[LINK] Nick Szabo: Beware Pascal’s Scams
Nick Szabo on acting on extremely long odds with claimed high payoffs:
Beware of what I call Pascal’s scams: movements or belief systems that ask you to hope for or worry about very improbable outcomes that could have very large positive or negative consequences. (The name comes of course from the infinite-reward Wager proposed by Pascal: these days the large-but-finite versions are far more pernicious). Naive expected value reasoning implies that they are worth the effort: if the odds are 1 in 1,000 that I could win $1 billion, and I am risk and time neutral, then I should expend up to nearly $1 million dollars worth of effort to gain this boon. The problems with these beliefs tend to be at least threefold, all stemming from the general uncertainty, i.e. the poor information or lack of information, from which we abstracted the low probability estimate in the first place: because in the messy real world the low probability estimate is almost always due to low or poor evidence rather than being a lottery with well-defined odds.
Nick clarifies in the comments that he is indeed talking about singularitarians, including his GMU colleague Robin Hanson. This post appears to revisit a comment on an earlier post:
In other words, just because one comes up with quasi-plausible catastrophic scenarios does not put the burden of proof on the skeptics to debunk them or else cough up substantial funds to supposedly combat these alleged threats.
This is a terrible misrepresentation. SI does not argue for donations on these grounds; Eliezer and other SI staff have explicitly rejected such Pascalian reasons, but instead argued that the risks that they wish to avert are quite probable.
Then it constitutes a serious PR problem.
Or is at least a symptom of bad PR.
It has to also be probable that their work averts those risks, which seem incredibly improbable by any reasonable estimate. If the alternative Earth was to adopt a strategy of ignoring prophetic groups of ‘idea guys’ similar to SI and ignore their pleads for donations so that they can hire competent researchers to pursue their ideas, I do not think that such decision would have increased the risk by more than a miniscule amount.
People currently understand the physical world sufficiently to see that supernatural claims are bogus, and so there is certainty about impossibility of developments predicated on supernatural. People know robust and general laws of physics that imply impossibility of perpetual motion, and so we can conclude in advance with great certainty that any perpetual motion engineering project is going to fail. Some long-standing problems in mathematics were attacked unsuccessfully for a long time, and so we know that making further progress on them is hard. In all these cases, there are specific pieces of positive knowledge that enable the inference of impossibility or futility of certain endeavors.
In contrast, a lot of questions concerning Friendly AI remain confusing and unexplored. It might turn out to be impossibly difficult to make progress on them, or else a simple matter of figuring out how to apply standard tools of mainstream mathematics. We don’t know, but neither do we have positive knowledge that implies impossibility or extreme difficulty of progress on these questions. In particular, the enormity of consequences does not imply extreme improbability of influencing those consequences. It looks plausible that the problem can be solved.
This kind of seems like political slander to me. Maybe I’m miscalibrated? But it seems like you’re thinking of “reasonable estimates” as things produced by groups or factions, treating SI as a single “estimate” in this sense, and lumping them with a vaguely negative but non-specified reference class of “prophetic groups”.
The packaged claims function to reduce SI’s organizational credibility, and yet it references no external evidence and makes no testable claims. For your “prophetic groups” reference class, does it include 1930′s nuclear activists, 1950′s environmentalists, or 1970′s nanotechnology activists? Those examples come from the socio-political reference class I generally think of SI as belonging to, and I think of them in a mostly positive way.
Personally, I prefer to think of “estimates” as specific predictions produced by specific processes at specific times, and they seem like they should be classified as “reasonable” or not on the basis of their mechanisms and grounding in observables in the past and the future.
The politics and social dynamics surrounding an issue can give you hints about what’s worth thinking about, but ultimately you have to deal with the object level issues, and the object level issues will screen off the politics and social dynamics once you process them. The most reasonable tool for extracting a “coherent opinion” from someone on the subject of AGI that is available to the public that I’m aware of is the uncertain future.
(Endgame: Singularity is a more interesting tool in some respects. It’s interesting for building intuitions about certain kinds of reality/observable correlations because it has you play as a weak but essentially benevolent AGI rather than as humanity, but (1) it is ridiculously over-specific as a prediction tool, and (2) seems to give the AGI certain unrealistic advantages and disadvantages for the sake of making it more fun as a game. I’ve had a vague thought to fork it, try to change it to be more realistic, write a bot for playing it, and use that as an engine for Monte-carlo simulator of singularity scenarios. Alas: a day job prevents me from having the time, and if that constraint were removed I bet I could find many higher value things to work on, reality being what it is, and people being motivated to action the way they are.)
Do you know of anything more epistemically helpful than the uncertain future? If so, can you tell me about it? If not, could you work through it and say how it affected your model of the world?
(Note that the Uncertain Future software is mostly supposed to be a conceptual demonstration; as mentioned in the accompanying conference paper, a better probabilistic forecasting guide would take historical observations and uncertainty about constant underlying factors into account more directly, with Bayesian model structure. The most important part of this would be stochastic differential equation model components that could account for both parameter and state uncertainty in nonlinear models of future economic development from past observations, especially of technology performance curves and learning curves. Robin Hanson’s analysis of the random properties of technological growth modes has something of a similar spirit.)
I think your estimate of their chances of success is low. But even given that estimate, I don’t think it’s Pascalian. To me, it’s Pascalian when you say “my model says the chances of this are zero, but I have to give it non-zero odds because there may be an unknown failing in my model”. I think Heaven and Hell are actually impossible, I’m just not 100% confident of that. By contrast, it would be a bit odd if your model of the world said “there is this risk to us all, but the odds of a group of people causing a change that averts that risk are actually zero”.
It is not just their chances of success. For the donations to matter, you need SI to succeed where without SI there is failure. You need to get a basket of eggs, and have all the good looking eggs be rotten inside but one fairly rotten looking egg be fresh. Even if a rotten looking egg is nonetheless more likely to be fresh inside than one would believe, it is highly unlikely situation.
