EA Forum Reader

I would say it a little differently. I would say that "judgmental" forecasting, the kind typically done on Metaculus or Good Judgement Open or similar platforms, CAN involve mathemtical models, but oftentimes people are just doing some simple math, if any at all. In cases where people do use models, sure it would make sense to link to them as sources, and I agree that would also be valuable to track for similar reasons. Guesstimate seems like the obvious place to do that.

I think that is separate from the proposition I intended to communicate for primarily text based research. 

I also wasn't anticipating any need to do scraping if this was implemented by the two platforms themselves. It should be easy enough for them to tell if a citation is linking to an EA forum post? Metaculus doesn't have a footnote/citation formatting tool today like the EA Forum's. (Although if you were to scrape, finding EA forum links within citations on this forum seems pretty well defined and achievable? idk, I don't write much code, thus me floating this out here for feedback)

Thanks for the thoughts!

Yeah, that’s where I asked first. No responses in the thread, but someone DMd me to suggest looking at universities: https://www.lesswrong.com/posts/kH2uJeQZMnEBKG9GZ/can-independent-researchers-get-a-sponsored-visa-for-the-us [LW · GW].

Good points. 

This post [EA · GW] comes to mind, which I cited in my nuclear GCR forecasts here, along with many other posts from that series. In general I expect posts from Rethink Priorities to be relevant. I've seen similar quality posts for AI risks and pandemics here. Most of my familiarity is with GCR's but I expected there to be strong overlap between popular forecasting topics and popular EA forum topics more generally. There are lots of GCR related questions on Metaculus, and you can find many cited in that link with my forecasts.

Still, I think you're right that this wouldn't be applicable to the majority of EA forum posts. Maybe it's only even displayed once a post is cited in a forecast, or only a particular tag is eligible for this in order to simplify the implementation. 

I do think making people's forecasting performance more obvious in different contexts would be very useful for the community (re: your brier scores in EAF profiles idea), and would love a central site that's sort of like a minimum viable linkedin that consolidates relevant metrics for an individual across the top forecasting platforms and has an API that makes it easy to connect to other accounts, or use with discord bots etc. I may write about this soon.

Generating forecasts associated with a post is interesting and I'm sure there are UX opportunities to make this easier / more common, but I need to think more about it.

Thanks for the thoughtful response!

This is an add-on to this response [EA(p) · GW(p)] I wrote and sort of to all the SBF-EA-related stuff I've written recently. I write it for personal reasons. I've argued that we should have patience when assigning blame for EA leadership and not assume leaders deserve blame or ever were necessarily incompetent in a way that counterfactual leaders would not have been. But this point is distinct from thinking there was nothing we could do or no signs to pay attention to. I don't want to be seen as arguing there was nothing that EAs in general could do, so here are my actual thoughts of what was, on it's own, enough to warrant taking action of distancing ourselves from SBF:] 

FWIW I do think mistakes were made around SBF. I'm just not willing to pin it on EA leaders specifically (yet), or even EAs/EA itself specifically (to the exclusion of others). Anyone, including journalists and finance people, could have seen red flags who watched SBF's interviews. IMO, the major red flags in retrospect were things anyone who was paying attention (I was only a bit?) could see: 

(1) SBF talking about ponzi schemes, and some of his testimony regarding crypto regulation (I think?) which apparently made the ponzi scheme possibility look more real. 

Personal take and regrets: I saw neither of these myself but my newly-EA gf thought they were morally troubling before the crash and told me. We had a couple short conversations about it which basically led to "Oof, IDK what to say" from me. I thought of looking it all up, or messaging prominent EAs on her word alone. But I did not, mostly due to confusion about what it meant... "Isn't this just the nature of crypto as an asset? Or is this unethical? Am I getting into the moral dilemma that EAs just shouldn't do finance to E2G? Is that a bullet I want to bite, because I think I'll have to argue that?". 

I didn't think of it as a red flag for upcoming crash and bankruptcy. I guess someone who knows about ponzi schemes can say if I was dumb to not think of this. But it was a red flag that he didn't care about FTX users, and he might not be "a good guy" (even by consequentialist standards, the balance gets way more complicated and you can't be anywhere close to sure enough to take such risks with citizen's money), and I still kick myself for not messaging someone. It wouldn't have been that much notice, a couple months maybe? But maybe could have helped EA distance itself proactively. Sigh. 

(2) SBF's violation of kelly criteria/biting bullets on St. Petersburg paradox. 

Personal take and expressing shock/light scolding: I never knew how "all-in" he was, but I'd have found that super alarming, and this I think I'd have tried (more seriously than about #1) to talk to someone about it. (Basically all I know about betting is that you "never bet it all. always leave enough to bet another day", but I know it as the golden rule). It still troubles me that EAs and others seemingly thought SBF's responses were philosophically neutral or something, when actually it was a glaring red flag that the companywould fail, even without fraud. And also a red flag that he was kind of self-deluding, or trying too hard to be clever via breaking rules. Like. If you want to make more money to do more good, just do the thing that is already known to make the most money in the long run (kelly), don't instead pull numbers out of your ass to reinvent a wheel, except much worse than before. This also tied into SBF acting way too morally sure of himself--personally I'd never bet earth's entire future without others' consent because of one moral theory coupled with the multiple universe theory, in regards to a situation that is called a paradox for good reason (it's not supposed to be an easy decision, which generally means you should defer to group consensus!).

