Paul_Christiano

Karma: 3,522

Paul_Christiano 29 Sep 2014 4:43 UTC
3 points
0 ∶ 0
on: Brainstorming thread: ideas for large EA funders
As Carl points out here, if big donors have better opportunities than small donors, the small donors can just purchase lotteries. In fact I think that you can get more than a 50% chance of doubling your money using typical risky investments, though this gets into a more complicated discussion about risk. $200,000 is small enough that you probably have access to the typical EA opportunities, so I suspect you shouldn’t do anything differently than smaller EA donors.

That said, I suspect that smaller EA donors should behave quite differently than they currently do, and thinking about what you might do with more money can serve to make this more clear.

edited to add: I also think it’s plausible that at the $100k level a donor should still be saving / trying to pool their money / being risk-loving / just giving it to existing EA orgs, because the amounts of time required to make a good decision even at the meta level—unless you’ve already found someone you trust who you can just hand the money to.

Paul_Christiano 29 Sep 2014 4:51 UTC
14 points
0 ∶ 0
on: Brainstorming thread: ideas for large EA funders
I suspect that funders in general, but especially altruists, and double-especially effective altruists, should be investing in prizes more on the margin. Briefly, some arguments in favor:
- It is much easier to evaluate interventions after the fact. In fact, that is what most EA donors are already doing, but in a way that forces an intervention to be very similar to something else that has already happened if it wants to be funded.
- The incentives created by prize funding tend to be much better.
- Standard mechanisms can overcome many of the costs of prizes (esp. the inability of prospective grantees / prize-recipients to fund their projects if they don’t win a prize), and the development of such institutions would be a huge public good.
- The evaluation process for prizes tends to substantially reinforce EA values, as compared to the more typical grant-making process.
There are a number of ways in which an economically sound EA prize would differ from usual philanthropic prizes (which are typically engineered with the PR impact in mind, rather than as a sustainable funding source). I hope to write a post about this eventually (both w.r.t. the economic arguments, and particularly promising opportunities at the moment). But I’m always interested in thoughts.
What links here?
- Ideas for new experimental EA projects you could fund! by Robert_Wiblin (2 Dec 2014 2:47 UTC; 9 points)

Paul_Christiano 29 Sep 2014 4:55 UTC
4 points
0 ∶ 0
in reply to: Ben Kuhn’s comment on: Brainstorming thread: ideas for large EA funders
I basically believe the efficient market hypothesis, but the inference/implication here doesn’t seem right to me.

The risk from investing in individual stocks rather than broad indices is pretty minor (and for philanthropic capital I think it is completely negligible), and if you accept the EMH the returns are the same. There are a number of other considerations (tax issues, small amounts of domain expertise or insight, hedging) that might lead one to purchase individual stocks instead. The amounts of money involved don’t have to be too large before it starts to look worthwhile to be thinking about these issues.

Paul_Christiano 29 Sep 2014 14:25 UTC
0 points
0 ∶ 0
in reply to: toron’s comment on: Brainstorming thread: ideas for large EA funders
If the returns were worse, the price would drop.

Paul_Christiano 29 Sep 2014 14:26 UTC
2 points
0 ∶ 0
in reply to: TopherHallquist’s comment on: Brainstorming thread: ideas for large EA funders
It could be worse, mostly by correlating your investment returns with the general success of the tech sector (with which your ordinary income is tightly linked, and moreover which drives much effective altruist philanthropy).

Paul_Christiano 29 Sep 2014 14:46 UTC
2 points
0 ∶ 0
in reply to: toron’s comment on: Brainstorming thread: ideas for large EA funders
What would that look like? Ex post the returns of different stocks are very different. Ex ante, if you think some stock is going to have low returns, sell it short.

Paul_Christiano 1 Oct 2014 22:59 UTC
2 points
0 ∶ 0
on: Career choice: Evaluate opportunities, not just fields
The main impact of early choices in a career may be determining what skills you develop, who you know, what you are respected for, and so forth. The value of these resources depends on how useful they will be down the line, i.e. on what opportunities you will have in 10 years. This seems to be an important consideration in favor of thinking about what area to be in.

Paul_Christiano 9 Oct 2014 5:42 UTC
7 points
0 ∶ 0
on: How a lazy eater went vegan
(I apologize in advance for a downer of a comment, especially for the parts that are poorly argued.)

In general, I think the risk of important but unknown nutrient deficiencies from cutting out a whole class of foods is probably much more important than the animal welfare benefits if you are spending much of your life working on good projects. This is a concern with vegetarianism, but it seems much more severe for veganism.

