MichaelA

• MichaelA 21 Jan 2021 11:48 UTC

  2 points

  on: Pro­pose and vote on po­ten­tial tags

  EA fellowships

  I think it might be useful to have a post on EA fellowships, meaning things like the EA Virtual Programs, which “are opportunities to engage intensively with the ideas of effective altruism through weekly readings and small group discussions over the course of eight weeks. These programs are open to anyone regardless of timezone, career stage, or anything else.” (And not meaning things like summer research fellowships, for which there’s the Research Training Programs tag.)

  I think this’d be a subset of the Event strategy tag.

  But I’m not sure if there are enough posts that are highly relevant to EA fellowships for it to be worth having this tag in addition to the Event strategy tag. And maybe a somewhat different scope would be better (i.e., maybe something else should be bundled in with this).

• MichaelA 20 Jan 2021 9:16 UTC

  4 points

  on: What is go­ing on in the world?

  Another possible story, which could underpin some efforts along the lines of patient altruism /​ punting to the future: “There will probably be key actions that need taking in the coming decades, centuries, or millennia, which will have a huge influence over the whole rest of the future. There are some potential ways to set up future people to take those actions better in expectation, yet very few people are thinking strategically and working intensely on doing that. So that’s probably the best thing we can do right now.”

  Those “potential ways” of punting to the future could be things like building a community of people with good values and epistemics or increasing the expected future wealth or influence of such people.

  And this story could involve thinking there will be a future time that’s much “higher leverage” /​ more “hingey” /​ more “influential”, or thinking that there are larger returns to some ways of “punting to the future”, or both.

  (See also.)

  (Personally, I find this sort of story at least plausible, and it influences me somewhat.)

• MichaelA 20 Jan 2021 9:07 UTC

  2 points

  on: What is go­ing on in the world?

  Another potential story could go something like this: “Advances in artificial intelligence, and perhaps some other technologies, have begun to have major impacts on the income, wealth, and status of various people, increasing inequality and sometimes increasing unemployment. This then increases dissatisfaction and instability with our political and economic systems. These trends are all likely to increase in future, and this could lead to major upheavals and harms.”

  I’m not sure if all those claims are accurate, and don’t personally see that as one of the most important stories to be paying attention to. But it seems plausible and somewhat commonly believed among sensible people.

• MichaelA 20 Jan 2021 9:04 UTC

  2 points

  on: What is go­ing on in the world?

  AI agents will control the future, and which ones we create is the only thing about our time that will matter in the long run. Major subplots: …

  I think there are plausible and plausibly important plots similar to this, and subplots similar to the subplots below it, but that differ in a few ways from what’s stated there. For example, I think I’m more inclined towards the following generalised version of that story:

  AI systems will control the future or simply destroy our future, and how our actions influence the way that plays out is the only thing about our time that will matter in the long run. Major subplots: …

  This version of the story could capture:

  • The possibility that the AI systems rapidly lead to human extinction but then don’t really cause any other major things in particular, and have no [other] goals

    • I feel like it’d be odd to say that that’s a case where the AI systems “control the future”

  • The possibility that the AI systems who cause these consequences aren’t really “agents” in a standard sense

  • The possibility that what matters about our time is not simply “which [agents] we create”, but also things like when and how we deploy them and what incentive structures we put them in

  One thing that that “generalised story” still doesn’t clearly capture is the potential significance of how humans use the AI systems. E.g., a malicious human actor or state could use an AI agent that’s aligned with the actor, or a set of AI services/​tools, in ways that cause major harm. (Or conversely, humans could use these things in ways that cause major benefits.)

• MichaelA 20 Jan 2021 8:54 UTC

  2 points

  on: What is go­ing on in the world?

  Personally, the simple stories that I pretty much endorse, and that are among the stories within which my choices would make sense, are basically “low-confidence”, “expected value”, and/​or “portfolio” versions of some of these (particularly those focused on existential risks). One such story would be:

  There’s a non-trivial chance that there are risks to the future of humanity (‘existential risks’), and that vastly more is at stake in these than in anything else going on. Meanwhile the world’s thinking and responsiveness to these risks is incredibly minor and they are taken unseriously. So, in expectation, it’d be a really, really good idea if some people acted to reduce these risks.

