MathWizard
Good things are good
No bio...
User ID: 164
MathWizard
Good things are good
1yr ago
Pretty much all legitimate justifications for racism rely on inaccurate proxies for other things we actually care about. I think you can make arguments in favor of using it in the absence of better knowledge, but once more direct signals have been acquired the race no longer serves a useful purpose.
Since I am white and was raised by white parents among mostly other white people, I can reasonably expect that the average white person is more likely to be similar to me than the average black person. We'll be more likely to have similar cultural knowledge, values, habits, etc. But my black neighbor who I actually know and happens to be a christian pastor has way more in common with me than the average white Californian.
In the past race was a very strong proxy for nationality, culture, and loyalty. In modern times it is a weak signal unless you live in a predominantly monoethnic country.
MathWizard
Good things are good
1yr ago
racismschool.tumblr.com links to a user on tumblr named "racismschool". It appears to be empty now, though I can't tell if that's because they deleted everything or because I don't have a tumblr account. Presumably they're the person who made the thing that he's referring to in this post.
MathWizard
Good things are good
1yr ago
But it specifically applies pressure against negative behaviors, at least according to the subjective perceptions of the mocker. X behavior is stupid/bad -> Y group of people mock it -> Z group of people care about Y's opinion and/or avoiding mockery in general and do X less or fail to start doing X -> less X exists. If the mocker has good subjective opinions and targets, then this is a net positive since it reduces the prevalence of stupid/bad behaviors. If the mocker has bad subjective opinions and targets, then this is a net negative since it reduces the prevalence of good behaviors that have been mislabeled.
Speed running.... makes speed running look cool? Like, maybe it encourages people to try really hard and dedicate themselves to a task, or peer into the underlying mechanics of games and pedantically look for flaws that they can exploit which maybe increases their ability as a hacker/programmer/anti-hacker? But the most likely outcome is that it makes people more likely to become speed runners. I suppose one could make a similar argument about a lot of hobbies, but a lot of hobbies have depth or broadly interesting components, while speedrunning is about pedantic details and weird edge cases.
Like, if someone has a hobby of using tweezers to arrange tiny colored grains of sand into beautiful artwork, that's kind of cool. I wouldn't do it, it seems like more time and effort than it's worth to me, but if someone else wants to do that good for them, and maybe at the end I'll look at the picture they make. If someone has a hobby of using tweezers to arrange tiny grains of sand into binary representations of the code to retro videogames, that's stupid. It takes similar levels of pedantic effort to perfectly arrange each grain of sand into the right shape, but in the end you have a bunch of dots of sand and the binary representation doesn't do anything because operating systems can't read sand, so it's functional equivalent to a random arrangement of sand. I suppose if someone had some property of their brain that makes this hobby enjoyable for them I'm not going to say they're not allowed to do it, but to me it's boring both to do AND to hear about or watch, while the colored sand piles are boring to do but might be worth watching a little bit. I feel that videogames are more analogous to the colored art sand: pragmatically useless towards survival in the real world but interesting to experience or view, while speedrunning is analagous to the binary representations: similarly complex in function but more pedantic and way less interesting.
All to say that pressure towards making people more interested in speedrunning is negative because it increases the amount of people with boring hobbies, which funges against more interesting hobbies that they could have. And while this is mostly a subjective opinion from me as someone who thinks speedrunning is boring, I think there is some way in which speedrunning is objectively worse than most hobbies, including broader videogaming, although I'm not entirely sure exactly how to formalize, hence vaguely gesturing at it via the above analogy.
MathWizard
Good things are good
1yr ago
Is that actually true? Like, it sounds plausible that it could be true, but it also seems plausible that it helps shape culture and behavioral norms, because people are less likely to do things that lead to scorn and mockery. Granted, people probably spend an inordinate amount of effort talking shit on the internet above and beyond its actual value, but there is the potential for actual value buried in there.
MathWizard
Good things are good
1yr ago
A very interesting read, thank you for posting this here. I think what I'm most interested in that was briefly mentioned but not gone into in detail is the nuances of being an unusually intelligent person within this group and the social dynamics that could result in. Did that make it harder to fit in because you're different from everyone else in some noticeable way? Is there stigma against intelligent people, or is it simply the stigma against perceived cowardice that correlates with intelligence? Did being a strong, competent, aggressive person like everyone else mean they respected you anyway and didn't care about you being smart, or did you have to prove yourself above and beyond what the more typical infantry people did? Or were you able to leverage your intelligence towards making your life even easier than everyone else?