I haven’t looked into the literature in creatine in any depth (largely because my diet is long on creatine), but the evidence for material cognitive effects looks weighty enough that I would want to be careful before considering returning to a low-creatine vegetarian diet. It seems like this should probably get more weight, unless everyone else knows something I don’t.

Based on rough estimates it seems like being a vegan is not going to be worth it on consequentialist grounds, at least for those who would e.g. be willing to pay an additional factor of 2 premium to eat meat periodically. For such people the question is: if raising an animal generates $X of economic value, can you generate a welfare offset for a small multiple of $X? It would be quite surprising to me if this weren’t possible (e.g. just subsidizing the creation of a small number of more humanely raised substitutes should work, which I would guess is orders of magnitude less effective than realistic approaches). I haven’t thought about this at too much length, largely because I think that the effects on animal welfare today are not too important either on contractarian grounds nor on utilitarian grounds (since they seem to have a minimal long-term impact, and the direct impacts are small compared to long-term considerations). But I do think that given the basic economic logic the burden of argument rests with the vegetarian advocate.

If it’s a bad idea on consequentialist grounds, I’m not convinced it’s a useful signaling exercise. Alternatives like “cut out the worst offenders,” “reduce consumption by 90%,” and “buy humanely produced products” seem much better both on consequentialist grounds and (consequently) also signaling grounds.

I second the recommendations for powersmoothies elsewhere (though I drink mine with milk, which I find more delicious and seems probably better health-wise).

Paul_Christiano 9 Oct 2014 15:25 UTC
1 point
0 ∶ 0
in reply to: Greg_Colbourn’s comment on: How a lazy eater went vegan
I agree that creatine is easy to get, but given that I object to a description of how to be a vegetarian that says “if it’s too much trouble don’t bother.”

Animals and non-animals are quite different nutritionally, and it seems quite likely that there are some things you need to supplement as a vegan, or at least that you need to be quite careful about.

The question of whether you can get by without meat at all then seems quite similar to the question of whether you can get by without any fruits or vegetables if you take all of the relevant nutrients via supplementation, for which the consensus appears to be “maybe, but don’t count on it given our limited understanding of nutrition.” Do you see a relevant disanalogy (other than scope, which may be somewhat larger for veggies than meat)?

Paul_Christiano 16 Oct 2014 5:25 UTC
6 points
0 ∶ 0
in reply to: AGB’s comment on: On Progress and Prosperity
My estimates for [1] and [2] together are less than 0.01% per year (for [1], they are very much lower!) So the quantitative effect of speeding up progress, via these channels, is quite small. You would have to be very pessimistic about the plausible trajectory-changing impacts of available interventions before such a small effect was relevant.

I can believe that most of the problems faced by people today are external, and indeed this is related to why I find this disagreement such a painful one. But why do you think that the long-term constraints are external? I’ve never seen a really plausible quantitative argument for this. Resource limitations are the closest, though (1) I’m quite skeptical as a very long-term factor, but this comes down mostly to views on the likely rate of technological progress over the next centuries, and (2) describing those as “external” rather than “a consequence of progress” is already pushing it.

I think that the “man-made” troubles have already easily surpassed the natural worries, via the risk of nuclear annihilation and anthropogenic climate change. Do you disagree with this? What natural problems do you think might be competitive? The unknown unknown?

Paul_Christiano 16 Oct 2014 15:58 UTC
5 points
0 ∶ 0
in reply to: AGB’s comment on: On Progress and Prosperity
In the worst case, you personally saving money and spending it later (in the worst case, to hasten progress at a time when the annual risk of doom has increased!) seems very likely to beat 0.01%, unless you have pretty confident views about the future.

I think research on improving institutional quality, human cognition, and human decision-making, also quite easily cross this bar, have had successes in the past, and have opportunities for more work on the margin. I’ve written about why I think these things would constitute positive changes. But it’s also worth pointing out that if you think there is a 0.01% / year chance of doom now, then improvements in decision-making can probably be justified just by short-term impacts on probability of handling a catastrophe soon

How long progress goes on for before stopping seems irrelevant to its value, according to the model you described. The value of an extra year of progress is always the per annum doom risk. Suppose that after some number of years T the per annum doom probability drops to 0. Then speeding up progress by 1 year reduces that number to T-1, reducing the cumulative probability of doom by the per annum doom probability. And this conclusion is unchanged if the doom probability continuously decreases rather than dropping to 0 all at once, or if there are many different kinds of doom, or whatever. It seems to be an extremely robust conclusion. Another way of seeing this is that speeding up progress by 1 year is equivalent to just pausing the natural doom-generating processes for a year, so naturally the goodness of a year of progress is equal to the badness of the doom-generating processes.