I’m afraid I’m not getting your meaning. Could you fill out what corresponds to what in the analogy? What are all the other eggs? In what way do they look good compared to SI?
All the other people and organizations that are no less capable of identifying the preventable risks (if those exist) and addressing them, have to be unable to prevent destruction of mankind without SI. Just like in the Pascal’s original wager, the Thor and other deities are to be ignored by omission.
On how the SI does not look good, well, it does not look good to Holden Karnofsky, or me for that matter. Resistance to feedback loops is an extremely strong point of his.
On the rationality movement, here’s a quote from Holden.
Could you give me some examples of other people and organizations trying to prevent the risk of an Unfriendy AI? Because for me, it’s not like I believe that SI has a great chance to develop the theory and prevent the danger, but rather like they are the only people who even care about this specific risk (which I believe to be real).
As soon as the message becomes widely known, and smart people and organizations will start rationally discussing the dangers of Unfriendly AI, and how to make a Friendly AI (avoiding some obvious errors, such as “a smart AI simply must develop a human-compatible morality, because it would be too horrible to think otherwise”), then there is a pretty good chance that some of those organization will be more capable than SI to reach that goal: more smart people, better funding, etc. But at this moment, SI seems to be the only one paying attention to this topic.
It’s a crooked game, but it’s the only game in town?
None of that is evidence that SI would be more effective if it had more money. Assign odds to hostile AI becoming extant given low funding for SI, and compare the odds of hostile AI becoming extant given high funding for SI. The difference between those two is proportional to the value of SI (with regards to preventing hostile AI).
SI being the only one ought to lower your probability that this whole enterprise is worthwhile in any way.
With regards to the ‘message’, i think you grossly over estimate value of a rather easy insight that anyone who has watched Terminator could have. With regards to “rationally discussing”, what I have seen so far here is pure rationalization and very little, if any, rationality. What the SI has on the track record is, once again, a lot of rationalizations and not enough rationality to even have had an accountant through it’s first 10 years and first over 2 millions dollars in other people’s money.
Note that that second paragraph is one of Holden Karnofsky’s objections to SIAI: a high opinion of its own rationality that is not so far substantiable from the outside view.
Yes. I am sure Holden is being very polite, which is generally good but I’ve been getting impression that the point he was making did not in full carry across the same barrier that has resulted in the above-mentioned high opinion of own rationality despite complete lack of results for which rationality would be better explanation than irrationality (and presence of results which set rather low ceiling for the rationality). The ‘resistance to feedback’ is even stronger point, suggestive that the belief in own rationality is, at least to some extent, combined with expectation that it won’t pass the test and subsequent avoidance (rather than seeking) of tests; as when psychics do believe in their powers but do avoid any reliable test.
Really, last time I checked Eliezer was refusing to name either a probability or a time scale.
I’m not seeing how you get from “doesn’t state an explicit probability or timescale publically” to “argues that SI should be supported on Pascalian grounds”.
It looked like just a response to you saying “instead argued that the risks that they wish to avert are quite probable.”
In Pascal’s original wager the ‘risk’ is 50% . Due to our natural tendency to see two alternatives, once presented in A vs B form, as closer to even odds when information is absent. That’s what wager is about—screwing up probabilities in absence of knowledge of how to calculate them, and arguing that it is quite probable when it’s not quite probable. Not so much the problem with decision process as the problem with ‘lets call vague feelings probabilities’. Between other gods, the possibility of better deals in the future, I have impression that mathematically it would work out to ‘do not pay’, if someone could actually do the math. With the made up probabilities that tend towards 50⁄50 when there’s unknowns, and when the propositions are maliciously generated to relieve you of your wallet, and when you are simply inserting a hypothesis into graph because you were told something (zero to nonzero update), a little wonder that some people get exploited.
You are grossly misinformed.
Er. We’re talking about Pascal’s wager, right? The one published in Pensees? The one which explicitly invokes infinities, where it doesn’t matter if the odds are 1 to 1 or 1,000,000,000 to 1, the argument still goes through?
The point is that it is still a Pascal’s wager even if you have mis-estimated probabilities and argued that it is actually likely that God exists.
In case of SI, even if we assume that risk exists it is still the case that one is to donate to group of people whom, in all likelihood, are entirely incapable of affecting the risk in any way what so ever (and are only offering risk reduction due to their incompetence combined with Dunning-Kruger effect. It never happened in the history that the first people to take money for cure would be anything but either self deluded or confidence tricksters)
You realize DK is a narrow effect which only obtains in certain conditions, is still controversial, and invoking it just makes you look like you’ll grab at any thing at all no matter how dubious in order to attack SI, right? (About on the same level as ‘Hitler was an atheist!’)
Seriously. In no area of research, medicine, engineering, or whatever, the first group to tackle a problem succeeded? Such a world would be far poorer and still stuck in the Dark Ages than the one we actually live in. I realize this may be a hard concept, but sometimes, the first person to tackle a problem—succeeds! In fact, sometimes multiple people tackling the problem all simultaneously succeed! (This is very common; called multiple discovery.)
Not every problem is as hard as fusion, or to put it another way, most hard problems are made of other, easier, problems, while if your hyperbolic statement were true, no progress would ever be made.
Dunning Kruger effect is likely a product of some general deficiency in the meta reasoning facility leading both to failure of reasoning itself and failure of evaluation of the reasoning; extremely relevant to people that proclaim themselves to be more rational, more moral, and so on than anyone else but do not seem to accomplish above mediocre performance at fairly trivial yet quantifiable things.
Ghmm. He said first people to take money, not first people to tackle.
The first people to explain the universe (and take some contributions for that) produced something of negative value, nearly all of the medicine until last couple hundred years was not only ineffective but completely harmful, and so on.