MathWizard
Good things are good
1yr ago
I strongly suspect that having a sane, competent government and culture increases the probability of friendly AI, which will require care, cooperation, and long-termism not typically present in a society dominated by extremists (from either side).
As such, planting a tree restabilizing balance in arts and its impact in culture (not tipping the balance oppositely to give right wing extremists an edge, but a return towards balance) may in fact help our grandkids actually retain control of the shade it produces. It's unlikely to be the sole determining factor, but it's one more marginal weight on the scale.
MathWizard
Good things are good
1yr ago
The best time to plant a tree was 20 years ago. The second best time is now.
This culture war has been happening for decades, it will likely continue for decades into the future. We need to do other stuff to help ourselves in the short term, but not doing this is just going to make the problem continue to get worse in the long run.
MathWizard
Good things are good
1yr ago
I wouldn't straight up cut someone off if they were already a friend for other reasons and that was the only thing about them I disliked. But it would be a yellow flag which would make me less comfortable around them. Because stuff like that rarely shows up in isolation. I've never actually had the issue show up, because the type of people I typically hang out with are so far from that archetype that it's not even a remote possibility.
MathWizard
Good things are good
1yr ago
-
There's a negative feedback loop here that prevents this from being reliably true. That is, in an environment where it is possible/easy/profitable to consistently get away with unethical behavior, more people do it until it becomes common enough that people respond and become less trusting in order to protect themselves. This is largely what distinguishes high-trust societies versus low trust. Additionally, the expected cost of unethical behavior is the probability of being caught multiplied by the penalty, which means that you can stabilize at higher levels of ethics by ramping up the penalties, be that financial, reputational, or justice. I think this is largely why upper and middle class communities tend to be higher trust than lower class communities. If you have lots of money, stable long-term friends, and a job that relies on maintaining a professional bearing and reputation, then you have more to lose even if you do unethical but technically legal things. None of my friends have, to my knowledge, ever shoplifted in their lives, and if I found out they did I would lose respect for them and shame them for it. Because that's not the kind of person I want to hang out with, even if they were stealing from some soulless megacorp and there's no risk of them stealing from me. Ethical people who can reliably recognize each other and group together can create better. happier, more stable subcommunities by filtering, which creates a hard-to-measure cost to being unethical.
-
The rational game theoretic perspective says to maximize your utility function, which if you are not a sociopath might itself contain a term for ethics. Don't fall into the trap thinking that people are profit-maximizing corporations, sometimes good deeds are their own reward. A large part of why I do ethical things even if I might get away with it is because one of my terminal values is the desire to be a good person. I feel guilty when I do bad things, and I feel good/proud/accomplished when I do good things, especially if there was a temptation to do a selfish bad thing and I chose to do the right thing anyway. Most people have something like that. The philosophical argument that you should be good because if everyone is bad you'd be worse off is weak, it was always weak. It's not the actual reason to be good, which is that it is good to be good, and if you're not an evil sociopath your utility function will care about that in its own right. If someone is an evil sociopath then there's not much the rest of us can do to convince them to care, all we can do is arrange society such that unethical behavior is punished harshly enough that the rationally selfish unethical behaviors we can't punish are rare and minor.
MathWizard
Good things are good
1yr ago
This in turn incentivizes would-be-dictators to make tons of broad and vague rules with selective enforcement. If there is such a byzantine maze of laws that almost never get enforced, then nobody pays attention to them and then everyone ends up technically in violation of some sort of rule. Then the enforcer can pick and choose who gets penalized based on their own internal reasons, but always has some sort of justification in the actual rules to pin it to.
MathWizard
Good things are good
1yr ago
Not sure if this belongs here or in SQS, but it could either be a small question I don't understand or a discussion depending on whether or not people disagree about the answer.
Why did support for Ukraine split along the left/right the way it did (at least in the U.S.), when typically one would expect it to go the other way. That is, the right is usually more pro-military, pro-military intervention, and patriotic defending of one's homeland. Even though the right tends to be more focused on domestic issues and oppose foreign aid, military support tends to be the exeption. Although there was bipartisan support of the Iraq war (at least in the aftermath of 9/11) the Republicans were more strongly in favor of it and stayed in favor of it for longer. If Russia had threatened to invade the U.S. the Republicans would have been not only gung-ho about repelling them but also about retaliating and obliterating them in revenge so that none would dare try ever again. So you would think they would sympathize with Ukrainians as similarly patriotic defenders of their home turf, while the left would be all peace and let's try to get along and diplomatically convince the invaders to stop without violence, or something like that.
