I'll take a shot.

What we choose to measure affects what we choose to do. If I adopt the definition above, and I ask a wish machine to "minimize sexism", maybe it finds that the cheapest thing to do is to ensure that for every example of institutional oppression of women, there's an equal and opposite oppression of men. That's...not actually what I want.

So let's work backwards. Why do I want to reduce sexism? Well, thinking heuristically, if we accept as a given that men and women are interchangeable for many considerations, we can assume that anyone treating them differently is behaving suboptimally. In the office in the example, the dress code can't be all that helpful to the work environment, or the women would be subject to it. Sexism can be treated as a pointer to "cheap opportunities to improve people's lives". The given definition cuts off that use.

I think one possible answer is that your model of sexism, while internally consistent, is useless at best and harmful at worst, depending on how you interpret its output.

If your definition of sexism is completely orthogonal to morality, as your last bullet point implies, then it's just not very useful. Who cares if certain actions are "sexist" or "blergist" or whatever ? We want to know whether our goals are advanced or hindered by performing these actions -- i.e., whether the actions are moral -- not whether they fit into some arbitrary boxes.

On the other hand, if your definition implies that sexist actions very likely to be immoral as well, then your model is broken, since it ignores about 50% of the population. Thus, you are more likely to implement policies that harm men in order to help women; insofar as we are all members of the same society, such policies are likely to harm women in the long run, as well, due to network effects.

EDIT: Perhaps it should go without saying, but in the interests of clarity, I must point out that I have no particular desire to commit violence against anyone. At least, not at this very moment.

TL;DR: Some evidence points, and the rest my mind fills in by type 1 / pattern-matching / bias / etc., towards hypothetical you being fundamentally broken somewhere crucial, at BIOS or OS level to use a computer metaphor, though probably you can be fixed. I feel very strongly that this hypothetical you is not even worth fixing. This is something about myself I'd like to refine and "fix" in the future.

Well, the type 1 processes in my brain tell me that the most expedient, least "troublesome" way to solve the "problem" is to eliminate the source of the problem entirely and permanently, namely Hypothetical::TheOtherDave. This implies that there is a problem, and that it originates from you, according to whatever built-in system is screaming this to my consciousness.

Tracing back, it appears that in this scenario, I have strong beliefs that there is a major systemic error in judgment that caused "sexism" to be defined in that manner, and if the person is a "Feminist" that only applies techniques to solve "that kind" of "sexism", without particular concern for things that I consider sexism beyond "they might be bad things too, but not any more than any other random bad things, thus as a Feminist I'm not fighting against them", then I apparently see it as strong evidence that there is a generalized problem - to make a computer metaphor, one of the low-level primary computing functionalities, perhaps even directly in the instruction set implementation (though much more likely to be in the BIOS or OS, since it's rarely that "hardwired"), is evidently corrupted and is spreading (perhaps virally) wrongful and harmful reasoning throughout the mental 'system'.

Changing the OS or fixing and OS error is feasible, but very rarely happens directly from within the system, and usually requires specific, sometimes complex user input - there needs to be certain contexts and situations, probably combined with particularly specific or strong action taken by someone other than the "mentally corrupted" person, in order for the problem to be corrected.

Since the harm is continuous, currently fairly high in that hypothetical, and the cost of fixing it "properly" is rather high, I usually move on to other things while bashing my head on a wall figuratively in my mind and "giving up" on that person - I classify them as "too hard to help becoming rational", and they get this tag permanently unless something very rare (which I often qualify as a miracle) happens to nudge them sufficiently hard that there appears to be a convenient hack or hotfix that can be applied to them.

Otherwise, "those people" are, to my type-1 mind, worth much less instrumental value (though the terminal value of human minds remains the same), and I'll be much less reticent to use semi-dark-arts on them or otherwise not bother helping or promoting more correct beliefs. I'll start just nodding absentmindedly at whatever "bullcrap" political or religious statements they make, letting them believe they've achieved something and convinced me or whatever they'd like to think, just so I can more efficiently return to doing something else.

Basically, the "source" of my very negative feelings is the intuition (very strong intuition, unfortunately) that their potential instrumental value is not even worth the effort required to fix a mind this broken, even if I had all the required time and resources to actually help each of those cases I encounter and still do whatever other Important Things™ I want/need to do with my life.

That is my true reason. My rationalization is that I have limited resources and time, and so must focus on more cost-effective strategies. Objectively, the rationalization is probably still very very true, and so would make me still choose to not spend all that time and effort helping them, but it is not my original, true reason. It also implies that my behavior is not exactly the same towards them as it would be if that logic were my true chain of reasoning.

All in all, this is one of those things I have as a long-term goal to "fix" once I actually start becoming a half-worthy rationalist, and I consider it an important milestone towards reaching my life goals and becoming a true guardian of my thing to protect. I meant to speak much more at length on this and other personal things once I wrote an intro post in the Welcome topic, but I'm not sure posting there would be appropriate anymore or whether I'll ever actually work myself up to actually write that post.

Edit: Added TLDR at top, because this turned into a fairly long and loaded comment.

I've also often noted, watching certain types of people responding this way to third parties disengaging, that the vigorous condemnation is frequently dropped as soon as the party is no longer in earshot.

Perhaps unrelatedly, I'm told the same thing is often true of small children throwing tantrums.

Fortunately this approach has the side effect that once out of earshot they are condemning you somewhere you don't have to listen to them!