  (“Non-trivial” probably understates my actual beliefs. When I forced myself to try to estimate total existential risk by 2120, I came up with a very tentative 13%. But I think I might behave similarly even if my estimate was quite a bit lower.)

  What I mean by “portfolio” versions is basically that I think I’d endorse tentative versions of a wide range of the stories you mention, which leads me to think there should be at least some people focused on basically acting as if each of those stories are true (though ideally remembering that that’s super uncertain). And then I can slot into that portfolio in the way that makes sense on the margin, given my particular skills, interests, etc.

  (All that said, I think there’s a good argument for stating the stories more confidently, simply, and single-mindedly for the purposes of this post.)

• MichaelA 20 Jan 2021 8:43 UTC

  4 points

  on: What is go­ing on in the world?

  The following statements from Luke Muehlhauser feel relevant:

  Basically, if I help myself to the common (but certainly debatable) assumption that “the industrial revolution” is the primary cause of the dramatic trajectory change in human welfare around 1800-1870, then my one-sentence summary of recorded human history is this:

  >Everything was awful for a very long time, and then the industrial revolution happened.

  (The linked post provides interesting graphs and discussion to justify/​flesh out this story.)

  Though I guess that’s less of a plot of the present moment, and more of a plot of the moment’s origin story (with hints as to what the plot of the present moment might be).

• MichaelA 20 Jan 2021 8:41 UTC

  3 points

  in reply to: Linch’s comment on: What is go­ing on in the world?

  Yeah, I agree with that.

  On this, I really like this brief post from Our World in Data: The world is much better; The world is awful; The world can be much better. (Now that I have longtermist priorities, I feel like another useful slogan in a similar spirit could be something like “The world could become so much better; The world could end or become so much worse; We could help influence which of those things happens.”)

• MichaelA 19 Jan 2021 13:34 UTC

  4 points

  in reply to: meerpirat’s comment on: Pro­pose and vote on po­ten­tial tags

  This seems plausibly useful to me.

  Obviously it’d overlap a lot with the Forecasting tag. But if it’s the case that several posts include Elicit forecasts but most posts tagged Forecasting don’t include Elicit forecasts, then I imagine a separate tag for Elicit forecasts could be useful. (Basically, what I’m thinking about is whether there would be cases in which it’d be useful for someone to find /​ be sent a collection of links to just posts with Elicit forecasts, with the Forecasting tag not covering their needs well.)

  But maybe a better option would be to mirror LessWrong in having a tag for posts about forecasting and another tag for posts that include actual forecasts (see here)? (Or maybe the latter tag should only include posts that quite prominently include forecasts, rather than just including them in passing here and there.) Because maybe people would also want to see posts with Metaculus forecasts in them, or forecasts from Good Judgement Inc, or just forecasts from individual EAs but not using those platforms. And I’d guess it’d make more sense to have one tag where all of these things can be found than to try to have a separate tag for each.

  (That’s just my quick thoughts in a tired state, though.)

  It could also be handy to have a tag for posts relevant to “Ought /​ Elicit”—I think it’d probably be good to bundle them together but note Elicit explicitly—similarly to how there’s now tags for posts relevant to each of a few other orgs (e.g. Rethink Priorities, FHI, GPI, QURI). So maybe the combination of a tag for posts that contain actual forecasts and a tag for Ought /​ Elicit would serve the role a tag for posts containing Elicit forecasts would?

• MichaelA 19 Jan 2021 13:28 UTC

  2 points

  in reply to: Aaron Gertler’s comment on: Should pretty much all con­tent that’s EA-rele­vant and/​or cre­ated by EAs be (link)posted to the Fo­rum?

  That all makes sense to me—thanks!