But that's not what happened. Why?
Is it just because the left has been harping on about Putin for years so hopped on the anti-Russia train too quickly and the right felt compelled to instinctively oppose them? If China had invaded Ukraine (for some mysterious reason) would the right be pro-Ukraine and the left opposing intervention because they don't want to piss off China (and accusing Ukraine of being nazis as an excuse)? That is, is there something specific to Ukraine/Russia that caused this divide here specifically, or am I misunderstanding the position of each side regarding military intervention in general (or has it changed in the past few decades and my beliefs used to be accurate but no longer are)?
MathWizard
Good things are good
1yr ago
Knowing very little about tax code, I think this shouldn't work because the jump from $50k -> $5million would be counted as profit in some sense, similar to if you buy $50k of stock and then sell it for $5 million. I think it's called an "asset appreciation tax"? So your taxable income would go up by 4.95 million from having an asset you paid $50k for go up in valuation, and then down by $5 million for the donation, giving you a net -$50k (because you spent $50k that you then donated). But I'm not certain this is how it actually works.
MathWizard
Good things are good
1yr ago
Okay but you don't seem to be arguing against categorizing people, you're mostly just suggesting that accurate categories are superior to inaccurate categories. I'm not especially familiar with Wonderlic, but some quick Googling suggests it's an employment-specific intelligence test. Which means it's is not literally measuring merit at a job, it's categorizing people based on questions that it thinks are a proxy for job skill (unless the job literally consists of answering Wonderlic questions). People don't go around politically identifying with in discrete groups based on their intelligence, but screening out unintelligent people is still a form of grouping people up and discriminating based on something that isn't directly merit, but is strongly correlated with it.
MathWizard
Good things are good
1yr ago
Except I'm not just coyly avoiding references to HBD but secretly referring to that alone, I'm deliberately including a broader range of groups like using someone's membership in the KKK to know you don't want to be friends with them, or using someone's presence on a sex offender registry to avoid hiring for daycare positions. This is also a form of using someone's group membership to pre judge them. Of course there are differences, I'm not trying to argue "these are the same therefore in order to be consistent you have to support or oppose both". My point is more "there are tons of differences between these things, which ones are the actually relevant distinctions and why?"
MathWizard
Good things are good
1yr ago
I'm not sure that's stable though, because it may inevitably slippery slope its way into progressivism. That is, this optimal state depends on universal but not-common knowledge: the utilitarian version has to actually be a secret. Because if you are publicly insisting on ignoring group memberships and everyone knows that person A is discriminating against group X in a not-secret way, then the public persona is forced to denounce them as a X-ist in order to maintain consistency. But if everyone using the utilitarian version in practice, then it's hard to keep that a secret from everyone else (who is doing the same thing). And if only the smart well-behaved rationalists who can be trusted to discriminate responsibly use the utilitarian version while everyone else uses the liberal version, then a higher fraction of smart well-behaved rationalists would be discovered and denounced as X-ist creating a stereotype against them.
Maybe it works if you restrict the secret utilitarian version to only cases where there's absolutely no conceivable way of being discovered.
MathWizard
Good things are good
1yr ago
I would describe her as economically socialist, socially conservative/centrist. Which ought to be broadly coherent as positions despite being rare in practice. She's all in on Bernie Sanders, redistributive policies, and taxing and regulating the crap out of corporations and rich people. None of which demands social justice censorship, feminism, or collectivization of people into little intersecting boxes and then blaming straight white men for all of society's ills.
MathWizard
Good things are good
1yr ago
Except smoking tobacco also kill you in some subtle and devious ways, and it took us hundreds of years to actually get formal scientific data for that. It seems entirely reasonable that smoking marijuana would have similar effects which are significantly above zero but non-obvious due to the long-term nature of it.
Absent hard data pointing in either direction, it seems reasonable to ballpark guess that it's probably has similar effects to things most similar to it, of which tobacco cigarettes seem like the closest comparison (which are a thing people smoke and inhale), not watermelons (which are a fruit that people eat).
More importantly, if we have no idea, you shouldn't report it in a table of things that we do have an idea about. You should leave it off the table or say "we have no idea"
MathWizard
Good things are good
1yr ago
In many cases it's where merit is difficult to measure up front. If you are looking at job applications, you can't literally perceive merit until you've already hired someone, and thus excluded the other candidates. If you're trying to avoid rapists, you can't perceive merit until they've literally attempted or succeeded at raping someone. If you're looking for romantic partners, a 1 minute analysis based on group membership is 120 times cheaper than going on a 2 hour date, and thus potentially worthwhile if the amount of information you can extract from it is 1% as much.
