MichaelPlant

• MichaelPlant 15 Aug 2022 16:29 UTC

  6 points

  on: The Parable of the Boy Who Cried 5% Chance of Wolf

  Not sure if this is essential to the parable, but wouldn’t it be useful to distinguish between the following cases?:

  (1) the boy says every day there’s 5% chance the wolf could come every evening, but isn’t saying the wolf is there right now

  and

  (2) the boy says there is a wolf in the village whenever he thinks there’s a 5+% chance this is true.

  If the boy is doing (1) and the villagers panic now then they’ve just misunderstood what he’s saying. If the boy is doing (2), then you’d understand why the villagers would start ignoring the boy (just like everyone ignores car alarms because they are so oversensitive).

  I’m not sure either is neatly analogous to the X-risk case, which is that different people give different estimates and these range from negligible to doom being apparently virtually certain. I guess that’s a bit like there being many different boys in the village, each of whom assigns a different percentage chance to the wolf appearing at some point (but none of whom are claiming it’s literally here now).

• MichaelPlant 13 Aug 2022 12:29 UTC

  16 points

  on: Com­mon-sense cases where “hy­po­thet­i­cal fu­ture peo­ple” matter

  I found this discussion, and these cases, objectionably uncharitable. It doesn’t offer the strongest version of person-affecting views, explain why someone might believe them, then offer the objections and how the advocate of the view might reply. It simply starts by assuming a position is true and then proposes some quick ways to persuade others to agree with it.

  An equivalent framing in a different moral debate would be saying something like “people don’t realise utilitarianism is stupid. If they don’t realise, just point out that utilitarians would kill someone and distribute their organs if they thought it would save more lives”. I don’t think the forum is the place for such one-sidedness.

• MichaelPlant 12 Aug 2022 16:35 UTC

  3 points

  in reply to: SiebeRozendal’s comment on: To WELLBY or not to WELLBY? Mea­sur­ing non-health, non-pe­cu­niary benefits us­ing sub­jec­tive wellbeing

  Yes for both a) and b). But the strategy is secret...

  (Basically, the idea is to show using measures of how people feel can and will give different priorities and therefore we should pay more attention to it)

• MichaelPlant 12 Aug 2022 16:31 UTC

  7 points

  in reply to: Vanessa’s comment on: To WELLBY or not to WELLBY? Mea­sur­ing non-health, non-pe­cu­niary benefits us­ing sub­jec­tive wellbeing

  I’m not sure I understand your point. Kahneman famously distinguishes between decision utility—what people do or would choose—and experience utility—how they felt as a result of their choice. SWB measures allow us to get at the second. How would you empirically test which is the better measure of preferences?

To WELLBY or not to WELLBY? Mea­sur­ing non-health, non-pe­cu­niary benefits us­ing sub­jec­tive wellbeing

• MichaelPlant 28 Jul 2022 17:51 UTC

  5 points

  in reply to: Stefan_Schubert’s comment on: Re­mu­ner­a­tion In Effec­tive Altruism

  Hmm, I don’t think you’ve engaged with my point: there’s something odd about very altruistically capable people requiring very high salaries, lest they choose to go and do non-impactful jobs instead. The charity section famously has lower salaries because the work is more intrinsically rewarding than regular corporate fare.

  The salaries might have an effect, but I don’t think you’ve shown that in this case—the linked tweet is anecdata. A possibility is that higher salaries in one EA org just pull the better candidates to that org. So I want evidence showing it is pulling in ‘new’ candidates.

  I’m not sure about the fraction of monetised impact bit of that relevant. As someone who runs an org, I only have access to my budget, not the monetised impact—a job might have ‘£1m a year of impact’, but that’s um, more than 4x HLI’s budget. For someone with enormous resources, eg Open Phil, it might make more sense to think like this.

  Of course, it might be that we just have different meanings of ‘high’ and I would have welcomed if you’d offered a operationalisation in your discussion. I’m not sure I disagree with your conclusion, I just don’t think you’ve proved your case.

• MichaelPlant 28 Jul 2022 4:51 UTC

  8 points

  on: Re­mu­ner­a­tion In Effec­tive Altruism

  Stefan, your most important argument seems to be that higher salaries will help with recruitment and motivation. But you don’t address the concern there’s something a bit puzzling about the most competent effective altruists being motivated by making money for themselves.