If you look at very narrow definitions, of course, the first to tackle nuclear bomb creation did succeed—but the first to tackle the general problem of weapon of mass destruction were various shamans sending a curse. If saving people from AI is an easy problem, then we’ll survive without SI; if it’s a hard problem, at any rate SI doesn’t start with a letter from Einstein to the government, it starts with a person with no quantifiable accomplishments cleverly employing oneself. As far as I am concerned, there’s literally no case for donations here; the donations happen via sort of decision noise similar to how NASA has spent millions on various antigravity devices, the power companies have spent millions on putting electrons in hydrogen orbitals at below ground level (see Mills hydrinos), and millions were invested in Steorn’s magnetic engine.
Speaking of yourself in the third person?
Dmytry, you are abusing sockpuppet accounts. Your use of private_messaging was questionable, but at least you declared it as an alias early on. Right now you are using a sockpuppet with the intent to deceive.
“Ghmm” is a hapax legomenon used solely by private_messaging/Dmytry and JaneQ.
Ah, nice. I’ve been looking for a phrase/word with this meaning.
“Ghmm” was the phrase that made me certain. Before that I found “edit: i.e.” and “etc etc etc”, but these were not quite as unique. (PB page)
You know, I’ve been wondering about that for a while now, but it never occurred to me to look for hapaxes (nor had I been aware of the phrase). I have just learned a new technique, for which I have learned a cool new word, which helps solve a problem I actually had. If I endorsed upvoting multiple times, I would upvote you multiple times; as it is, you’ll have to settle for an upvote and my gratitude.
It’s just one of the little-known advantages to reading critical analysis of classical or Biblical literature! Although I’d be hard-pressed to name a second advantage.
The embarrassing thing, now that you mention it, is that I’m acquainted with this technique in that context (identifying source texts and common authors and so forth) but it still didn’t occur to me to apply it here, despite it being the same problem even at a surface level.
(sigh) Corrupted hardware sucks. It ain’t the things I don’t know that irritate me. It’s not even the things I do know that just ain’t so. It’s the things I know, that are so, and that somehow don’t present themselves to be reasoned with when I need them.
That seems unlikely. Leading both?
Mediocrity is sufficient to push them entirely out of the DK gap; your thinking DK applies is just another example of what I mean by these being fragile easily over-interpreted results.
(Besides blatant misapplication, please keep in mind that even if DK had been verified by meta-analysis of dozens of laboratory studies, which it has not, that still only gives a roughly 75% chance that the effect applies outside the lab.)
Without specifics, one cannot argue against that.
So you’re just engaged in reference class tennis. (‘No, you’re wrong because the right reference class is magicians!’)
Seems straightforward to me: Eliezer’s unwarranted self importance did result in him not pursuing education or for that matter proper self education, and simultaneously to believing he’s awesome and selling existential risk reduction that nobody else would sell. edit: The alternative explanation is the level of resistance to self deception so high that the process of the self education transcended the necessity to seek objective feedback on the progress (which one gets if one e.g. tries to prove mathematical theorems, as here an un-intelligent process of checking a proof can validate one’s powers of reasoning).
Did it ever occur to you that one has to actually do something incompatible with the broad reference class to get into much much smaller reference class? E.g. you are in reference class ‘people’, not reference class ‘people with IQ>=150’ unless you take IQ test or take other test with very low false positive rate. Likewise, the reference class is ‘people with grand promises’ until you actually do something that moves you into microscopic sub class of ‘people with grand promises who deliver’.
Suppose one were to grant that for Eliezer. Out of curiosity, I would be interested in hearing how Nick Bostrom & FHI are similarly deluded and in the reference class of magicians.
Speculation of sufficiently advanced future technologies is indistinguishable from magical thinking.
Unless there is scientific method in what you’re doing, and unless you’re producing something testable and testing it, you are certainly not in the reference class of scientists. Unless there is plenty of rigour, you are not in reference class of people using mathematical methods (even if you have formulas in your papers). Not up to grabs here either. Maybe the reference class is philosophers. If you wish, philosophers with PhD.
As they do tend to honestly make actual arguments rather than just try to manipulate for profit the people who already agree with core ideas, one can examine actual argumentation. Which is not very good. E.g. simulation argument of his looks watertight at first glance but is really reliant on assumption about physics (suppose MWI is correct, now the counting patently doesn’t work for probabilities), and does not account for potentially enormous number of simulated beings that can easily tell their reality is simulated as the simulator cuts some corners. Typical example of philosophy, making arguments that seem true for mere lack of alternative propositions to made up assertions. You only spend time formalizing such stuff if you can’t see that you are building false precision, making up far too many assumptions for the results to be meaningful in any way (in the field where intuition is unlikely to work, too). If you don’t see that you are making a lot of extremely shaky assumptions when you are making a lot of extremely shaky assumptions, you’ll be susceptible to generating false precision precisely as per Nick Szabo’s article.
Their musings about singularity are in precise agreement with the hypothesis that at the current point in time it is early enough that the only people ‘working’ on this are those who for some reason fail to see when they aren’t making progress. I’m pretty sure all this movement will look very silly in 100 years—there will be dangers they did not see and there won’t be dangers they focused on.
What is the correct counting for MWI, exactly?
I think you are misunderstanding the SA, which is surprising since it’s formally pretty simple.
The SA is a trilemma; finding evidence that strongly supports one leg of the trilemma is not a problem with the trilemma itself. It’s just reasons for you to bite a particular bullet: “the SA says ‘either X or Y or Z’, and here our reality looks like a cutrate simulation, so I guess Z was right after all!′
So the trilemma remains ‘watertight’ even if the specific paper in enumerating reasons to believe X (or Y, or Z) fails to cover some favorite bit of reasoning of yours. The reasoning still fits inside the trilemma framework and is not an argument against the framework itself.
