially if they disagree with where the bounty pursuer might want to direct the donation).

I think the best way to handle this is that bounty pursuers would issue short proposals — saying that they are interested in collecting the bounty, what their relevant experience is, on what timeframe the post can be done, a "karma" system reflecting satisfaction of previous bounty counterparties, etc. Then, if the issuer agrees, they lock-in the arrangement and the issuer commits to paying out a bounty when the research is done. Even if this fails some small percentage of the time, there could be an insurance fund where the EA forum covers bounties in the event that the issuer walks back their agreement. This proposal system also minimizes duplicative effort.





More posts like this

Sorted by Click to highlight new comments since: Today at 12:54 AM

Executive summary: A donation-directing bounty board on the EA Forum could incentivize valuable research and help coordinate between those with important questions and those with the skills to answer them.

Key points:

  1. The EA Forum has a high density of research talent and good ideas for valuable write-ups, but many are left unexplored due to unclear impact and weak incentives.
  2. A donation-directing bounty board could allow users to offer donations to effective charities in exchange for posts answering specific research questions.
  3. Donation-directing bounties fit the EA movement's spirit, reduce incentives for gaming the system, and may be perceived as less costly by bounty issuers.
  4. Bounty pursuers can direct donations to their preferred causes and organizations, rewarding those supported by people with strong research skills and judgment.
  5. To avoid low-quality or duplicative work, bounty pursuers could submit proposals, and an insurance fund could cover bounties if issuers fail to follow through.



This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.

Curated and popular this week
Relevant opportunities