  If someone says “look, I’ll do the work, and I will be excellent, but you have to pay me $150k a year or I walk” I would doubt that were that serious about helping other people. They’d sound more like your classic corporate lawyer than an effective effective altruist.

• MichaelPlant 28 Jul 2022 3:03 UTC

  5 points

  in reply to: Mo Putera’s comment on: De­worm­ing and de­cay: repli­cat­ing GiveWell’s cost-effec­tive­ness anal­y­sis

  Ah, I was waiting for someone to bring these up!

  On cluster Vs sequence, I guess I don’t really understand what the important distinction here is supposed to be. Sometimes, you need to put various assumptions together to reach a conclusion—cost-effectiveness analysis is a salient example. However, for each specific premise, you could think about different pieces of information that would change your view on it. Aren’t these just the sequence and cluster bits, respectively? Okay, so you need to do both. Hence, if someone were to say ‘that’s wrong—you’re using sequence thinking’ and I think the correct response is to look at them blankly and say ‘um, okay… So what exactly are you saying the problem is’?

  On cost-effectiveness, I’m going to assume that this is what GiveWell (and others) should be optimising for. And if they aren’t optimising for cost-effectiveness then, well, what are they trying to do? I can’t see any statement of what they are aiming for instead.

  Also, I don’t understand why trying to maximize cost-effectiveness will fail to do so. Of course, you shouldn’t do naive cost-effectiveness, just like you probably shouldn’t be naive in general.

  I appreciate that putting numbers on things can sometimes feel like false precision. But that’s a reason to use confidence intervals. (Also, as the saying goes, “if it’s worth doing, it’s worth doing with made up numbers”). Clearly, GiveWell do need to do cost-effectiveness assessments, even if just informally and in their heads, to decide what their recommendations are. But the part that’s just as crucial as sharing the numbers is explaining the reasons and evidence for your decision so people can check them and see if they agree. The point of this post is to highlight an important part of the analysis that was missing.

De­worm­ing and de­cay: repli­cat­ing GiveWell’s cost-effec­tive­ness anal­y­sis

• MichaelPlant 25 Jul 2022 10:47 UTC

  5 points

  on: The Charle­magne Effect: The Longter­mist Case For Neartermism

  [I read this quickly, so sorry if I missed something]
  
  First, on the scope of the argument. When you talk about “Traditional neartermist interventions”, am I right in thinking you *only* have life-saving interventions in mind? Because there are “traditional neartermist interventions”, such as alleviating poverty, that not only do not save lives, but also do not appear to have large effects on future population size.

  If your claim applies only to the traditionally near-term interventions which increase the population size (in the near term), then you should make that clear. (Aside, I found the “TLIs” and “TNIs” thing really confusing because the two are so similar and this would be a further reason to change them).

  Second, on your argument itself. It seems to rest on this speculative claim.

  Those children will go on to have children, and those children will have children, and so on, following an exponential curve. The more time goes by, the more the number of descendants will accelerate until there are millions, billions, or even more future people.

  But, as an empirical matter, this is highly unlikely. If you look at the UN projections for world population, it looks like this will peak around 2100. As countries get richer, fertility—the number of children each woman has—goes down. Fertility rates are below 2 in nearly every rich country (e.g. see this), which is below the ‘Replacement level fertility’ of 2.1, the average number of children per woman you need to keep the population constant over time. As poorer countries get richer, we can expect fertility to come down there too. So, your claim might be true if the Earth’s population was set to keep growing in perpetuity, but it’s not. On current projections, saving a life in a high fertility county might lead to one to two generations of above replacement level fertility. (You mention interstellar colonisation, but it’s pretty unclear what relationship that would have with Earth’s population size).

  All this is really before you get into debates about optimum population, a topic there is huge uncertainty about.

  

• MichaelPlant 23 Jul 2022 11:40 UTC

  15 points

  on: Lean­ing into EA Disillusionment

  I really appreciated you writing this up. A couple of thoughts.

  First, a useful frame: when people are unhappy about a group, any group, they have two choices. They can either ‘exit’ or ‘voice’. This was originally discussed in relation to consumers buying goods, but it’s general and as simple as it sounds: you leave or you complain. Which of those people do is more complicated: it depends, afaict, on how loyal they are to the group and also whether they think voicing their concerns will work.