(My own impression is that your cutrate suggestion wouldn’t go very far since it’s not clear—to me, anyway—what a cutrate simulation would look like, and whether our own universe is not cutrate. One could validly argue that our failure to find good clear evidence that we’re in a simulation is evidence against being in a simulation, but quantifying how much evidence this would be is even harder. And given how many crude simulation we run for science & business & pleasure, and the Fermi paradox, it seems especially unlikely that this point is strong enough to move someone from biting the we’re-in-a-simulation bullet to biting one of the other bullets.)
Everyone looks silly from 100 years on. That’s not a useful point to make.
MWI: we don’t know what is that works, but we can tell if something doesn’t work. Probabilities don’t seem to work out if you just count distinct observers. Plus, the number of distinct observers grows very rapidly with time, so you get extreme case of doomsday paradox. If you aren’t just counting distinct observers but count copies twice then your probabilities could as well depend to e.g. thickness of wires in the computer, not just the raw number of simulated realities.
Furthermore, more significantly, under MWI it is not even clear what first two statements could even mean.
We are discussing Nick Bostrom, and I take the http://en.wikipedia.org/wiki/Nick_Bostrom to be at least somewhat representative of his contribution to simulation argument
The trilemma as stated is:
No civilization will reach a level of technological maturity capable of producing simulated realities.
No civilization reaching aforementioned technological status will produce a significant number of simulated realities, for any of a number of reasons, such as diversion of computational processing power for other tasks, ethical considerations of holding entities captive in simulated realities, etc.
Any entities with our general set of experiences are almost certainly living in a simulation.
I assumed that the last statement is to be taken as ‘we should expect to be in a sim if first two conditions are false and given our general set of experiences’, by assumption of at least rudimentary relevance of this trilemma.
In such case there is the fourth possibility with probability overwhelmingly higher than of this entire argument: the wild guess that there would ever be a good reason to believe that we should be among most numerous, with same weight for real thing and simulator (or same weight for different types of simulator), is not spot on. Collaborated also by us not being among those in weird sims of any kind (we’d detect a god speaking to us every day).
Furthermore, the distinction between perfect simulator and reality strikes me as nonsensical. Until there is a measurement that we are in a simulation, we may most sensibly assume we are in both (this time merely drawing inspiration from the sort of intuitions we might have had if we believed in MWI). The probability of measurement that we are in a simulation, well that has an exceptionally good chance of being a much more complicated matter than assumed.
edit: To clarify, my point is that even putting this sort of stuff in words or equations is a great example of false precision that Nick Szabo complains about. Too many assumptions have to be made without noticing that those assumptions are made, for the statements to have meaning at all.
Those who aren’t grossly wrong (Newton for example) don’t look as silly as the silly I am speaking of.
Then we can pardon Bostrom for not taking them into account.
Wikipedia is pretty bad on philosophy (the SEP is much better), and in this case, there’s no reason not to read Bostrom’s original paper and the correction: he writes clearly, and they are readily available on his website.
What? Could you write this more clearly, I have no idea what you’re trying to say.
Newton is actually a great example; if you don’t choose to ignore the areas which make him look bad, he looks like an incredible fool in many respects. His constant pursuit of alchemy even then was the object of derision, and while we no longer would hang or exile him for his bizarre theology & eschatology, we (even most Christians) would regard them as hilarious. Then there are his priority disputes...
If even Newton looks this foolish, what hope can the rest of us have? No, the suggestion ‘would this make me look foolish in 100 years?’ does us no good in practice.
When pondering the possibilities where we live given lack of grand unified theory ‘of everything’, you can’t assume your physical intuitions hold true. In fact you should assume the opposite. The MWI is just an example of how philosophical argument that looks entirely invariable admits defeat from possible physics, in a way in which the mathematics—which philosophy mimics—does not. That means validity of argument requires validity of intuitions, which are reasonably very unlikely to be valid in any grand sense. There’s also historical example: a lot of philosophers assuming Euclidean geometry is the only logically possible kind of geometry, without even noticing that they are making such assumption, up until mathematicians came up with alternative.
You assume that the probability of being among either group depends to number of yous within that group (rather than something entirely different), to do anthropic reasoning beyond tautological ‘we can’t observe universes where we can’t be alive’. In my opinion this is a case of wild guess over unknowns, totally false precision.
I came up with a clearer example of how something totally different may actually make a lot more sense: Solomonoff induction on codes that model various universes. A model of the universe outputting the data matching your internal self-perception must not only contain yourself, but must include the code that finds yourself within that model, so that output begins with sense data. It is then clear that the code that picks one of yous out of a model full of all sorts of simulations may easily be larger than the code that picks you out of the world where there is just one you, but the number of various others is much smaller.
I’m not arguing that Bostrom is bad for a philosopher. I am outlining how in philosophy you just make all sorts of assumptions that you don’t notice are wild guesses, and how the field is basically built on false precision. I.e. you assume that the probability of being within some huge set full of yous and non-yours is independent of number of non-yous, which is just a wild guess. Connect together half a dozen implicit wild guesses with likelihood of correctness of overly generous 1 in 100 each , and we’re speaking of probability of correctness in the range of 10^-12 . Philosophy is generally like this.
I believe this falls under Nick Szabo’s complaint about false precision.
Also, a paper by Steven Weinberg on usefulness of philosophy.
It seems to me that funding philosophical works in this field may actually be actively harmful, due to establishment of such false precision and prejudices. It’s like funding ‘embracing bias, imprecision, and making your mind up before checking where the mathematics will lead you’.
That’s the whole point: very low probability of being right. There’s a crucial difference: the methods Newton employed managed to achieve a non-zero (and not negligible) truth finding rate. So he made something that does not look silly. Even with this, most of stuff was quite seriously wrong.
