Brian Wang

88 karmaJoined Sep 2015



    How does this change affect the eligibility of near-term applicants to LTFF/EAIF (e.g., those who apply in the next 6 months) who have received OpenPhil funds in the past / may receive funds from OpenPhil in the future? Currently my understanding is that these applicants are ineligible for LTFF/EAIF by default – does this change if EA funds and Open Philanthropy are more independent?

    Estimates of the mortality rate vary, but one media source says, "While the single figures of deaths in early January seemed reassuring, the death toll has now climbed to above 3 percent." This would put it roughly on par with the mortality rate of the 1918 flu pandemic.

    It should be noted that the oft-cited case-fatality ratio of 2.5% for the 1918 flu might be inaccurate, and the true CFR could be closer to 10%:

    EDIT: Also see this twitter thread:

    It seems that there are two factors here leading to a loss in altruistic belief:

    1. Your realization that others are more selfish than you thought, leading you to feel a loss of support as you realize that your beliefs are more uncommon than you thought.

    2. Your uncertainty about the logical soundness of altruistic beliefs.

    Regarding the first, realize that you're not alone, that there are thousands of us around the world also engaged in the project of effective altruism – including potentially in your city. I would investigate to see if there are local effective altruism meetups in your area, or a university group if you are already at university. You could even start one if there isn't one already. Getting to know other effective altruists on a personal level is a great way to maintain you desire to help others.

    Regarding the second, what are the actual reasons for people answering "100 strangers" to your question? I suspect that the rationale isn't on strong ground – that it is mostly borne out of a survival instinct cultivated in us by evolution. Of course, for evolutionary reasons, we care more about ourselves than we care about others, because those that cared too much about others at the expense of themselves died out. But evolution is blind to morality; all it cares about is reproductive fitness. But we care about so, so much more. Everything that gives our lives value - the laughter, love, joy, etc. – is not optimized for by evolution, so why trust the answer "100 strangers" if it is just evolution talking?

    I believe that others' lives have an intrinsic value on par with my own life, since others are just as capable of all the experiences that give our lives value. If I experience a moment of joy, vs. if Alice-on-the-other side-of-the-world-whom-I've-never-met experiences a moment of joy, what's the difference from "the point of view of the universe"? A moment of joy is a moment of joy, and it's valuable in and of itself, regardless who experiences it.

    Finally, if I may make a comment on your career plan – I might apply for career coaching from 80,000 hours. Spending 10 years doing something you don't enjoy sounds like a great recipe for burnout. If you truly don't think that you'll be happy getting a machine learning PhD, there might be better options for you that will still allow you to have a huge impact on the world.

    I think the central "drawing balls from an urn" metaphor implies a more deterministic situation than that which we are actually in – that is, it implies that if technological progress continues, if we keep drawing balls from the urn, then at some point we will draw a black ball, and so civilizational devastation is basically inevitable. (Note that Nick Bostrom isn't actually saying this, but it's an easy conclusion to draw from the simplified metaphor). I'm worried that taking this metaphor at face value will turn people towards broadly restricting scientific development more than is necessarily warranted.

    I offer a modification of the metaphor that relates to differential technological development. (In the middle of the paper, Bostrom already proposes a few modifications of the metaphor based on differential technological development, but not the following one). Whenever we draw a ball out of the urn, it affects the color of the other balls remaining in the urn. Importantly, some of the white balls we draw out of the urn (e.g., defensive technologies) lighten the color of any grey/black balls left in the run. A concrete example of this would be the summation of the advances in medicine over the past century, which have lowered the risk of a human-caused global pandemic. Therefore, continuing to draw balls out of the urn doesn't inevitably lead to civilizational disaster – as long as we can be sufficiently discriminate towards those white balls which have a risk-lowering effect.

    Interesting idea. This may be worth trying to develop more fully?

    Yeah. I'll have to think about it more.

    I'm still coming at this from a lens of "actionable advice for people not in ea". It might be that the person doesn't know many other trusted individuals, what should be the advice then?