  It seems like what you’re saying is that people come into EA, become loyal to the idea, get disillusioned because the reality doesn’t live up to the hype, but then exit rather than voice. The reason they exit is some combination not wanting to harm the project or be seen as a bad actor, and because they don’t think their criticisms will be listened to.

  Second, lots of this really resonates with me. EA sort of sells itself as being full of incredibly smart, open-minded, kind, dedicated people. And, for the most part, they are—at least by the starts of the rest of the world. But they are people still: prone to ego, irrationality, distraction, championing their pet projects, sticking to their guns, and the rest. And these people work together in groups we call ‘organisations’. And even with the best people, getting them to work together and work differently is a struggle… It is a recipe for disillusionment.

  (I recognise I’m not offering any solutions here, sorry...)

• MichaelPlant 23 Jul 2022 11:02 UTC

  29 points

  in reply to: Askell’s comment on: Lean­ing into EA Disillusionment

  I don’t buy this. Perhaps I don’t understand what you mean.

  To press, imagine we’re at the fabled Shallow Pond. You see a child drowning. You can easily save them at minimal cost. However, you don’t. I point out you could easily have done so. You say “I know but I don’t want to do that”. I wouldn’t consider that a satisfactory response.

  If you then said “I don’t need to justify my choices on effective altruist grounds” I might just look at your blankly for a moment and say “wait, what has ‘effective altruism’ got to do with it? What about, um, just basic ethics?” Our personal choices often do or could affect other people.

  I don’t think that people should endlessly self-flaggelate about whether they are doing enough. You need to recognise it’s a marathon, not a sprint, and there are serious limits to what we can will ourselves to do, even if think it would be a good idea in principle. And it’s important to be as kind, forgiving, and accepting to ourselves as we think we should be to others we love. But what youve said, taken as face value, seems carte blanche for not trying.

• MichaelPlant 20 Jul 2022 18:36 UTC

  45 points

  in reply to: Alexander_Berger’s comment on: A philo­soph­i­cal re­view of Open Philan­thropy’s Cause Pri­ori­ti­sa­tion Framework

  Thanks very much for these comments! Given that Alex—who I’ll refer to in the 3rd person from here—doesn’t want to engage in a written back and forth, I will respond to his main points in writing now and suggest he and I speak at some other time.

  Alex’s main point seems to be that Open Philanthropy (OP) won’t engage in idle philosophising: they’re willing to get stuck into the philosophy, but only if it makes a difference. I understand that—I only care about decision-relevant philosophy too. Of course, sometimes the philosophy does really matter: the split of OP into the ‘longtermism’ and ‘global health and wellbeing’ pots is an indication of this.

  My main reply is that Alex has been too quick to conclude that moral philosophy won’t matter for OP’s decision-making on global health and wellbeing. Let me (re)state a few points which show, I think, that it does matter and, as a consequence, OP should engage further.

  1. As John Halstead has pointed out in another comment, the location of the neutral point could make a big difference and it’s not obvious where it is. If this was a settled question, I might agree with Alex’s take, but it’s not settled.

  2. Relatedly, as I say in the post, switching between two different accounts of the badness of death (deprivationism and TRIA) would alter the value of life-extending to life-improving interventions by a factor of perhaps 5 or more.

  3. Alex seems to object to hedonism, but I’m not advocating for hedonism (at least, not here). My main point is about adopting a ‘subjective wellbeing (SWB) worldview’, where you use the survey research on how people actually experience their lives to determine what does the most good. I’m not sure exactly what OP’s worldview is—that’s basically the point of the main post—but it seems to place little weight on people’s feelings (their ‘experienced utility’) and far more on what they do or would choose (their ‘decision utility’). But, as I argue above, these two can substantially come apart: we don’t always choose what makes us happiest. Indeed, we make predictable mistakes (see our report on affective forecasting for more on this).

  4. Mental health is a problem that looks pretty serious on the SWB worldview but appears nowhere in the worldview that OP seems to favour. As noted, HLI finds therapy for depressed people in LICs is about 10x more cost-effective than cash-transfers in LICs. That, to me, is sufficient to take the SWB worldview seriously. I don’t see what this necessarily has to do with animals.

  5. Will the SWB lens reveal different priorities in other cases? Very probably—pain and loneliness look more important, economic growth less, etc. - but I can’t say for sure because attempts to apply this lens are so new. I had hoped OP’s response would be “oh, this seems to really matter, let’s investigate further” but it seems to be “we’re not totally convinced, so we’ll basically ignore this”.

