Yeah, it doesn’t (obviously) follow. See the appendix on equality. It made the proof simpler and I thought most readers would not find it objectionable, but if you have a suggestion for an alternate simple proof I would love to hear it!
Ben_West
Karma: 2,174
Thanks!
I don’t think the theorem provides support for total utilitarianism, specifically, unless you add extra assumptions about how to deal with populations of different sizes or different populations generally. Average utilitarianism is still consistent with it, for example.
Well, average utilitarianism is consistent with the result because it gives the same answer as total utilitarianism (for a fixed population size). The vast majority of utility functions one can imagine (including ones also based on the original position like maximin) are ruled out by the result. I agree that the technical result is “anything isomorphic to total utilitarianism” though.
You might be interested in Teruji Thomas’ paper
I had not seen that, thanks!
In that case, it would IMO be better to change “total utilitarianism” to “utilitarianism” in the article. Utilitarianism is different from other forms of consequentialism in that it uses thoroughgoing aggregation. Isn’t that what Harsanyi’s theorem mainly shows?
Hmm, it does show that it’s a linear addition of utilities (as opposed to, say, the sum of their logarithms). So I think it’s stronger than saying just “thoroughgoing aggregation”.
Yeah, my point was that ex-ante utility was valued equally, but I think that was confusing. I’m just going to remove that section. Thanks!
Thanks for the comment!
Also, you suggest that this result lends support to common EA beliefs.
Hmm, I wasn’t trying to suggest that, but I might have accidentally implied something. I would be curious what you are pointing to?
First, it leads to preference utilitarianism, not hedonic utilitarianism
I used preferences about restaurants as an example because that seemed like something people can relate to easily, but that’s just an example. The theorem is compatible with hedonic utilitarianism. (In that case, the theorem would just prove that the group’s utility function is the sum of each individual’s happiness.)
Second, EAs tend to value animals and future people, but they would arguably not count as part of the “group” in this framework(?).
I don’t think that this theorem says much about who you aggregate. It’s just simply stating that if you aggregate some group of persons in a certain way, then that aggregation must take the form of addition.
Third, I’m not sure what this tells you about the creation or non-creation of possible beings (cf. the asymmetry in population ethics).
I agree it doesn’t say much, see e.g. Michael’s comment.
Harsanyi’s simple “proof” of utilitarianism
I expect bets about a large number of Global Catastrophic Risks to be of great importance, and to similarly be perceived as “ghoulish” as you describe here.
The US government attempted to create a prediction market to predict terrorist attacks. It was shut down basically because it was perceived as “ghoulish”.
My impression is that experts think that shutting down the market made terrorism more likely, but I’m not super well-informed.
I see this as evidence both that 1) markets are useful and 2) some people (including influential people like senators) react pretty negatively to betting on life or death issues, despite the utility.
Thanks for the suggestions. There are some community-organized events like meetups or parties in the days around the conference. Due to some past issues (e.g. someone sending every attendee a promotional message about their organization on the event app, or confusion about who is actually present at the event to meet with), we’re wary of expanding app access beyond the actual conference attendees. (See also Ellen’s comment here, which is a somewhat similar idea.)
Thanks for sharing this! It seems like a really exciting project, and I hope you continue to post updates. Very cool that you have explicit success metrics.
A semi-research thing I’m interested in is putting more information on Wikipedia. I wrote a little bit about this here. I suspect that for people who are new to research, or aren’t entirely sure what subject they want to research, making existing research accessible is a similar task which is also quite useful for the world.
Thanks for asking Ozzie! The current bottlenecks limiting our ability to make a larger EA Global are not things that community members can easily help with.
That being said, we recently published a post on other types of events. I would encourage community members to read that and consider doing one-on-one’s, group socials, or other events listed there. Even though EA Global in particular is not something that can be easily scaled by the community, many other types of events can be.
More involved community members may also consider doing a residency. I believe you and I first met when I stayed in the Bay for a few weeks many years ago, and to this day I’m still more closely connected with people I met on that trip than many I met at EA Global.
Thanks for the detailed thoughts Oli.
I think having common knowledge of norms, ideas and future plans is often very important, and is better achieved by having everyone in the same place. If you split up the event into multiple events, even if all the same people attend, the participants of those events can now no longer verify who else is at the event, and as such can no longer build common knowledge with those other people about the things that have been discussed.
Interesting, this doesn’t fit with my experience for two reasons: a) attendance is so far past Dunbar’s number that I have a hard time knowing who attended any individual EA Global and b) even if I know that someone attended a given EA Global, I’m not sure whether they attended any individual talk/workshop/etc. (since many people don’t attend the same talks, or even any talks at all).