    Yeah, for people outside EA I think structures could be set up such that reaching consensus (or at least a majority vote) becomes a standard policy or an established norm. E.g., if a journal is considering a manuscript with potential info hazards, then perhaps it should be standard policy for this manuscript to be referred to some sort of special group consisting of journal editors from a number of different journals to deliberate. I don't think people need to be taught the mathematical modeling behind the unilateralist's curse for these kinds of policies to be set up, as I think people have an intuitive notion of "it only takes one person/group with bad judgment to fuck up the world; decisions this important really need to be discussed in a larger group."

    One important distinction is that people who are facing info hazards will be in very different situations when they are within EA vs. when they are out of EA. For people within EA, I think it is much more likely to be the case that a random individual has an idea that they'd like to share in a blog post or something, which may have info hazard-y content. In these situations the advice "talk to a few trusted individuals first" seems to be appropriate.

    For people outside of EA, I think those who are in possession of info hazard-y content are much more likely to be embedded in some sort of larger institution (e.g., a research scientist or a journal editor looking to publish something), where perhaps the best leverage is setting up certain policies, rather than trying to teach everyone the unilateralist's curse.

    As I understand it you shouldn't wait for consensus else you have the unilateralist's curse in reverse. Someone pessimistic about an intervention can block the deployment of an intervention needed to avoid disaster.

    You're right, strict consensus is the wrong prescription. A vote is probably better. I wonder if there's mathematical modeling that you could do that would determine what fraction of votes is optimal, in order to minimize the harms of the standard unilateralist's curse and the curse in reverse? Is it a majority vote? A 2/3s vote? l suspect this will depend on what the "true sign" of releasing the potentially dangerous info is likely to be; the more likely it is to be negative, the higher bar you should be expected to clear before releasing.

    If there is a single person with the knowledge of how to create safe efficient nuclear fusion they cannot expect other people to release it on their behalf.

    Ah right. I suppose the unilateralist's curse is only a problem insofar as there are a number of other actors also capable of releasing the information; if you are a single actor then the curse doesn't really apply. Although one wrinkle might be considering the unilateralist's curse with regards to different actors through time (i.e., erring on the side of caution with the expectation that other actors in the future will gain access to and might release the information), but coordination in this case might be more challenging.

    What the researcher can do is try and build consensus/lobby for a collective decision making body on the internal climate heating (ICH) problem. Planning to release the information when they are satisfied that there is going to be a solution in time for fixing the problem when it occurs.

    Thanks, this concrete example definitely helps.

    I think I am also objecting to the expected payoff being thought of as a fixed quantity. You can either learn more about the world to alter your knowledge of the payoff or try and introduce things/insituttions into the world to alter the expected payoff. Building useful institutions may rely on releasing some knowledge, that is where things become more hairy.

    This makes sense. "Release because the expected benefit is above the expected risk" or "not release because the vice versa is true" is a bit of a false dichotomy, and you're right that we should be more thinking about options that could maximize the benefit while minimizing the risk when faced with info hazards.

    Also as the the unilaterlist's curse suggests discussing with other people such that they can undertake the information release, sometimes increases the expectation of a bad out come. How should consensus be reached in those situations?

    This can certainly be a problem, and is a reason not to go too public when discussing it. Probably it's best to discuss privately with a number of other trusted individuals first, who also understand the unilateralist's curse, and ideally who don't have the means/authority of releasing the information themselves (e.g., if you have a written up blog post you're thinking of posting that might contain info hazards, then maybe you could discuss in vague terms with other individuals first, without sharing the entire post with them?).

    The unilateralists curse only applies if you expect other people to have the same information as you right?

    My understanding is that it applies regardless of whether or not you expect others to have the same information. All it requires is a number of actors making independent decisions, with randomly distributed error, with a unilaterally made decision having potentially negative consequences for all.

    You can figure out if they have the same information as you to see if they are concerned about the same things you are. By looking at the mitigation's people are attempting. Altruists should be attempting mitigations in a unilateralist's curse position, because they should expect someone less cautious than them to unleash the information. Or they want to unleash the information themselves and are mitigating the downsides until they think it is safe.