  6. Alex says “we don’t think that different measures of subjective wellbeing (hedonic and evaluative) neatly track different theories of welfare” but he doesn’t explain or defend that claim. (There are a few other places above where he states, but doesn’t argue for, his opinion, which makes it harder to have a constructive disagreement.)

  7. On the total view, saving lives, and fertility, we seem to be disagreeing about one thing but agreeing about another. I said the total view would lead us to reduce the value of saving lives. Alex says it might actually cause us to increase the value of saving lives when we consider longer-run effects. Okay. In which case, it would seem we agree that taking a stand on population ethics might really matter. In which case, I take it we ought to see where the argument goes (rather than ignore it in case it takes us somewhere we don’t like).

  8. It seems that Alex’s conclusion that moral philosophy barely matters relies heavily on the reasoning in the spreadsheet linked to in footnote 50 of the technical update blog post. The footnote states “Our [OP’s] analysis tends to find that picking the wrong moral weight only means sacrificing 2-5% of the good we could do”. I discussed this above in footnote 3, but I expect it’s worth restating and elaborating on that here. The spreadsheet isn’t explained and it’s unclear what the justification is. I assume the “2-5%” thing is really a motte-and-bailey. To explain, one might think OP is making a very strong claim such as “whatever assumptions you make about morality makes almost no difference to what you ought to do”. Clearly, that claim is implausible. If OP does believe this, that would be an amazing conclusion about practical ethics and I would encourage them to explain it in full. However, it seems that OP is probably making a much weaker claim, such as “given some restrictions on what our moral views can be, we find it makes little difference which ones we pick”. This claim is plausible, but of course, the concern is that the choice of moral views has been unduly restricted. What the preceding bullet points demonstrate is that different moral assumptions (and/​or ‘worldviews’) could substantially change our conclusions—it’s not just a 2-5% difference.

  I understand, of course, that investigating—and, possibly, implementing—additional worldviews is, well, hassle. But Open Philanthropy is a multi-billion dollar foundation that’s publicly committed to worldview diversification and it looks like it would make a practical difference.

• MichaelPlant 16 Jul 2022 14:24 UTC

  3 points

  in reply to: Harry R. Lloyd’s comment on: Wheel­ing and deal­ing: An in­ter­nal bar­gain­ing ap­proach to moral uncertainty

  Ah, this is very interesting! My only comment on this, which develops the idea, is that the agents would realise that the other agents are also certain, so won’t change their minds whatever the oracle pronounces. If we conceive of there as being undecided sub-agents, then the decided sub-agents could think about the value of getting information that might convince them.

• MichaelPlant 16 Jul 2022 14:20 UTC

  3 points

  in reply to: MichaelStJules’s comment on: Wheel­ing and deal­ing: An in­ter­nal bar­gain­ing ap­proach to moral uncertainty

  Ah, I wasn’t aware of the slides. I actually had a section discussing the distinction between IB and moral parliaments which I cut at the last minute because I wasn’t sure what to say. In the Greaves and Cotton-Barrett paper, they make no reference to moral parliaments. Newberry and Ord (2021) cite Greaves and Cotton-Barrett but don’t explain what they take the distinction between the moral parliament and bargaining theoretic approach to be.

  Thinking about it now, both approaches capture the intuition we should imagine what would happen if we distributed resources to representatives of moral views and then some bargaining occurs. The differences appear to be in the specifics: do we allow agents to bargain over time vs only at an instance? Do we allow grand bargains vs restrict the discussion to a limited set of issues? What are the decision-making procedure and the disagreement point? What resources are awarded to the agents (i.e. votes vs money/​time)?

  One problem with the parliamentary approach, where we have to specify a voting procedure is, is how we could possibly justify one procedure over another. After all, to say it was right on the basis of some first-order moral theory would be question-begging.

  Relatedly, as I note in a comment above, is that agents may well prefer to be able to engage in grand, life-long bargains than to be restricted (as Ord and Newberry suggest) to convening parliaments now and then to deal with particular issues. So we can see why agents, starting from something like a ‘veil of ignorance’, would not agree to it.

  It’s for these reasons I suspect the internal bargaining version is superior. I suspect there would be more to say about this after further reflection.