The optimal Bayesian thing to from a purely selfishly rational perspective seems to be using immediately identifiable group membership as a first screening pass (establishing the prior) and then update with more direct merit measures as/if they become available.
MathWizard
Good things are good
1yr ago
-
Does this suggest that if such an obnoxious rational person does have the statistical sophistication to draw such distinctions and make decisions in secret, then it's okay? That is, stereotypes are bad in general if widespread because normies will abuse them, but the actual rational analysis is fine if used in an isolated and secret way that normies don't find out about?
-
What defines "obnoxious" here? Is rationalism itself defined as obnoxious because it cares about pedantic details that normies don't? Or is it merely the social obliviousness of nerdy rationalists who oversimplify everything and miss the forest for the trees, such that a more sophisticated rational intelligence that understands and compensates for normies would not be obnoxious?
MathWizard
Good things are good
1yr ago
I think I defined it fairly unambiguously:
Let's suppose that we know with certainty that people in group X have a statistically higher rate of bad feature Y compared to the average population, whether that be criminality, laziness, low intelligence, or are just unpleasant to be around. I'm taking the fact that this is accurate as an axiom. The actual proportion of people in group X with feature Y is objectively (and known to you) higher than average, but is not universal. That is, Y is a mostly discrete feature, and we have 0 < p < q < 1 where p is the probability of a randomly sampled member of the public has Y, and q is the probability that a randomly sampled member of q has Y.
It's "accurate" in that the literal proportion of people with trait Y in the general population and the group, in real life are p and q respectively, with p < q, and we also believe this to be true. As opposed to an inaccurate stereotype representing a false belief. In-so-far as Y actively impacts merit, then membership in X does provide a real signal correlated with merit.
Obviously actually measuring merit directly is superior to imperfect correlations, but if you are, for instance, hiring someone for a job, imperfect correlations are the only thing you have up until you actually hire someone and watch them perform the job. Literally everything you judge on is going to be an imperfect correlation of some form, so it's just a question of which ones you use and how much weight you put on each.
MathWizard
Good things are good
1yr ago
In what contexts are accurate prejudice/biases acceptable justification for discrimination?
I want to consider a broad range of groups including both involuntary/innate characteristics such as race, gender, and IQ, as well as more voluntary categories such as religion, political ideology, or even something like being in the fandom for a certain TV show, expressing a preference for a certain type of food, or having bad personal grooming. This is a variable that your answer might depend upon.
Let's suppose that we know with certainty that people in group X have a statistically higher rate of bad feature Y compared to the average population, whether that be criminality, laziness, low intelligence, or are just unpleasant to be around. I'm taking the fact that this is accurate as an axiom. The actual proportion of people in group X with feature Y is objectively (and known to you) higher than average, but is not universal. That is, Y is a mostly discrete feature, and we have 0 < p < q < 1 where p is the probability of a randomly sampled member of the public has Y, and q is the probability that a randomly sampled member of q has Y. Let's leave the causation as another variable here: maybe membership in X increases the probability of Y occurring, maybe Y increases the probability of joining X (in the case of voluntary membership), maybe some cofactor causes both. This may be important, as it determines whether discouraging people from being in group X (if voluntary) will actually decrease the prevalence of Y or whether it will just move some Ys into the "not X" category.
Another variable I'll leave general is how easy it is to determine Y directly. Maybe it's simple: if you're interacting with someone in person you can probably quickly tell they're a jerk without needing to know their membership in Super Jerk Club. Or maybe it's hard, like you're considering job applications and you only know a couple reported facts, which include X but not Y and you have no way to learn Y directly without hiring them first.
When is it okay to discriminate against people in group X? The far right position is probably "always" while the far left would be "never", but I suspect most people would fall somewhere in the middle. Few people would say that it would be okay to refuse to hire brown-haired people if it were discovered that they were 0.1% more likely to develop cancer and thus leave on disability. And few people would say that it's not okay to discriminate against hiring convicted child rapists as elementary school teachers on the basis that they're a higher risk than the average person. (if you are such a person though, feel free to speak up and explain your position).