Do you think it’s “generous” to assign only 99% probability in the negation of “the probability of being within some huge set full of yous and non-yours is independent of number of non-yous” where “you” is interpreted to include all your observations? That seems like insane overconfidence in a view that goes haywire in simple finite discrete cases.
Typical philosophy: tear strawman alternatives to prove a wild guess. (Why strawman: because can be also dependent on anything else entirely like positions of copies)
Also, the 99% confidence is not in the dependence on non yous (and yous BUT nothing else), the 99% confidence is in the wild guess of independence from everything else and dependence to count of yous, to be wrong. Also, consider two computer circuits real nearby, running identical you, separated by thin layer of dielectric. Remove dielectric, 1 copy with thicker wires. Conclusion: it may depend to thickness of wires of a copy or maybe to the speed of the copy.
Hell, the probability of being in a specific copy may just as well be undefined entirely until a copy figures out which copy it is, and then depend solely to how it was figured out.
Let’s suppose that the probability of being sampled out of model is sum of 2^-l over all codes that pluck you out of the model, like in Solomonoff induction. May well be dependent on the presence or absence of stone dummies (provided those break some simple method of locating you). Will definitely depend to your position. Go show it broken.
edit : actually, this alternative distribution for observers (and observer-moments) based on Solomonoff-type prior has been proposed here before by Wei_Dai , and has also been mentioned by Marcus Hutter. I’m not at all impressed by Nick Bostrom, that’s the point, or philosophy for that matter. The conclusions of philosophers—given relative uselessness of philosophy compared to science—ought to be taken as very low grade evidence.
Thank you for linking to Hutter’s talk, what an astounding mind. What a small world it is, I remember being impressed by him when I sat through his courses back at grad school, little knowing how much of my future perspective on map-building would eventually depend on his and his colleagues’ school of thought.
That presentation should be mandatory reading. In all Everett branches.
What is the reasonable probability you think I should assign to the proposition by some bunch of guys (with at most some accomplishments in highly non-gradable field of philosophy) led by a person with no formal education nor prior job experience nor quantifiable accomplishments, that they should be given money to hire more people to develop their ideas on how to save the world from a danger they are most adept at seeing? The prior here is so laughably low you can hardly find a study so flawed it wouldn’t be a vastly greater explanation for the SI behavior than it’s mission statement taken at face value, even if we do not take into account SI’s prior record.
Reference class is not up for grabs. If you want narrower reference class you need to substantiate why it should be so narrow.
edit: Actually, sorry it comes as unnecessarily harsh. But do you recognize that SI genuinely has a huge credibility problem?
The donations to SI only make sense if we are to assume SI has extremely rare survival ability vs the technological risks. Low priors for extremely rare anything are a tautology, not an opinion. The lack of other alternatives is evidence against SI’s cause.
What is this, the second coming of C.S. Lewis and his trilemma? SI must either be completely right and demi-gods who will save us all or they must be deluded fools who suffer from some psychological bias—can you really think of no intermediates between ‘saviors of humanity’ and ‘deluded fools who cannot possibly do any good’, which might apply?
I just wanted to point out that invoking DK is an incredible abuse of psychological research and does not reflect well on either you or Dymtry, and now you want me to justify SI entirely...
Alternatives would also be evidence against donating, too, since what makes you think they are the best one out of all the alternatives? Curious how either way, one should not donate!
No, it comes out of what SI members claim about themselves and their methods—better than science, we are more rational, etc etc etc. That really drives down the probability of anything in the middle between the claimed excellence and the incompetence compatible with the fact of making those claims (you need sufficient incompetence to claim extreme competence). If they didn’t want that sort of dichotomy they should have kept their extreme arrogance from surfacing. (Or alternatively they wanted this dichotomy, to drive some people into fallacies from politeness)
Do you have a disagreement besides fairly stupid rhetoric? The lack of alternatives is genuinely evidence against SI’s cause, whereas presence of alternatives would genuinely make it unlikely that either of them is necessary. Yep, it’s very curious, and very inconvenient for you. The logic is sometimes impeccably against what you like. Without some seriously solid evidence in favour of SI, it is a Pascalian wager as the chance of SI making a difference is small.
I’ll rephrase: your argument from alternatives is as much bullshit as invoking Dunning-Kruger. Both an argument and its opposite cannot lead to the same conclusion unless the argument is completely irrelevant to the conclusion. If alternatives matter at all, there must be some number of alternatives which reflect better on SI than the other numbers.
It’s not an argument and it’s opposite. One of the assumptions in either argument is ‘opposite’, that could make distinction between those two assumptions irrelevant but the arguments themselves remain very relevant.
I take as other alternatives everyone who could of worked on AI risk but didn’t, because I consider it to be an alternative not to work on AI risk now. Some other people take as other alternatives people working on precisely the kind of AI risk reduction that SI works on. In which case the absence of alternatives—under this meaning of ‘alternatives’ - is evidence against SI’s cause—against the idea that one should work on such AI risk reduction now. There should be no way how you can change—against same world—meanings of the words and arrive at different conclusion; it only happens if you are exercising in the rationalization and rhetoric. In electromagnetism if you are to change right hand rule to left hand rule every conclusion will stay the same; in reasoning if you wiggle what is ‘alternatives’ that should not change the conclusion either.
This concludes our discussion. Pseudologic derived from formal maxims and employing the method of collision (like in this case, colliding ‘assumption’ with ‘argument’) is too annoying.
I hope you don’t mind if I don’t reply to you any further until it’s clear whether you’re a Dmytry sockpuppet.
If an account isn’t actualy Dmytry but instead just someone who thinks the same way that Dmytry does there seems we can just treat them the same way anyhow. After all, ten people who act like Dmytry seems just as bad as Dmytry with ten accounts.
They could just think the same way and be borrowing vocabulary and ideas without actually being as bad as him.