• MichaelPlant 16 Jul 2022 13:40 UTC

  2 points

  in reply to: AidanGoth’s comment on: Wheel­ing and deal­ing: An in­ter­nal bar­gain­ing ap­proach to moral uncertainty

  On regress: perhaps I’m misunderstanding you, but this seems to me to be a universal problem in that we will always be uncertain about how we should make decisions under moral uncertainty. We might have credences in MFT, MEC and IB, but which of these (if any) should we use to decide what to do under uncertainty about what to do under moral uncertainty (and so on...)?

  I didn’t really explain myself here, but there might be better vs worse regress problems. I haven’t worked out my thoughts enough yet to write something useful.

• MichaelPlant 16 Jul 2022 13:20 UTC

  2 points

  in reply to: AidanGoth’s comment on: Wheel­ing and deal­ing: An in­ter­nal bar­gain­ing ap­proach to moral uncertainty

  I’m a little uncomfortable with the distinction between conflicting priorities and unrelated priorities because unrelated priorities are conflicting once you account for opportunity costs: any dollars spent on theory A’s priority can’t be spent on theory B’s priority (so long as these priorities are different). However, I think you’re pointing at something real here and that cases you describe as “conflicting priorities” will tend to lead to spending resources on compromise options rather than splitting the pot, and that the reverse is true for cases you describe as “unrelated priorities”.

  Agree the distinction would be tightened up. And yes, important bit seems to be whether agents will just ‘do their own thing’ vs consider moral trade (and moral ‘trade wars’)

• MichaelPlant 16 Jul 2022 13:18 UTC

  2 points

  in reply to: AidanGoth’s comment on: Wheel­ing and deal­ing: An in­ter­nal bar­gain­ing ap­proach to moral uncertainty

  I like how you go through how a few different scenarios might play out in Section 2 but while I think intuition can be a useful guide, I think it’s hard to how things would play out without taking a more formal approach.

  I don’t really disagree. However, as I stated, my purpose was to give people ‘a feel’ for the view I doubt they would get from Greaves and Cotton-Barrett’s paper (and I certainly didn’t get when I did). The idea was to sketch a ‘quick-and-dirty’ version of the view to see if it was worth doing with greater precision.

• MichaelPlant 16 Jul 2022 13:17 UTC

  3 points

  in reply to: AidanGoth’s comment on: Wheel­ing and deal­ing: An in­ter­nal bar­gain­ing ap­proach to moral uncertainty

  My biggest worry is with the assumption that resources are distributed among moral theories in proportion to the agent’s credences in the moral theories. It seems to me that this is an outcome that should be derived from a framework for decision-making under moral uncertainty, not something to be assumed at the outset

  I don’t think I understand the thinking here. It seems fairly natural to say “I am 80% confident in theory A, so that gets 80% of my resources, etc.”, and then to think about what would happen after that. It’s not intuitive to say “I am 80% confident in utilitarianism, that gets 80% ‘bargaining power’”. But I accept it’s an open question, if we want to do something internal bargaining, what the best version of that is.

  One concern with IB that you don’t mention is that the NBS depends on a “disagreement point” but it’s not clear what this disagreement point should be. The disagreement point represents the utilities obtained if the bargainers fail to reach an agreement. I think the random dictator disagreement point in Greaves and Cotton-Barratt (2019) seems quite natural for many decision problems, but I think this dependence on a disagreement point counts against bargaining approaches.

  I do mention the challenge of the disagreement point (see footnote 7). Again, I agree that this is the sort of thing that merits further inquiry. I’m not sold on the ‘random dictator point’, which, if I understood correctly, is identical to running a lottery where each theory has a X% chance of getting their top choice (where X% represents your credence in that theory). I note in part of section 2 that bargaining agents will likely think it preferable, by their own lights, to bargain over time rather than resolve things with lotteries. It’s for this reason I’m also inclined to prefer a ‘moral marketplace’ over a ‘moral parliament’: the former is what the sub-agents would themselves prefer.

• MichaelPlant 16 Jul 2022 13:16 UTC

  3 points

  in reply to: AidanGoth’s comment on: Wheel­ing and deal­ing: An in­ter­nal bar­gain­ing ap­proach to moral uncertainty

  Hello Aidan. Thanks for all of these, much food for thought. I’ll reply in individual comments to make this more manageable.