That all said: I think EAs' philosophical naivety here, or brushing it off, is disappointingly normal? As proven by no one else in the world writing a hit piece about SBF (that I'm aware of). Bystander effect too maybe, since that stuff is way more public than ex-Alameda employee complaints, it'd be normal to think "Well, lots of people are seeing this, and if no one else, including FTX investors, sees it as a problem, I guess it must be okay." Idk. I'd like EAs and non-EAs to do better at pinpointing problematic actors in this regard (and we can only control EA so we should focus on this failure mode a bit), but it's all qualitatively different from what the Time article [EA · GW]is talking about.

I expect I'm not the only one who feels this way and had/has feelings like this, including personal guilt, even though I was by all accounts a total outsider. And if I'm not the only one, that's one reason it feels very weird to me to pin it on EA leaders (as of right now)]

I could say a lot more, and be more precise, and doublecheck some stuff in #1 which I still never did, but this is just a shortform. 

Glad you're thinking about improving research and forecasting! My very quick take after only skimming your post is that it just happens too rarely that an EA forum post significantly informs specific Metaculus questions. Consequently investing in such a feature seems not sufficiently useful. (Though that might change of course if there were many more Metaculus questions that relate to topics in EA.) Maybe you could list some concrete examples for such EA forum - Metaculus connections? (Sorry if I missed you listing examples!)

Some other random ideas about forecasting on the forum:

• Display relative Brier Scores from Metaculus in EAF profiles -> incentivize forecasting, enable forum readers to roughly weigh comments by epistemic track records
• Hook up the forum to the GPT-4 API and allow forum writers to generate forecasting questions at the end of their posts.
  • Scott Alexander does this regularly and I find it useful to get a concrete bottom line of his uncertainties and concrete things he's forecasting based on the post, e.g. see here.

Few Things

•  https://forum.effectivealtruism.org/graphiql [? · GW] if people want to scrape.
• I only skimmed your post (let me know if I'm misunderstanding) but I have an issue with this idea. Many forecasts require complicated mathematical models to describe. You can't simply link to sources. You also need to link to a model. Blog posts/txt files, which are essentially what the forum is, are extremely hard to scrape and parse unless everyone starts adopting conventions. So you max you functionality out at linking, this isn't very automated. 
• If you are recommending connecting a full mathematical model from the forum, let me suggest that rather than connecting Metaculus to the forum, you connect it to https://www.getguesstimate.com/models, as this is much more scalable and clear. 
• thank you for thinking about these things, it inspired me to make my own post. 

FWIW the former CEO of FTX US also claimed this:

In early April 2022, my eleventh month, I made one last try. I made a written formal complaint about what I saw to be the largest organizational problems inhibiting FTX’s future success. I wrote that I would resign if the problems weren’t addressed.

29/49 In response, I was threatened on Sam’s behalf that I would be fired and that Sam would destroy my professional reputation. I was instructed to formally retract what I’d written and to deliver an apology to Sam that had been drafted for me.

The threat model is still unclear, but this is at least somewhat corroborating evidence that Sam is not above using threats in such situations.

I don’t know the answer, but I suggest asking this on LessWrong as well.

(Sorry I took so long to come back) Thanks for clarifying. Hm I'm surprised then that it really seems like the journalist didn't turn up such quotes about fraud. I do think you are right that many of them expected a crash-and-burn... of some sort. I feel like I should have written something more precise like "crash and burn 3 years later, after making 15B on paper" which comes with so many signals over the years that if I were such a person I'd end up discounting my early suspicions. If I were in their or CEA's shoes I'd probably have expected something like what happened with Tara's company, a crash and burn pretty soon after (2019?), so I'd be assuming something got fixed along the way if not. Especially given how an ex-employee(s?) talked about burning through the Asian arbitrage dollars with bad trade decisions.. they'd have had to fix it, right, or they'd have gone belly-up way sooner? I guess crypto was just that much of a gold rush and so few "adults in the room" that they could keep fudging their numbers for that long..?

Maybe the investigation was worse than useless in the end, but reasonably any action taken was going to start with an investigation. It depends the quality of the investigation, but for now I'm much more comfortable considering this a bug of the world than something to blame CEA for.

[Edit: This isn't to say that I think no mistakes were made. But my complaints are not focused on EA leaders specifically (I'm hesitant to call out any single person til CEA's commissioned investigation is complete), and are different from what the article discusses. I discuss that in shortform [EA(p) · GW(p)]]

I value the “it is something that everyone in EA can work on“-sentiment.

Particularily in these times, I think it is excellent to find things that (1) seem robustly good and (2) we can broadly agree on as a community to do more of. It can help alleviate feelings of powerlessness (and help with this is, I believe, one of the things we need.)

This seems to be one of those things. Thanks!