If you believe a 0.01% / year chance of doom by natural catastrophes, addressing response capabilities to those catastrophes in particular generally seems like it is going to easily dominate faster progress. On your model reducing our probability of doom from natural disasters by 1% over the next year is going to be comparable to increasing the overall rate of progress by 1% over the next year, and given the relative spending on those problems it would be surprising if the latter was cost-effective. I can imagine such a surprising outcome for some issues like encountering aliens (where it’s really not clear what we would do), but not for more plausible problems like asteroids or volcanos (since those act by climate disruptions, and there are general interventions to improve society’s robustness to massive climate disruptions).

Whether we’ve missed the boat on climate change or not, it would undermine your claim that historically the main problems have been nature-made and not man-made (which I find implausible), which you were invoking to justify the prediction that the same will hold true in the future. I’m also willing to believe the risk of extremely severe outcomes is not very large, but you don’t have to be very large to beat 0.01% / year.

One reason I’m happy ignoring aliens is that the timing would have to work out very precisely for aliens to first visit us during any particular 10k year period given the overall timescales involved of at least hundreds of millions of years and probably billions. There are further reasons I discount this scenario, but the timing thing alone seems sufficient (modulo simulation-like hypotheses where the aliens periodically check up on Earth to see if it has advanced life which they should exterminate, or some other weird thing).

Paul_Christiano 18 Oct 2014 16:02 UTC
0 points
0 ∶ 0
in reply to: AlexRichard’s comment on: On Progress and Prosperity
I discussed some quantitative estimates of this here, with a general argument for why it would be small in light of model uncertainty. Overall it seems at least a few orders of magnitude smaller than other issues that favor faster progress.

Paul_Christiano 18 Oct 2014 16:08 UTC
4 points
0 ∶ 0
in reply to: Helen’s comment on: On Progress and Prosperity
I think that most good things people do push on differential progress in one way or another, it’s just a different standard for evaluation. Those do stand out as things that contribute particularly to differential progress.

I would guess that on average progress in the social sciences is a net benefit in terms of differential progress, while progress in the hard sciences is a net cost in terms of differential progress. I think the bulk of improvement in institutions comes from people working within organizations trying to help them run better (with the positive developments then propagated through society), and then economists and the social sciences in a distant second.

Paul_Christiano 22 Oct 2014 20:37 UTC
12 points
0 ∶ 0
in reply to: AlasdairGives’s comment on: Should Giving What We Can change its Pledge?

one of the reasons I joined was infact specifically because GWWC was committed only to the narrow cause and not a club for EA types in general.

I think that cause agnosticism is probably the most important novel ingredient of effective altruism, so seeing this kind of sentiment is disheartening. (I don’t have strong views on the pledge itself.)

Paul_Christiano 22 Oct 2014 20:43 UTC
14 points
0 ∶ 0
in reply to: Vincent_deB’s comment on: Should Giving What We Can change its Pledge?
The issue isn’t one of fitting several goals in one pledge. If you take the current GWWC pledge to literally require supporting interventions in developing countries, then that’s not something that a cause-agnostic donor should be willing to agree to early in their life, even if they currently think that interventions in developing countries are most promising.

Which cause you support should be open to change as you learn and as the available opportunities change.

Paul_Christiano 22 Oct 2014 21:13 UTC
6 points
0 ∶ 0
in reply to: AlasdairGives’s comment on: Should Giving What We Can change its Pledge?
I don’t see how this is consistent with pledging to support the cause indefinitely. (I’m not objecting to GWWC being a poverty-focused community.)

It’s worth noting that your stance towards evidence appears to be unusual amongst modern philanthropists, and in particular the standard of “provable” seems both counterproductive and radical. I hope that this stance doesn’t become a standard part of what makes effective altruism distinctive.

I am glad that the open philanthropy project (formerly givewell labs) exists; given that foothold, I think that overreliance on measurement is a significantly less likely failure mode than it otherwise would be. Low epistemic standards and insufficient skepticism seem like more plausible failure modes, and I think we are on the same page concerning those issues. (I agree that more openness exacerbates these difficulties, though I am skeptical that an exclusive focus on poverty per se is too helpful.)