So the choice then would be whether to give the potential sockpuppet the benefit of the doubt and allow treatment of them to asymptotically approach the treatment of known Dmytry accounts to the extent that and for as long as they make Dmytry-like posts and for as long as it looks like the comments could be anomalies. ie. There is the expectation of either a regression to the mean or that an actual new user will be capable of learning from feedback.
If it is assumed that the accounts are sockpuppets then they immediately get treated without the benefit of doubt and with the additional loading penalty given to sockpuppets for being sockpuppets.
My comment on Dunning-Kruger effect is the second highest ranked comment in my post history or so.
Also: this thread is too weird.
And also similar issues with English as a second language. I agree it’s Dmytry, but not a sockpuppet. He didn’t go to great lengths to hide that he was private_messaging. The new JaneQ account posted some posts (as opposed to comments), thanks to its positive karma balance. I figure that Dmytry wanted to make a post, so created a new account without huge negative karma (and thus be able to post). I.e. I think he’s just trying to circumvent the karma system, not deceive people.
“Just”.
People are welcome to abandon an account when they realize they have irrevocably destroyed their reputation and wish to start again and try not being an asshat. They are not welcome to use multiple accounts to subvert the karma system.
Please review the context. You will notice that gwern is arguing with two ‘people’ in this discussion. Both of them, by your own prediction, are Dmytry. Using multiple accounts to support each other in a single argument is exactly what ‘sockpuppetry’ is all about. To put it mildly: I don’t like it.
You’re right, I missed that.
Yeah, I’ve been considering that theory for a while myself, as they share a few salient characteristics, but I’ve been unable to work out to my satisfaction what evidence would make it clear one way or the other. I’d be interested in your thoughts on the subject.
Let’s avoid inflationary use of Pascal’s wager.
Ah well, those can’t be the singularitarians he’s talking about then. He doesn’t name any names, leaving it to Anonymous to do so, then responds by saying “I wasn’t going to name names, but...” and then continuing not to name names. I predict a no true Scotsman path of retreat if you take your argument to him.
It’s not clear to me how approaching your response with an assumption of bad faith will convince him or his readers of the correctness of your position. Let us know how it works out for you.
I’m not assuming bad faith, just observing a lack of specifics about who he is talking about. But I’m not intending to make any response there, not being as informed as, say, ciphergoth on the SI’s position.
I would have preferred that he use my (even more passive-aggressive) approach, which is to say, “I’m not going to name any names[1]”, and then have a footnote saying “[1] A ‘name’ is an identifier used to reference a proper noun. An example of a name might be ‘Singularity Institute’.”
Get it? You’re not “naming names”, you’re just giving an example of name in the exact neighborhood of the accusation! Tee hee!
(Of course, it’s even better if you actually make the accusation directly, but that’s obviously not an option here.)
Who has constant marginal utility of money up to $1,000,000,000?
Consequences can’t be inflated to make up for arbitrarily low probabilities. Consequences are connected: if averting human extinction by proliferation of morally valueless machinery is super valuable because of future generations, then the gains of averting human extinction by asteroids, or engineered diseases, will be on the same scale.
It cost roughly $100 million to launch a big search for asteroids that has now located 90%+ of large (dinosaur-killer size) asteroids, and such big impacts happen every hundred million years or so, accompanied by mass extinctions, particularly of large animals. If working on AI, before AI is clearly near and better understood, had a lower probability of averting x-risk reduction per unit cost than asteroid defense, or adding to the multibillion dollar annual anti-nuclear proliferation or biosecurity budgets, or some other intervention, then it would lose.
“Some nonzero chance” isn’t enough, it has to be a “chance per cost better than the alternatives.”
I should have said something about marginal utility there. Doesn’t change the three tests for a Pascal scam though.
The asteroid threat is a good example of a low-probability disaster that is probably not a Pascal scam. On point (1) it is fairly lottery-like, insofar as asteroid orbits are relatively predictable—the unknowns are primarily “known unknowns”, being deviations from very simple functions—so it’s possible to compute odds from actual data, rather than merely guessing them from a morass of “unknown unknowns”. It passes test (2) as we have good ways to simulate with reasonable accuracy and (at some expense, only if needed) actually test solutions. And best of all it passes test (3) -- experiments or observations can be done to improve our information about those odds. Most of the funding has, quite properly, gone to those empirical observations, not towards speculating about solutions before the problem has been well characterized.
Alas, most alleged futuristic threats and hopes don’t fall into such a clean category: the evidence is hopelessly equivocal (even if declared with a false certainty) or missing, and those advocating that our attention and other resources be devoted to them usually fail to propose experiments or observations that would imrove that evidence and thus reduce our uncertainty to levels that would distinguish them from the near-infinity of plausible disaster scenarios we could imagine. (Even with just the robot apocalypse, there are a near-infinity of ways one can plausibly imagine it playing out). Same, generally speaking, for future diseases—there may well be a threat lying in there, but we don’t have any general ways of clearly characterizing specifically what those threats might be and thus distinguishing them from the near-infinity of threats we could plausibly imagine (again generally speaking—there are obviously some well-characterized specific diseases for which we do have such knowledge).
Who has constant marginal utility of people up to 1000000000 people? (To answer the rhetorical question—no one.)
This reminds of of Jaynes and transformation groups—establish your prior based on transforms that leave you with the same problem. I find this makes short work of arbitrary assertions that want to be taken seriously.
Someone who’s already got many billions? (But then again, for such a person a 1/1000 chance of getting one more billion wouldn’t even be worth the time spent to participate in such a lottery, I suppose.)
From zero up to $1,000,000,000.
I do, in that there are nonfatal actions that i would not take in exchange for that much money. Of course, at numbers over several hundred thousand, money loses unit utility very fast. One billion dollars has significantly less than one thousand times the value to me of one million dollars, because the things I can buy with a billion dollars are less than one thousand times as valuable to me as the things I can buy with a million.