• MichaelA 19 Jan 2021 2:28 UTC

  3 points

  on: Train­ing Bot­tle­necks in EA (pro­fes­sional skills)

  One small point of superficial feedback on this post: I feel like it could be useful to change the title to something like “Professional Skills Training Bottlenecks in EA”, or “Professional Skills Training: A Bottleneck in EA”?

  I say this because I think training bottlenecks for things other than professional skills—e.g., learning about EA concepts or academic fields or research skills—are also quite important topics that are often discussed on the Forum. (E.g., in the posts tagged Research Training Programs.) And I think it’d also be good to someday have a post collecting ideas for addressing that broader set of training bottlenecks in EA.

  So even though at the start you said “I’m focusing on professional skills such as fundraising or management, rather than learning about concepts in effective altruism, for which there seems to be a number of excellent programs happening”, I think I still kept sort-of forgetting that the scope was intended to be limited in that way as I read the rest of the post. And I think that might’ve been due to how the title set my expectations.

  (Though maybe this is just me and my sleepy brain!)

• MichaelA 19 Jan 2021 2:14 UTC

  7 points

  on: Train­ing Bot­tle­necks in EA (pro­fes­sional skills)

  On the point of finding useful courses, Edo Arad asked earlier What are some good online courses relevant to EA?, mentioned a couple, and got a few more answers. Though most weren’t focused on “professional skills such as fundraising or management”.

  Maybe if people have additional course ideas, they could mention them as answers to that question post? Or maybe someone should make a new post to collect such course suggestions?

  (This would be very much only a partial solution, but could be helpful.)

  What links here?

  • What are some good on­line courses rele­vant to EA? by EdoArad (14 Apr 2020 8:36 UTC; 16 points)

• MichaelA 19 Jan 2021 2:10 UTC

  2 points

  on: Train­ing Bot­tle­necks in EA (pro­fes­sional skills)

  Thanks for this post.

  Some thoughts on the “Systematising mentorship” stuff:

  • My impression is that WANBAM and Effective Thesis have essentially done some things that are similar to some of the things proposed in this section of this post. I imagine other EA- or EA-aligned orgs/​people might’ve done so as well, in addition to presumably many non-EA orgs/​people. So there might be a lot that could be learned from their approaches, successes, failures, thinking, etc.

  • In a recent podcast interview, Rob Wiblin and Spencer Greenberg proposed a potentially easier way to capture part of the benefits that “Systematising mentorship” would also aim to capture: Just setting up weekly meetings with someone else who’s at roughly the same level of seniority and who also wants more “management”/​”mentorship”

    • I summarised some of what they said about that here

    • This of course wouldn’t capture all the benefits of getting mentorship from a person with more expertise in an area than the mentee has, but it might capture the benefits that basically just require any “line manager” type person

  • “A paid mentorship program could harm other mentorship programs and organic mentoring in the community by setting up an expectation that people providing mentorship be paid.” That does seem to me plausible and worth noting.

    • But it also seems plausible that a paid mentorship could lead to people coming to see mentorship in general (even when provided for free) as more a more valuable, desired, “substantial” way of being helpful and having impact. And that could perhaps lead to more mentorship being offered, it being offered by higher-calibre mentors, or mentors being more motivated (again, even when the mentorship is free.)

    • On the other hand, the existence of some paid mentorship could also lead to people implicitly assuming that free mentorship is lower quality or something like that, which could have bad effects (e.g., leading to less supply of and demand for free mentorship).

    • (I’m therefore not sure what the net effect of this consideration would be.)

• MichaelA 19 Jan 2021 0:45 UTC

  4 points

  in reply to: Aaron Gertler’s comment on: Should pretty much all con­tent that’s EA-rele­vant and/​or cre­ated by EAs be (link)posted to the Fo­rum?

  I don’t especially care whether the content is written by community members, though I suppose that’s slightly preferable (as community members are much more likely to respond to comments on their work).

  >an article on nuclear risk from a non-EA academic

  Heck yeah.