I’m curious if you have examples of “norms, ideas, or future plans” which were successfully shared in 2016 (when we had just the one large EA Global) that you think would not have successfully been shared if we had multiple events?
I have been to 3 EAGx events, all three of which seemed to me to be just generally much worse run than EAG, both in terms of content and operations
We have heard concerns similar to yours about logistics and content in the past, and we are providing more support for EAGx organizers this year, including creating a “playbook” to document best practices, having monthly check-in calls between the organizers and CEA’s events team, and hosting a training for the organizers (which is happening this week).
At least in recent years, the comparison of the Net Promoter Score of EAG and EAGx events indicate that the attendees themselves are positive about EAGx, though there are obviously lots of confounding factors:
(More information about EAGx can be found here.)
The value of a conference does scale to a meaningful degree with n^2… I think there are strong increasing returns to conference size
Echoing Denise, I would be curious for evidence here. My intuition is that marginal returns are diminishing, not increasing, and I think this is a common view (e.g. ticket prices for conferences don’t seem to scale with the square of the number of attendees).
Group membership is in significant parts determined by who attends EAG, and not by who attends EAGx, and I feel somewhat uncomfortable with the degree of control CEA has over that
Do you have examples of groups (events, programs, etc.) which use EA Global attendance as a “significant” membership criterion?
My impression is that many people who are highly involved in EA do not attend EA Global (some EA organization staff do not attend, for example), so I would be pretty skeptical of using it.
Meta Note
To clarify my above responses: I (and the Events team, who are currently running a retreat with the EAGx organizers) believe that more people being able to attend EA Global is good, all other things being equal. Even though I’m less positive about the specific things you are pointing to here than you are, I generally agree that you are pointing to legitimate sources of value.
Great – I appreciate your dedication to transparency even though you have so many other commitments!
Thanks for writing this up despite all your other obligations Oli! If you have time either now or when you do the more in-depth write up, I would still be curious to hear your thoughts on success conditions for fiction.
I wanted to share an update: for the past month, our events team (Amy, Barry, and Kate) have been brainstorming ways to allow more people to attend EA Global SF 2020. Our previous bottleneck was the number of seats available for lunch: even with us buying out the restaurant next to the dome (M.Y. China), we only had space for 550 people. (Tap 415, another nearby restaurant which we had used in prior years, has gone out of business.)
We have now updated our agreements with the venue and contractors and brainstormed some additional changes that will allow more attendees in sessions and at lunch. This has increased our capacity by 70 (from 550 to 620).
(As a reference point: EA Global SF had 499 attendees in 2019.)
Thanks for the feedback, Oliver. Do you have opinions on our hypothesis that we should focus on EAGx over more/bigger EA Globals?
We don’t have any current plans to split EA Global into multiple sub-conferences. We have used the fact that not everyone attends talks to increase attendance (for example, at EA Global London 2019, we accepted more attendees than could fit in the venue for the opening talk on the assumption that not all of them would attend the opening).
We will keep the sub-conference idea in mind for the future.
Thanks for the questions. We have adjusted our promotion – for example: the application page and form lists who we believe EA Global to be a good fit for, and we send group leaders an email with this set of criteria and some FAQs about why group members may not be admitted. Conversely, we send emails to people we expect to accept (e.g. Community Building Grant recipients), to encourage them to apply. We try to make community members aware when applications open and convey who the event is aimed at, but we don’t try to promote it as strongly as we did in some past years.
Despite this, we know that there are still many people who would be a good fit for EA Global who do not apply, and others who apply and feel disappointed when they are not accepted. We want to express our appreciation to everyone who applies.
Regarding themes: in 2017 EA Global Boston had a theme of “expanding the frontiers of EA”, EA Global London had an academic theme, and EA Global SF had a community theme and had looser admission standards than the other two. We found that people primarily applied to the conference they were geographically closest to and did not seem to have strong preferences about themes. We’ve also run smaller targeted retreats on specific topics like organizing EA groups or working in operations.
This blog post does some calculations and estimates 250,000 animals per year.
They don’t share the details, but it sounds like a pretty noisy estimate, e.g.:We have to make a number of other assumptions as well, for example, we take the word of business analyst, Andrew Charles, who is widely quoted as suggesting Burger King could sell 50 Impossible Whoppers per day, and apply that same figure to all the QSRs in our model
Thanks, this is worth thinking about. Among similar applicants (e.g. students from the same university group) we’re choosing fairly arbitrarily but we’ve never tried true randomization.
Ah, my mistake – I had heard this definition before, which seems slightly different.
Thanks for the suggestion – always tricky to figure out what a “straightforward” consequence is in philosophy.
I changed it to this – curious if you still find it jarring?