Hide table of contents

Here's an example of what I'm asking about:

I just came across this comment by Geoffrey Miller from 9 hours ago:

Once the FTX debacle gets resolved, and the PR blowback dies down, it might be useful for EAs to try to nudge Bezos to broaden his cause areas -- on the principle that tackling climate change may be moderately important in scope and severity, but is far from neglected, and seems rather intractable politically.

It would be great to see some of the Amazon money going to higher-impact cause areas.

The comment is essentially a truism within EA, yet as of this moment it has 9 karma with 11 votes and only 2 agreement karma with 7 votes, meaning a nontrivial fraction of EA Forum users voting on it disagreed with it.

It's not very often that I come across posts/comments on the Forum where both (1) the karma/agreement voting behavior seems very off the mark from what it seems to me it should be, and (2) I further don't have a good explanation for why the current vote is what it is.

However, in the last few days since the FTX news broke I have noticed more than a few unexpected karma/agreement vote counts (though I didn't save any examples), hence why I'm asking this now.

Has anyone else noticed karma/agreement votes being off from what you'd expect recently too? In particular, have you seen more comments with karma and disagreement downvotes despite no obvious reason (or commented justification) for disagreement?

Or am I wrong in suspecting that voting patterns may have changed recently? (I'm only basing this suspicion based on a few surprising examples in the last few days.)

If it seems like voting behavior has changed/degraded, is it a very recent phenomena (since the FTX news broke?), or has this been a more gradual change that you've noticed for a while?




New Answer
New Comment

4 Answers sorted by

As a new user I am baffled (and slightly amused) by the fact that this comment is at -1/-9:


In the comment, a) I refer to an allegation that has been made about the FTX team, and b) make the extremely obvious point that if such an allegation were true and known, people would have been much more reluctant to place their trust in FTX.

What baffles me is that comments making an essentially identical point did not receive negative votes, and in some cases received strong positive votes.

I'm guessing it got downvoted because it was interpreted as an attack on polyamory, or relationships between EAs in general, coming from someone new to/outside the community, and didn't spell out -- when it should be obvious(!) -- that the reason that it's a problem is conflict of interest (I spelled it out elsewhere in the thread and got upvoted. Shame that there is this fairly widespread blindspot in EA when it comes to conflicts of interest).

Paul Currion
Yes, that sounds plausible. Perhaps I should have been clearer that it wasn't an attack on polyamory, but a red flag indicating a high possibility of conflict of interest regardless of whether Ellison was in a romantic or sexual relationship with any of the other residents. The fact that you needed to spell it out in your comment (which was one of the comments I was referring to) is concerning, but I'll take more care in the future to spell out any points I',m trying to make.

In said comment, you say:

An unverified suggestion in one article suggests that Ellison may have been in some form of relationship with other residents of the apartment

Are you referring to this article?

All 10 are, or used to be, paired up in romantic relationships with each other. That includes Alameda CEO Caroline Ellison, whose firm played a central role in the company's collapse – and who, at times, has dated Bankman-Fried, according to people familiar with the matter.

This doesn't read to me like a suggestion that Caroline was "in some form of relationsh... (read more)

Paul Currion
Thanks for the reply. It's not starting rumors to refer to information that is in the public domain, especially if that reference is heavily qualified as "an unverified suggestion" and the following point is clearly preceded by "If true". Per Greg_Colbourn's reply to me, however, I recognise that there is sensitivity about polyamory on this board. I deliberately didn't refer to polyamory because it wasn't relevant to my point, which is that any kind of relationship between Ellison as CEO of Alameda and any senior staff member of FTX - and to be honest, even simply co-habitation - should be a huge red flag to anybody entrusting their money. Other comments have communicated that better than mine!

This is interesting. Hopefully there are indeed changes happening within EA, not only concerning voting patterns but also real epistemic and epistemological changes. Miller's comment, for example, seems to me like an excellent example of an intellectual over-confidence which has been all too common in EA circles - that reasoning from first principles can lead to better conclusions than painstakingly following empirical evidence or established best-practices in relevant knowledge communities. We can for example be very certain by now that climate change is not only "moderately severe", but an extremely severe challenge. The claim that it is "politically intractable" is not, AFAIK, based on empirical evidence - there are clear differences between countries and polities concerning climate policies and their effectiveness. The claim that it is "far from neglected" is also not in line with survey evidence and qualitative research on societal elites, which shows that it's quite often far down on the priority list for many people. There are therefore good reasons to believe that the position which is here referred to as a truism within EA circles is mistaken. 