So for the most part our variables are:
-Group membership voluntariness
-Feature Y's severity and relevance to the situation
-The situation itself (befriending, hiring, electing to office)
-Ease of determining feature Y without using X as a proxy
-Causality of X to Y
Personally, I'm somewhere between the classically liberal "it's okay to discriminate against voluntary group membership but not involuntary group membership" and the utilitarian "it's okay to discriminate iff the total net benefit of the sorting mechanism is higher than the total cost of the discrimination against group members, taking into account that such discrimination may be widespread", despite the latter being computationally intractable in practice and requiring a bunch of heuristics that allow bias into the mix. I don't think I'm satisfied with the classically liberal position alone because if there were some sufficiently strong counterexample, such as someone with a genetic strain that made them 100x more likely to be a pedophile, I think I'd be okay with refusing child care positions to all such people even if they had never shown any other risk factors. But if there were a similar strain that made them 10% more likely I don't think it would be fair to do this, because it's such a low base rate that 10% doesn't do much to offset the cost of the discrimination. Also the utilitarian position allows for stricter scrutiny applied for more serious things like job applications (which have a huge cost if systematically discriminating against X) versus personal friendships (if people refuse to befriend X because they don't like Y, those people can more easily go make different friends or befriend each other, so the systemic cost is lower)
But I'd love to hear more thoughts and perspectives, especially with reasoning for why different cases are and are not justified under your philosophical/moral framework.
MathWizard
Good things are good
1yr ago
Easy solution: save backups of chat histories even when people block users or delete accounts.
MathWizard
Good things are good
1yr ago
Pushing back a bit on Gungeon. I've been playing it a lot with my fiance and although we have a lot of fun, in some ways it's actually more difficult than single player (for me) because she's not very good, which means she dies a lot. This means she spends a decent amount of time as a ghost and not a normal player, and also the way to resurrect a dead ally is by using a chest. Although it doesn't cost a key, it also uses up the chest so you don't get a weapon from it, meaning by the late game after she's died and been resurrected 4-5 times we are significantly underpowered.
OP does not specify the precise age of his son, but it's implied to be young, so I suspect the kid would spend the majority of time as a ghost given the difficult gameplay. There are easier co-op games out there.
MathWizard
Good things are good
1yr ago
The best argument I've heard in favor of unions is that the equivalent bargaining power of "a company" isn't "an employee" it's "all the employees".
Suppose we remove the distinction of capital versus labor, and suppose that we have two groups of people with disproportionate level of bottleneck in a production process. That is, if we have X people from the first group, and Y people from the second group, then the level of production is something like
f(X,Y) = A sqrt(X)P(Y)
where A is some constant, and P is 0 if Y is 0 and 1 if Y >= 1
That is, you only need one Y (the employer), but can have as many X as you want, but the more X you have in the same job the more diminishing returns you get. For each production process people can gather together and organize and form mutually consensual agreements to find some equilibrium level of X that makes this efficient. BUT, Y has disproportionate bargaining power here. If any individual X threatens to quit, their quitting drops the profits of the process by some small amount. But less than their average. The other X essentially pick up the slack, and the production keeps on going. But X is now unemployed and has 0 income, which is catastrophically awful and wasteful, as all of their potential labor is essentially being wasted unused. X quitting hurts themselves more than it hurts Y. But if Y threatens to quit then everything stops and everyone is at 0, so it's a credible threat.
But if all of the X form a union and threaten to quit/strike together, then again production stops entirely, just as if Y threatened to quit. So now they have equal bargaining power.
I'm pretty sure whoever I read this sort of argument from explained it way better than I just did, but I don't remember who or where (it might have been on the motte, so if whoever it was recognizes this argument as their own and can find the post, feel free to repost it and claim credit).
All priors collapse towards each other in the face of increasing amounts of evidence. Maybe you start 7% vs 33% that a random white vs black man is a violent criminal, and then if you learn they dress well and speak proper English it drops to 4% vs 5%. Or if you learn they have anger problems and are covered in tattoos it might go to 55% vs 60%. Given that genes have no almost direct causal impact on behavior except indirectly through other means such as IQ, personality, and cultural upbringing, it seems pointless to consider them when those things can be observed directly.
I agree with you that for random people on the street signals are more costly than they're worth and race can be useful as a quick hack to ballpark guess, I said as much in my previous post. But none of this implies that it
It only serves value in that it lets you guess at the direct predictors more quickly and easily than costly signals would. No priors carry immense amounts of residual predictive value even after controlling for more direct predictors. That's what makes them priors. The direct predictors are what we actually care about, and race is only useful in-so-far as it might be a faster way to guess at them if you don't already have them and don't want to spend the time and effort to acquire them properly. Which sounds reasonable for strangers, but less so for people you actually know.
More options
Context Copy link