Paul_Christiano 24 Oct 2014 22:13 UTC
1 point
0 ∶ 0
in reply to: ESRogs’s comment on: Risk aversion and investment (for altruists)
log(X+d) ~ log(X) + d/X, for small d. So maximizing E[d/X] is equivalent to maximizing E[log(d+X)].

Paul_Christiano 25 Oct 2014 6:37 UTC
2 points
0 ∶ 0
on: Should Giving What We Can change its Pledge?
I would prefer see “I recognise that I can use part of my income to do a significant amount of good in the developing world” remain in the pledge. On similar grounds, I do not really like the language “beyond poverty.”

Many commenters have focused on the effects of a change on GWWC’s branding and its ability to attract additional members. These may be important effects. But depending on how seriously you take the pledge and on how narrowly you read it, the direct effect may also be quite important. Namely, the a narrow reading of the current pledge entails a substantial commitment to a particular cause, which does not seem compatible with a cause agnostic outlook and a reasonable degree of epistemic humility—especially not for donors early in life (including much of GWWC’s current audience). Though GWWC need not and probably should not broaden its scope to be a general EA organization, I do think that being in direct tension with effectiveness would be bad.

Relatedly, many future-focused EA’s have signed the pledge on the basis of views such as:

The change is not likely to make a difference to people who think that the best way to help others is to ensure that the future will go well, since the pledge already explicitly includes people who will live in the future, as well as those alive now.

I think this is a plausible view in some cases, but in others it is a stretch. For example, I think it is plausible that our altruistic priorities may be quite different if we consider the very long-term, rather than restricting our attention to the coming centuries. One could debate whether such effects can be classified as “helping people in developing countries,” but (1) I suspect that for most precise definitions of developing countries, they couldn’t, (2) in any case, a casual listener would be very unlikly interpret the phrase this way. I would not want to be in the position of defending this interpretation in a discussion of my own giving, and I think that many people would view it as an intentional misrepresentation.

That said, I would probably not sign a modified version of the pledge either. Though I currently give substantially more than 10% of my income away, I am not confident that donation will remain the most effective way to do good (I understand that in principle investment in profitable enterprises or in my own human capital could be carried out in a way consistent with a sufficiently broad reading of the pledge, but I would be even more uncomfortable with this than with donating to non-development causes). I suspect that (a modest probability) * (10% of my income) is more important than the possible positive effects. Hopefully my view can be informative, but I don’t think it should be given much weight as such.

Paul_Christiano 11 Nov 2014 16:43 UTC
3 points
0 ∶ 0
in reply to: Owen Cotton-Barratt’s comment on: Certificates of impact
A philanthropist (or funding agency) gives certificates value by their efforts to acquire certificates.

I.e. rather than funding a bunch of research that it thinks looks promising, the NSF tries to purchase research output. People may buy certificates (or hire researchers in exchange for a fraction of their certificates) because they expect the NSF to buy them later.

The expectation is that the NSF won’t subsequently resell all of these certificates. Doing so would be an explicit preference reversal (unless the value of the certificates grew faster than the NSF’s rate of return, in which case other donors have decided that the NSF funded good things, and the NSF might decide to sell them).

ETA: this subsumes the proposal of occasional, stochastic valuations. A funder interested in tying the value of certificates to some explicit benchmark X can periodically buy certificates at a value determined by the benchmark X.

Paul_Christiano 11 Nov 2014 16:49 UTC
2 points
0 ∶ 0
in reply to: UriKatz’s comment on: Certificates of impact
If you buy and then sell a certificate, you aren’t funding the charity, the ultimate holder of the certificate is. They will only buy the certificate if they are interested in funding the charity.

You could pretend you are funding the charity, but that wouldn’t be true—the person you sold the certificate to would otherwise have bought it from someone else, perhaps directly from the charity. So your net effect on the charity’s funding is 0. I could just as well give some money to my friend and pretend I was funding an effective charity.

(I’m setting aside the tax treatment for now.)

You would pay an employee with certificates for the same reason a company might pay an emplyee in equity. If there is no secondary market, this can be better for the company for liquidity reasons, and can introduce a component of performance pay. But even if there is a secondary market (e.g. for Google stock), it can still be a financially attractive way for a company to pay a large part of its salary, because it passes some of the risk on to the employee without having to constantly adjust dollar-denominated salaries. (There are also default effects, where paying employees in certificates would likely lead to them holding some certificates.)