Seems like there is more going on than just “Do transhumanists endorse Pascalian bargains?” Because of confidence levels inside and outside an argument, the fact that someone (e.g. SI) makes an argument that a particular risk has a non-negligible probability does not mean that someone examining this claim should assign a non-negligible probability. It’s very possible for someone thinking about (e.g.) AI risk to assign low probabilities and thus find themselves in a Pascalian situation even if SI argues that the probability of AI risk is high.
Indeed. As to why I find extreme consequences from general AI highly unlikely, see here. Alas, my main reason is partly buried in the comments (I really need to do a new post on this subject). It can be summarized as follows: for basic reasons of economics and computer science, specialized algorithms are generally far superior to general ones. Specialized algorithms are what we should hope for or fear, and their positive and negative consequences occur a little at a time—and have been occurring for a long time already, so we have many actual real-world observations to go by. They can be addressed specifically, each passing tests 1-3, so that we can solve these problems and achieve these hopes one specialized task at a time, as well as induce general theories from these experiences (e.g. of security), without getting sucked into any of the near-infinity of Pascal scams one could dream up about the future of computing and robotics.
It would be better to present, as your main reason, “the kinds of general algorithms that humans are likely to develop and implement, even absent impediments caused by AI-existential risk activism, will almost certainly be far inferior to specialized ones”. That there exist general-purpose algorithms which subsume the competitive abilities of all existing human-engineered special-purpose algorithms, given sufficient advantages of scale in number of problem domains, is trivial by the existence proof constituted by the human economy.
Put another way: There is some currently-unsubstitutable aspect of the economy which is contained strictly within human cognition and communication. Consider the case where the intellectual difficulties involved in understanding the essence of this unsubstitutable function were overcome, and it were implemented in silico, with an initial level of self-engineering insight already equal to that which was used to create it, and with starting capital and education sufficient to overcome transient learning-curve effects on its initial success. There would then be some fraction of the economy directed by the newly engineered process. Would this fraction of the economy inevitably be at a net competitive advantage, or disadvantage, relative to the fraction of the economy which was directed by humans?
If that fraction of the economy would have an advantage, then this would be an example of a general algorithm ultimately superior to all contemporarily-available specialized algorithms. In that case, what you claim to be the core of your argument would be defeated; the strength of your argument would instead have to come from a focus on the reasons why it were improbable that anyone had a relevant chance of ever achieving this kind of software substitute for human strategy and insight (that is, before everyone else was adequately prepared for it to prevent catastrophe), and that even to the point that supposing otherwise deserves to be tarred with a label of “scam”. And if the software-directed economy would have a disadvantage even at steady state, then this would be a peculiar fact about software and computing machinery relative to neural states and brains, and it could not be assumed without argument. Digital software and computing machinery both have properties that have made them, in most respects, much more tractable to large returns to scale from purposeful re-engineering for higher performance than neural states and brains, and this is likely to continue to be true into the future.
I don’t understand your reasoning here. If you have a general AI, it can always choose to apply or invent a specialized algorithm when the situation calls for that, but if all you have is a collection of specialized algorithms, then you have to try to choose/invent the right algorithm yourself, and will likely do a worse (possibly much worse) job than the general AI if it is smarter than you are. So why do we not have to worry about “extreme consequences from general AI”?
Skill at making such choices is itself a specialty, and doesn’t mean you’ll be good at other things. Indeed, the ability to properly choose algorithms in one problem domain often doesn’t make you an expert at choosing them for a different problem domain. And as the software economy becomes more sophisticated these distinctions will grow ever sharper (basic Adam Smith here—the division of labor grows with the size of the market). Such software choosers will come in dazzling variety: they like other useful or threatening software will not be general purpose. And who will choose the choosers? No sentient entity at all—they’ll be chosen they way they are today, by a wide variety of markets, except that there too the variety will be far greater.
Such markets and technologies are already far beyond the ability of any single human to comprehend, and that gap between economic and technological reality and our ability to comprehend and predict it grows wider every year. In that sense, the singularity already happened, and long ago.
Can you expand on this? The way you say it suggests that it might be your core objection to the thesis of economically explosive strong AI. -- put into words, the way the emotional charge would hook into the argument here would be: “Such a strong AI would have to be at least as smart as the market, and yet it would have been designed by humans, which would mean there had to be a human at least as smart as the market: and belief in this possibility is always hubris, and is characteristically disastrous for its bearer—something you always want to be on the opposite side of an argument from”? (Where “smart” here is meant to express something metaphorically similar to a proof system’s strength: “the system successfully uses unknowably diverse strategies that a lesser system would either never think to invent or never correctly decide how much to trust”.)
I guess, for this explanation to work, it also has to be your core objection to Friendly AI as a mitigation strategy: “No human-conceived AI architecture can subsume or substitue for all the lines of innovation that the future of the economy should produce, much less control such an economy to preserve any predicate relating to human values. Any preservation we are going to get is going to have to be built incrementally from empirical experience with incremental software economic threats to those values, each of which we will necessarily be able to overcome if there had ever been any hope for humankind to begin with; and it would be hubris, and throwing away any true hope we have, to cling to a chimerical hope of anything less partial, uncertain, or temporary.”
Would you agree that humans are in general not very good at inventing new algorithms, many useful algorithms remain undiscovered, and as a result many jobs are still being done by humans instead of specialized algorithms? Isn’t it possible that this situation (i.e., many jobs still being done by humans, including the jobs of inventing new algorithms) is still largely the case by the time that a general AI smarter than human (for example, an upload of John von Neumann running at 10 times human speed) is created, which at a minimum results in many humans suddenly losing their jobs and at a maximum allows the AI or its creators to take over the world? Do you have an argument why this isn’t possible or isn’t worth worrying about (or hoping for)?