And Bezos is no idiot. Ok, he's doing lots of bad things in his business practice. But he is no idiot, and we can have fairly high confidence that he has discussed his future funding with very smart people. Still, some EA people still assume that they just know better, and are in a position to provide advice to Bezos on what to do. This is even more questionable after the complete disaster of FTX, where very important EA leaders displayed spectacular failures of judgment.

So the assumption that EAs are now in a position to weigh in on optimal allocation of funding for Bezos, away from climate change, strikes me as odd and not intellectually justified. 

So my hope is that new voting patterns in this case indeed reflect an intellectual reorientation among some EAs, hopefully in the direction of more intellectual humility.

Thanks for the helpful answer; this gives me some insight into the vote in this case!

To share my thoughts on this point with you:

The claim that it is "far from neglected" is also not in line with survey evidence and qualitative research on societal elites, which shows that it's quite often far down on the priority list for many people.

If you compare global annual spending on climate change to global annual spending on things like risks from AI, nuclear risk, or farmed animal welfare, it definitely seems to me like climate change is relatively extremely cro... (read more)

Your claim about climate change being not neglected be true but my guess is that many users of this forum disagree with it. Thus I guessed that the downvoting of Geoffrey’s commwnt was good faith disagreeing that climate change was nonneglected and worth steering Bezos away from. I.e. people who support more funding for climate stuff.
Good point. I assumed it was good faith disagreement, but surprising disagreement nonetheless. But perhaps I shouldn't be too surprised. If I recall correctly, the EA Survey consistently finds that a significant minority of EAs rate climate change as a high priority, and with so many new comers and the popularity of climate change as an issue among people who want to have a social impact in general, I suspect you're right there are a significant number of Forum users who think climate change is neglected and a top or near-top priority.
Thanks, that's a thoughtful response!  Let me say that straight away that I'm very much in favor of work on nuclear risk, animal welfare, poverty reduction etc (on AI I struggle to see an established mechanism between the work that is being done and the desired outcome, but I'm very open that I may be wrong on that one). Just to expand a bit on spending and cost-effectiveness on climate change vs. other causes: It all depends what you mean. Climate change is not one single issue, given that it's about our whole way of production and consumption. Reducing animal agriculture is also very important for fighting climate change, for example, not just for animal welfare. But if we break it down to different domains, we still see a lack of funding in many areas:  * renewables - not to mention nuclear energy, which I love! - still gets much less investment and funding than fossil fuels. Collecting data on this is not easy, of course. But one of the most thorough reports I know of is for funding and financing in the Netherlands, one of the greenest countries on earth. It revealed that fossil fuels received much more funding than renewables Financing of fossil fuels vs renewables by Dutch FIs EGW 211022_FIN (eerlijkegeldwijzer.nl) * Fossil fuel companies have spent enormous amounts of money on lobbying. There is no comparable funding for climate change lobbying, not even close Fossil Fuel Industry Has Spent Nearly $2 Billion on Lobbying to Kill Climate Laws - Oil Change International (priceofoil.org) * Fossil fuel companies have spent enormous amounts of money on advertising. There is no comparable funding for climate change advertising. Fossil Fuel Trade Associations Spent $1.4 Billion on Ads in Past Decade (therealnews.com) * Concerning activism and political fights, fossil fuel companies generally donate a whole lot to PACs and politicians etc, and there is no comparable funding for green politicians. Fossil fuel political giving outdistances renewables 13 to one »
It's fine; this post left the front page in about an hour so I expect new people joining will be limited. I will refrain from responding to your comment at length due to not wanting to stray far off topic, but know that I did read it, and my one sentence response is that I think the amount of money Bezos pledged toward climate change (not mentioning the far greater amount of money that has already been spent on it (Edit: >$300 billion annually) is more than an order of magnitude more money than the total amount of money that has ever been spent on AI safety (Edit: $10 million annually, same source), so I'm not very persuaded by your comparison of some areas of climate change to fossil fuel spending.