So, so true. I used to think it was the "least bad" / optimal choice until I figured out that it was much more Fun™ to just mess with them (and/or break their mind, if you're so inclined).

I'm Eliezer Yudkowsky! Do you have any idea how many distinct versions of me there are in Tegmark Levels I through III?

Lately I automatically lower my opinion of anyone who self-identifies with a broad enough group without reservation.

I'm a human. (IOW, I agree with what I think you're thinking of, but I don't think “broad enough” is the actual criterion you're using.)

I'm meta-contrarian

Out of curiosity, where do "I'm a humanist" and "I'm a transhumanist" scale?

But yes, outright claiming membership gratuitously for pretty much any wide group without further descriptors or evidence that this affiliation is somehow relevant to the discussion is usually not something to look favorably upon.

I wouldn't quite say it in itself lowers my opinion score of someone, but it might give me some light evidence towards adopting a lower-opinion-estimate model of that someone, which effectively would reduce the "expected opinion for that expected mental model".

Hehe. Once you realize that someone has condemned you guilty a priori, there's all kinds of nifty semi-Dark Arts tricks you can do.

My favorite is to begin agreeing with them more and more anyway, granting them authority and righteousness inch by inch even though it fuels their knowledge that I'm Evil, until I've lured them all the way into a fanatical position that is obviously absurd even to them.

At which point a simple "Yes, you've been right all along!" with a smile is usually all it takes for them to shut up and start agreeing with me instead - their mind is too busy trying to figure out what went wrong to protest, and the autopilot tells them to comply with whatever authority happens to bother telling them anything.

Of course, the effect is temporary, but you usually manage to slip in a few positive beliefs into their subconscious during that window of opportunity.

I'm curious what other LWers think of behaviors like this. I don't trust myself enough yet to ask myself the question (i.e. do a proper crisis of faith), and I fear more rationalization might make me sink into a very dangerous hole if this happens to be a Very Bad™ thing to do. It's something I've been doing (and enjoyed doing) since my early teens, after all. I even have a 'nickname' for it: Shadowdancing.

Yes. Meatspace-only for what I describe in this particular thread.

I've only had three cyberspace interactions with "ID'd-as" female feminists, or so they claimed, and two of these were both trollish and obviously a one-sided preacher throwing regular rage at The Internet with whatever topic they had in mind, while the other was, well, at the time already a much better rationalist than I was, wasn't primarily a "feminist" so much as having that as one of her colors, and is otherwise a subject I'm not quite ready to discuss on LessWrong (a melancholy story of grief and loved ones).

Basically, they're not even valid data points as far as I can tell, for reasons that might not be clear or obvious for the third case but would probably require much more detail than I'm willing to go into to explain why.

Yes, I've unfortunately fallen into that "trap" at least once.

However, the observations persist after modifying the behavior I attempt to output. Either I fail in a somewhat spectacular manner and there's a hard denial-of-denial bomb preventing me from noticing that I'm always acting in such a manner (though I would expect this mechanism to be much more widespread and not restricted specifically to "feminism", which is far from a particularly important point of focus for me among other possible points of focus).

My observations point to a strong causal link between such behavior and the response, but it seems like a sufficient cause, and by far not a required one. The example things I've mentioned (agreeing, disagreeing, nodding, staying silent, going away) are things I've actually tried in separate occasions, as my very first reaction to the topic, if my memory isn't being blurred, and they had the results described. My memory suggests two or three of those might have happened with the same person simply at separate times, but I'm not certain.

Overall, I think the hypothetical reaction you describe might pass a turing test, but I'm throwing that at my own mental emulator, so it's not much of a confirmation. Your mental model seems better detailed than mine, too.

Well, some recent hindsight analysis (during the eridu radical-feminist debacle) allowed me to notice that it seems highly likely that nearly all female feminists I've encountered in person with common knowledge of such were mostly of the kind that had one or few strong very bad near-type personal experiences with men, or many small but memorable such near-type experiences. The kinds you'd probably expect from a stereotypical scenario of "The Father is Master and Law of the House" or a poor waitress working late shifts at a café on the same street corner as a strip club.

So in my case I probably wasn't dealing only with "feminists", but at the same time with individuals taken with a widespread personal fear or anger towards men, in nearly all the cases that produced these kinds of strong reactions. This might be due to statistical coincidence (not that particularly unlikely) or to some behavior that causes other types of feminists to not identify themselves as such when dealing with me, or to some other cause.

It may very well be that the A scenario you describe actually does happen to me sometimes, but with the other participant(s) simply not identifying themselves as feminists at all. If so, I either never ran them through my mental model of feminists for a pattern-matching, reverse-ideological-turing-test thinghy, or my model is sufficiently incorrect/imprecise that they actually failed said test.

In my personal (social) experiences, feminists overall are not as vicious most of the time =)

I kind of suspected this to be the case, because if the contrary were true, the feminist movement as a whole would be spectacularly self-hindering and shooting itself in the foot constantly, since such behavior as I've observed would basically cause very destructive conflict and wouldn't actually help further their goals.

More generally, the idea that taking a potentially damaging action with respect to a vulnerable target is morally distinguishable from taking the exact same action against a well-defended target is relatively uncontroversial even without reference to feminism at all.

Thanks, that puts in context what you were talking about. Radical feminists is just the first thing that was mentally available when I looked for "identifiable ideological social group".