To answer your second sentence on, one consideration is that it is highly questionable whether scanning and uploading is even possible in any practical sense, as people who actually work with brain preservation on a daily basis and would love to be able to extract state from the preserved material seem to consider the matter: It’s “possible” philosophically, but not at all practically. This suggests that it’s low enough feasibility at present that even paying serious attention to it may be a waste of time of the “Pascal’s scam” form described in the linked post (whether the word “scam” is fair or not).
If uploads are infeasible, what about other possible ways to build AGIs? In any case, I’m responding to Nick’s argument that we do not have have to worry about extreme consequences from AGIs because “specialized algorithms are generally far superior to general ones”, which seems to be a separate argument from whether AGIs are feasible.
When some day some people (or some things) build an AGI, human-like or otherwise, it will at that time be extremely inferior to then-existing algorithms for any particular task (including any kind of learning or choice, including learning or choice of algorithms). Culture, including both technology and morality, will have changed beyond any of our recognitions long before that. Humans will already have been obsoleted for all jobs except, probably, those that for emotional reasons require interaction with another human (there’s already a growth trend in such jobs today).
The robot apocalypse, in other worlds, will arrive and is arriving one algorithm at a time. It’s a process we can observe unfolding, since it has been going on for a long time already, and learn from—real data rather than imagination. Targetting an imaginary future algorithm does nothing to stop it.
If, for example, you can’t make current algorithms “friendly”, it’s highly unlikely that you’re going to make the even more hyperspecialized algorithms of the future friendly either. Instead of postulting imaginary solutions to imaginary problems, it’s much more useful to work empirically, e.g. on computer scecurity that mathematically prevents algorithms in general from violating particular desired rights. Recognize real problems and demonstrate real solutions to them.
The phrasing suggests a level of certainty that’s uncalled for for a claim that’s so detailed and given without supporting evidence. I’m not sure there is enough support for even paying attention to this hypothesis. Where does it come from?
(Obvious counterexample that doesn’t seem unlikely: AGI is invented early, so all the cultural changes you’ve listed aren’t present at that time.)
All of these kinds of futuristic speculations are stated with false certainly—especially the AGi-is-very-important argument, which is usually stated with a level of certainty that is incredible for an imaginary construct. As for my evidence, I provide it in the above “see here” link—extensive economic observations have been done on the benefits of specialization, for example, and we have extensive experience in computer science with applying specialized vs. generalized algorithms to problems and assessing their relative efficiency. That vast amount of real-world evidence far outweighs the mere speculative imagination that undergirds the AGI-is-very-important argument.
Given the benefits of specialization, how do you explain the existence of general intelligence (i.e. humans)? Why weren’t all the evolutionary niches that humans current occupy already taken by organisms with more specialized intelligence?
My explanation is that generalized algorithms may be less efficient than specialized algorithms when specialized algorithms are available, but inventing specialized algorithm is hard (both for us and for evolution) so often specialized algorithms simply aren’t available. You don’t seem to have responded to this line of argument...
The belief that an error is commonly made doesn’t make it OK in any particular case.
(When, for example, I say that I believe that AGI is dangerous, this isn’t false certainty, in the sense that I do believe that it’s very likely the case. If I’m wrong on this point, at least my words accurately reflect my state of belief. Having an incorrect belief and incorrectly communicating a belief are two separate unrelated potential errors. If you don’t believe that something is likely, but state it in the language that suggests that it is, you are being unnecessarily misleading.)
To rephrase my question, how confident are you of this, and why? It seems to me quite possible that by the time someone builds an AGI, there are still plenty of human jobs that have not been taken over by specialized algorithms due to humans not being smart enough to have invented the necessary specialized algorithms yet. Do you have a reason to think this can’t be true?
ETA: My reply is a bit redundant given Nesov’s sibling comment. I didn’t see his when I posted mine.
I am far more confident in it than I am in the AGI-is-important argument. Which of course isn’t anywhere close to saying that I am highly confident in it. Just that the evidence for AGI-is-unimportant far outweighs that for AGI-is-important.
The upload thread talks about the difficulties in making an upload of a single specific adult human, which would have the acquired memories and skills from the biological human reproduced exactly. (Admittedly, “an upload of John von Neumann”, taken literally, is exactly this.) A neuromorphic AI that skips the problem of engineering a general intelligence by copying the general structure of the human brain and running it in emulation doesn’t need to be based on any specific person, though, just a general really very good understanding of the human brain, and it only needs to be built to the level of a baby with the capability to learn in place, instead of somehow having memories from a biological human transferred to it. The biggest showstopper for practical brain preservation seems to be preserving, retrieving and interpreting stored memories, so this approach seems quite a bit more viable. You could still have your von Neumann army, you’d just have to raise the first one yourself and then start making copies of him.
I don’t think Nick made a good case as to why these movements / belief systems deserve to be call “scams” and more importantly deserve to be ignored (in favor of “spending more time learning about what has actually happened in the real world”). The fact that certain hopes and threats share certain properties (which Nick numbered 1-3 in his post) is unfortunate, but I didn’t find any convincing arguments in his post showing why these hopes and threats should therefore be ignored.
(My overall position, which I’ll repeat in case anyone is curious and doesn’t want to dig up my past comments, is that Pascal’s Wager is a difficult unsolved philosophy problem, whose solution probably requires a great deal of progress in decision theory (e.g., how to handle logical uncertainty and deal with running on unreliable hardware) and ethics (e.g., how much utility should I really assign to certain outcomes). In the face of such philosophical uncertainty, my instinct is to spend a portion of my resources on the philosophical problem itself, a portion on the most promising “wager”, and the rest on other more mundane priorities. I don’t have a good argument for why this is the right (or rational) thing to do, but neither do I see any good arguments against it.)