I'm pretty sure I've seen people here arguing that the FTX case means that EA shouldn't get involved with billionaires again. There's plausibly an even larger group who think Geoffrey's comment may be correct but "too soon."

(I hadn't voted, I'm just reading it now and hazarding a guess.)

Here's a meta-answer for you to agreement-karma vote on (i.e. use the "X" and "check mark," (which affects "agreement karma") not the "<" and ">" (which affect "overall karma" also called just "karma"):

"Agree" with this if it seems like there has been some degradation in EA Forum user voting behavior over the last year or so.

"Disagree" if it seems like no degradation in EA Forum user voting behavior has taken place in the last year.

Strong Agree/Disagree only if you're particular confident in your observations.

In particular, don't Agree or Disagree to adjust the overall Agreement karma to what you think it should be. The point is to use this as a poll, not for the final Agreement karma to necessarily reflect the truth (which you may know more about than other Forum users).

Abstain from voting if you're not sure or object to my framing in this poll or my claim that the example I gave in the question is indicative of poor voting behavior.

My meta-answer currently has 3 karma with 17 votes despite my explanation of how to do agreement-karma voting.

Did people downvote my answer intentionally for some reason?

Or  does this show that a lot of EA Forum users voting somehow still don't understand the difference between regular karma and agreement karma and downvoted thinking that was how to participate in the poll?

Edit: My guess is the latter, based on priors and the fact that the top level post (the question) itself doesn't have downvotes (14 karma with 6 votes) and none of the presumably several people who downvoted the meta-answer commented to explain why.

I think there have been some weird examples of voting recently, but this isn't one! I've long been annoyed with the naivite of EAs who think unsolicited outreach to billionaires will have a positive impact on the world.

Thanks! That's a fair reason for a downvote / disagreement vote I hadn't considered in this case.
Sorted by Click to highlight new comments since:

This seems like progress to me. Something highly upvoted but disagreement-downvoted means to me "we appreciate this comment's existence and want to incentivize that, but disagree with it factually". I think voting has degraded due to the sheer influx of people and content, but that this feature is swimming uphill against that and has noticeably improved discourse.

I also want to emphasize that the main thing I wanted to highlight with the example was not that the regular karma was downvoted, but rather that the disagreement karma was, and that the content of the comment was not one that I was pretty confident an informed EA wouldn't disagree with (namely that climate change is not neglected and that directing money to more effective causes was valuable.) 

I saw this as another example of a pattern I think I may have perceived of "agreement-karma voting degradation." My impression was that the voting patterns I've seen (lower-quality agreement-karma voting) is evidence that EA Forum voters are less familiar with EA ideas than they once were.

Note though that Kirsten pointed out an alternative hypothesis for disagreement in her answer: "I think there have been some weird examples of voting recently, but this isn't one! I've long been annoyed with the naivite of EAs who think unsolicited outreach to billionaires will have a positive impact on the world."

So I've updated toward thinking that the example I gave as evidence of my perceived trend in bad agreement-karma voting wasn't actually a good example. In this case, the disagreement votes could be explained by people with Kirsten's view, rather than people disagreeing that climate change was neglected. (Though there are also EA Forum who do think climate change is neglected, per other answers this question received.)

Something highly upvoted but disagreement-downvoted means to me "we appreciate this comment's existence and want to incentivize that, but disagree with it factually". 

I agree with this sentence.

But it doesn't apply in this case because:

The example comment I gave was not "highly upvoted but disagreement-downvoted". Specifically, it was downvoted rather than "highly upvoted." (It had 9 karma with 11 votes meaning it was downvoted. Highly upvoted comments have much more karma than votes. Comments with equal karma to votes or less (as was in this case) were downvoted (since many users' votes necessarily deliver >1 karma).)

I just came across this. Same article, different voting behaviour.

https://forum.effectivealtruism.org/posts/6aXfG57FvbcrKf9Qd/make-the-drought-evaporate – 8 karma from 10 votes (no critical comments) https://www.lesswrong.com/posts/wttyWyKqkPyncuoCm/make-the-drought-evaporate – 32 karma from 24 votes

Curated and popular this week
Relevant opportunities