I recently read a post which:

• I thought was treating the reader like an idiot
• I thought was below-par in terms of addressing the considerations of the topic it broached
• I would nonetheless expect to be influential, because [censored]

Normally, I would just ask if they wanted to get a comment from this account. Or just downvote it and explain my reasons for doing so. Or just tear it apart. But today, I am low on energy, and I can't help but feel: What's the point? Sure, if I was more tactful, more charismatic, and glibber, I might both be able to explain the mistakes I see, and provide social cover both for myself and for the author I'm critizicing. But I'm not, not today. And besides, if the author was such that they produced a pretty sloppy piece, I don't think I'm going to change their mind.

Negative thoughts on a proposal for a database of project ideas

Written: Jun 3, 2021.

Epistemic status: See NegativeNuno's profile.

Hey [person],

Essentially, I think that it is quite likely that this will fail (70% as a made up number with less than 15 mins of thought; I'm thinking of these as "1 star predictions"); I don't think that the "if I build it they will come" theory of change is likely to work. In particular, I would be fairly surprised if (amount of time people spent working of stuff from your database) / (amount of time you spend creating that database) was greater than 1. Other commenters on both google docs seemed to share this perspective, but maybe the project is worth going through with anyways if you feel that the value of information is high enough. 

Also, your MVP proposal is too large. A google sheet with much fewer rows could serve as a decent MVP, and it would take much less time to set-up and test. It's also not clear why an MVP should be central, or even all that large. 

Part of my pessimism is that I tried a variant of this (more centered around the forecasting part). You can find the google doc which I used for part of the project here: https://docs.google.com/spreadsheets/d/1YHaO-vmjrfM6xbLwa1ljd15KHuCelS03saV81kKYvW4/edit?usp=sharing. Note that this is more for small projects, rather than for research projects. Also note that I did do the obvious thing of first getting the volunteers and then investing the time to gather the projects and estimate their impact.

My ~10 volunteers flaked (except one, who is probably going to be writing her masters thesis on the project I assigned to her, so I don't consider this to be a total waste of time [update: she flaked as well]). I also tried to get a research group going on in Austria, without much success. Part of this was that I underestimated the difficulty of finding volunteers to carry out small projects, and that I overestimated their potential commitment.
 

If I were doing this differently, and this is something you might want to explore yourself, I would first get a **strong** commitment from an already existing cohesive group of people to spend a certain number of hours on a project I decide on. For instance, you could talk with Edo, or with some local EA group leaders about starting a local research group, and initially create your database for that group, and then expand the project to be a central repository for all EAs. For this, your MVP doesn't have to be particularly elaborate; the research projects you propose just have to be better than whatever your volunteers would have otherwise done.

On the other hand, it's possible that the initial publicity of it being a "central EA repository" and perhaps being mentioned in e.g., the EA newsletter might be enough to get the ball rolling, which is something that I didn't try. So that's a  judgment call you have to make.
 

To elaborate on this, an instructive EA forum post to which I keep coming back is Jan Kulveit's What to do with people?. He proposes a "hierarchical networked structure", in which e.g., city EA groups are coordinated by a national EA leadership, which would be coordinated by e.g. regional offices (a la JPAL), which would be coordinated by a central brain. Instead, your central repository, as you currently describe it, would have a pretty decentralized structure (anybody can search it, anybody could edit it (after some quality filtering)), which has its pros and cons. So there is a judgment call to make between doing your project in a more decentralized way (forum post + EA newsletter) or in a more hierarchical way (coordinating with an already existing local group).
 

The above feels somewhat unedited and stream of consciousness; let me know if something doesn't sound right or if you have some different models somewhere, or if I've misunderstood something.

Best,

--

This market (which starts at 15%) is done in the spirit of this account: Will I find that the PIBBSS Fellowship was a success? @ Mantic Markets.