    I agree that having dangerous information released by those who are in a position to mitigate the risks is better than having a careless actor releasing that same information –– but I disagree that this is sufficient reason to preemptively release dangerous information. I think a world where everyone follows the logic of "other people are going to release this information anyway but less carefully, so I might as well release it first" is suboptimal compared to a world where everyone follows a norm of reaching consensus before releasing potentially dangerous information. And there are reasons to believe that this latter world isn't a pipe dream; after all, generally when we're thinking about info hazards, those who have access to the potentially dangerous information generally aren't malicious actors, but rather a finite number of, e.g., biology researchers (for biorisks) who could be receptive to establishing norms of consensus.

    I'm also not sure how the strategy of "preemptively release, but mitigate" would work in practice. Does this mean release potentially dangerous information, but with the most dangerous parts redacted? Release with lots of safety caveats inserted? How does this preclude the further release of the unmitigated info?

    I've not had the best luck reaching out to talk to people about my ideas. I expect that the majority of new ideas will come from people not heavily inside the group and thus less influenced by group think. So you might want to think of solutions that take that into consideration.

    I'm not sure I'm fully understanding you here. If you're saying that the majority of potentially dangerous ideas will originate in those who don't know what the unilateralist's curse is, then I agree –– but I think this is just all the more reason to try to spread norms of consensus.

    The relevance of unilateralist's curse dynamics to info hazards is important and worth mentioning here. Even if you independently do a thorough analysis and decide that the info-benefits outweigh the info-hazards of publishing a particular piece of information, that shouldn't be considered sufficient to justify publication. At the very least, you should privately discuss with several others and see if you can reach a consensus.

    I wonder how much the "spend 1 year choosing and 4 years relentless pursuing a project" rule of thumb applies to having a high-impact career. Certain career paths might rely on building a lot of career capital before you can have high-impact, and career capital may not be easily transferable between domains. For example, if you first decide to relentlessly pursue a career in advancing clean meat technology for four years, and then re-evaluate and decide that influencing policymakers with regards to AI safety is the highest-value thing for you to do, it's probably going to be difficult to pivot. There's a sense in which you might be "locked in" to a career after you spend enough time in it. My sense is that, for career-building in the face of uncertainty, it might be best to prioritize keeping options open (e.g., by building transferable career capital) and/or spending more time on the choosing phase.

    Yes, I accept that result, and I think most EAs would (side note: I think most people in society at large would, too; if this is true, then your post is not so much an objection to the concept of EA as it is to common-sense morality as well). It's interesting that you and I have such intuitions about such a case – I see that as in the category of "being so obvious to me that I wouldn't even have to hesitate to choose." But obviously you have different intuitions here.

    Part of what I'm confused about is what the positive case is for giving everyone an equal chance. I know what the positive case is for the approach of automatically saving two people vs. one: maximizing aggregate utility, which I see as the most rational, impartial way of doing good. But what's the case for giving everyone an equal chance? What's gained from that? Why prioritize "chances"? I mean, giving Bob a chance when most EAs would probably automatically save Amy and Susie might make Bob feel better in that particular situation, but that seems like a trivial point, and I'm guessing is not the main driver behind your reasoning.

    One way of viewing "giving everyone an equal chance" is to give equal priority to different possible worlds. I'll use the original "Bob vs. a million people" example to illustrate. In this example, there's two possible worlds that the donor could create: in one possible world Bob is saved (world A), and in the other possible world a million people are saved (world B). World B is, of course, the world that an EA would create every time. As for world A, well: can we view this possible world as anything but a tragedy? If you flipped a coin and got this outcome, would you not feel that the world is worse off for it? Would you not instantly regret your decision to flip the coin? Or even forget flipping the coin, we can take donor choice out of it; wouldn't you feel that a world where a hurricane ravaged and destroyed an urban community where a million people lived is worse than a world where that same hurricane petered out unexpectedly and only destroyed the home of one unlucky person?

    If so, then why give tragic world A any priority at all, when we can just create world B instead? I mean, if you were asked to choose between getting a delicious chocolate milkshake vs. a bee sting, you wouldn't say "I'll take a 50% chance of each, please!" You would just choose the better option. Giving any chance, no matter how small, to the bee sting would be too high. Similarly, giving any priority to tragic world A, even 1 in 10 million, but be too high.

    Load more