  The main reason I sort-of suggest “written by community members” as a possible criterion for deciding whether to linkpost things here is that it seems like, without that criterion, it might be very hard to decide how much to linkpost here. There are huge numbers of articles on nuclear risk from non-EA academics. If someone decided to linkpost all of them here, or to linkpost all of the peer-reviewed non-EA articles on any one of many other EA-relevant topics, that batch of linkposts might suddenly become a large fraction of all posts that year.

  We could go with something like “linkpost all especially high quality articles on nuclear risk that are especially relevant to the most extreme risk scenarios (not just e.g. the detonation of 1 or a few bombs by terrorists)”. But that’s a murkier principle, and it seems like it could easily end up “going too far” (or at least seeming weird). And I think worrying that I’m going too far might lead me to hold back more than is warranted.

  Maybe this could be phrased as “Making the decision partly based on whether the content was created by an EA could help in establishing a Schelling fence that avoids a slippery slope. And the existence of that fence could help people be more comfortable with beginning to travel down the slope, knowing they won’t slip too far.”

• MichaelA 18 Jan 2021 4:02 UTC

  3 points

  in reply to: kokotajlod’s comment on: Les­sons from my time in Effec­tive Altruism

  (Btw, this post and comment thread has inspired me to make a question post to hopefully collect links and views relevant to how much time EAs should spend engaging with people inside vs outside the EA community.)

• MichaelA 18 Jan 2021 3:55 UTC

  2 points

  on: How much time should EAs spend en­gag­ing with other EAs vs with peo­ple out­side of EA?

  Some factors that could influence answers to this cluster of questions:

  (This is just a very quick, non-exhaustive list, with very little commentary. I’m hoping other people can add a bunch more factors, commentary, etc.)

  • Concerns about value drift

  • The value of bringing people who are currently outside of EA into the EA movement

  • The value of introducing people who are (and might remain) outside of EA to EA ideas

    • The post Community vs Network feels somewhat relevant here

  • The value of getting a diversity of views

    • The post Community vs Network feels somewhat relevant here too

  • How much a person should be focused on working at EA vs non-EA orgs

    • There are various ways this could be relevant

      • One is that, if one is more focused on working at non-EA orgs, that might increase the value of building strong networks in specific communities outside of EA (e.g., in specific policy or academic communities).

  • Specific things to do with personal preferences and circumstances

• MichaelA 18 Jan 2021 3:49 UTC

  2 points

  on: How much time should EAs spend en­gag­ing with other EAs vs with peo­ple out­side of EA?

  Some related questions:

  How much should EAs focus on working at EA orgs vs at other orgs?

  • And how does this vary based on various aspects of the individual/​situation in question? And what are the factors driving answers to these questions?

  • We could also ask about working in roles explicitly in vs not explicitly in the EA community, if we wanted to also capture independent work, founding new orgs, etc.

  • Prior discussion:

    • There’s a lot, and I haven’t taken the time to compile it here

    • One recent discussion occurred in this post and some of the comments on it: My mistakes on the path to impact

  How much time should EAs spend consuming content (e.g., papers, articles, podcasts) created by EAs vs content not created by EAs?

  • I assume this has been discussed before in various places, though I can’t immediately recall where

  (Presumably many other questions could also be mentioned.)

• MichaelA 18 Jan 2021 3:42 UTC

  2 points

  on: How much time should EAs spend en­gag­ing with other EAs vs with peo­ple out­side of EA?

  Some prior statements/​discussion of this cluster of questions:

  • As noted, a recent post by Richard Ngo argued that he should’ve engaged more with people outside the EA community in prior years

  • Daniel Kokatajlo commented on that same post saying that the opposite seems true for Daniel

  • I think this post was relevant: Objections to Value-Alignment between Effective Altruists

  • I seem to recall there being relevant parts of the 80,000 Hours anonymous answers series

    • I just very quickly skimmed through those posts again, and this post definitely has relevant parts, while the following also seem decently likely to:

      • https://​​80000hours.org/​​2020/​​04/​​anonymous-answers-diversity/​​

      • https://​​80000hours.org/​​2020/​​02/​​anonymous-answers-effective-altruism-community-and-growth/​​

      • https://​​80000hours.org/​​2019/​​11/​​anonymous-answers-most-overrated/​​

      • https://​​80000hours.org/​​2020/​​02/​​anonymous-answers-general-life-advice/​​

      • https://​​80000hours.org/​​2020/​​02/​​anonymous-answers-flaws-80000hours/​​

      • https://​​80000hours.org/​​2020/​​06/​​anonymous-answers-myths-and-other-questions/​​

  (I’m sure there are many other relevant links that have slipped my mind or that I never saw, which is why this is a question post rather than just a post!)

[Question] How much time should EAs spend en­gag­ing with other EAs vs with peo­ple out­side of EA?

• MichaelA 18 Jan 2021 2:25 UTC

  5 points

  in reply to: MichaelA’s comment on: Les­sons from my time in Effec­tive Altruism

  A related matter: I get the impression from the section “2. I should have prioritised personal fit more” that you (Richard) think it would’ve been better if you’d skipped trying out engineering-style roles and gone straight into philosophy-style roles. Do you indeed think that?

  It seems plausible that going in an engineering direction for a couple years first was a good move ex ante, because you already knew you were a fit for philosophy but didn’t know whether you were a fit for things more along the lines of engineering? So maybe it was worth checking whether something else was an even better fit for you, or whether something else was a good enough fit that your comparative advantage (including your interest as a factor) would be things that somehow draw on both skillsets to a substantial degree?

  I.e., even if ex post it appears that “exploiting” in the philosophy path is the best move, perhaps, ex ante, it was worth some exploration first?

  (Of course, I don’t know the details of your career, plans, or your own knowledge several years ago of your skills and interests. And even if the answers to the above questions are basically “yes”, it’s still plausible that it would’ve been better to explore for less time, or in a way more consciously focused on exploration value—which might’ve entailed different roles or a different approach.)

• MichaelA 18 Jan 2021 2:12 UTC

  11 points

  in reply to: kokotajlod’s comment on: Les­sons from my time in Effec­tive Altruism

  I very much agree with Daniel’s paragraph on personal fit.

  The message on personal fit I got from 80k was basically “Personal fit is really important. This is one reason to find an area and type of work that you’ll be very passionate about. But that doesn’t mean you should just follow your current passion; you might discover later that you don’t remain passionate about something once you actually do it for a job, and you might discover that you can become passionate about things you aren’t yet passionate about, or haven’t even heard of. So it’s generally good to try to explore a lot early on, figure out what you have strong personal fit for, and then do one of those things.”

  I wrote that from my remembered impression, but then googled “80,000 Hours personal fit”, and here’s part of their summary from 2017 career guide article on “How to find the right career for you” (which was the first hit):

  • Your degree of personal fit in a job depends on your chances of excelling in the job, if you work at it. Personal fit is even more important than most people think, because it increases your impact, job satisfaction and career capital.

  • Research shows that it’s really hard to work out what you’re going to be good at ahead of time, especially through self-reflection.

  • Instead, go investigate. After an initial cut-down of your options, learn more and then try them out.

  I think that that’s great advice, and has been really helpful for me. There are some things I thought I’d like/​be good at but wasn’t, and vice versa. And there are many things I hadn’t even considered but turned out to like and be good at.

  Unfortunately, it seems like it’s common for people to round off what 80k said to “Personal fit and passion don’t matter”, even thought they explicitly argue against that. (80k’s 2014-2015 review does say that they think they previously hadn’t emphasised personal fit enough; perhaps this common misinterpretation can be traced back to ripple effects from 80k’s early messaging?)

  Of course, it’s still necessary to figure out precisely how important personal fit and interest are relative to other things, and so it’s still possible and reasonable for someone to “emphasise personal fit and interest significantly more than 80k does, when giving career advice”. But I’m pretty confident that 80k would already agree, for example, that “passion about a field is a very important component of doing world-class research in it”.