Heh, when arguing for the case that people should be careful with their wording, I'm challenged for a careless choice of wording. :-)

Innocuous in the sense of emerging from different-gendered people on average having different preferences and on average making different choices as a result. Me eating french fries every day, because I want to, is an innocous reason for eating french fries every day (though such behavior will probably cause health problems in the long term). Eating french fries every day because somebody pressures me into doing so, or because I genuinely can't afford anything else, is a non-innocous reason.

OK, I think I finally understand.

What was said was:

E.g. I've often heard it claimed that the difference in average pay between women and men is mostly attributable to differences in ambition and time voluntarily spent at home with children. I haven't looked at the matter enough to know if this is true. But if it is, then denying any population-level differences between men and women seems harmful, because it implies that something that actually has an innocuous explanation is because of discrimination.

One common explanation of harm and utilities is that the "real" or important utility function held by a human is that implied by the humans actions. If a human chooses A over B, that means to the human A has a higher value than B to the human. This runs us into problems, for example when humans choose B over C and C over A, but there is no agreed upon way to discuss the relationship of humans to utility functions. We just don't know how to extract the human and cut the nonsense without cutting the human! This is despite extensively discussing extrapolate volition. One way to get people to actually choose consistently among A, B, and C is to teach them about this paradox, but let's just say for our purposes here that it's clearly not out of line to discuss people's "true" preferences being something other than what they choose.

Vaniver: Ok: let's suppose he intended the primary definition of innocuous, "not harmful." If a choice is made voluntarily, then by the assumption of revealed preferences it is the least 'harmful.' If we forced women to choose with the same distribution that men do, then on net women would be worse off- i.e. harmed by our force.

Clarica: I think that calling the choice to spend more or less time doing financially unrecompensed work in the home an innocuous gender difference, is careless. The harms of the various choices have not been evaluated that well.

One issue is that language is flexible, and it is common to see "innocuous explanation" as a way of discussing the motives of a person causing the things the explanation explains, rather than according to the usual adjective-noun relationship where the adjective modifies the noun.

For example: a video teaching "how to fold a shirt" with the audio 50 decibels is a harmless explanation. The same video with the audio at 125 decibels is a harmful explanation.

No one argues that the explanation itself would have only good consequences, the discussion is instead what sort of harmlessness is meant instead. Whether the author's intent is clearly that, if it is discovered that women's actions alone cause the statistical difference, i) employers are doing no harm in the hypothetical case, or ii) if a similarly plausible interpretation is that no one is suffering harm, for had they chosen as men, there would be no disparity.

Context points to the first explanation as the best contrast with "discrimination", what employers are allegedly doing, and what hypothetical evidence would clear them of, but it's easy to see why someone intending the second point might have used the same words.

The sentence might be rewritten: "But if it is, then denying any population-level differences between men and women seems harmful, because it implies that something fully explained by innocuous behavior is because of discrimination."

The principle of charity protects us in similar cases where we happen to only see one interpretation and it is the wrong one.

Someone (Eliezer?) once said something like: if you tell me exactly what it is that an Artificial Intelligence can't do, I can build an AI to do exactly that. If a person who believes in a fundamental difference at that sort of level between machines and animals can precisely define something, a computer can follow that definition.

It doesn't work quite as well here. But if someone gives a good enough answer for their question of "who", with exactly why an animal wouldn't count, or a computer, or a corporation, they may make their question so complicated that it only has one answer or no answers as asked.

Point the second - Hypothetically, if this:

the difference in average pay between women and men is mostly attributable to differences in ambition and time voluntarily spent at home with children.

is true, then gender pay inequities do have an innocuous explanation- namely, the above. Kaj_Sotala made no claims beyond that, certainly not to the extent of claiming the above statement is true in the real world.

Hypothetically, I agree with you.

I think this might be confusing pedanterrific because if I read you right above you don't agree with him. I thought your position was similar to the one I made here that that explanation of pay inequality, even if true, is not innocuous because the reason why men and women make different choices about work and home life could be harmful social pressure, or some other reason that we don't think people should have to face in an ideal world. But I could have misread you when you wrote this:

I don't think I'd use the word innocuous with the example of this reason for this gender difference. If it is a rational choice, why don't both genders make similar choices?

Do you agree?

Yes, of course. That's trivially true and not in dispute.

I still think you're rather missing the point, however. I don't see how it makes sense to object to the phrase 'mostly attributable' when that's a premise of the hypothetical. Let's look at the original comment in context:

E.g. I've often heard it claimed that the difference in average pay between women and men is mostly attributable to differences in ambition and time voluntarily spent at home with children. I haven't looked at the matter enough to know if this is true. But if it is, then denying any population-level differences between men and women seems harmful, because it implies that something that actually has an innocuous explanation is because of discrimination.

That is, IF [the difference is mostly attributable to something innocuous], THEN [denying population-level differences seems harmful]. That's all that was said. Kaj_Sotala never claimed the innocuous explanation was true.

Editeditedit: I apologize for my horrible social skills.

...Oh. I think my sincerity detectors might be broken.

Do you agree?

Yes, of course. That's trivially true and not in dispute.

I still think you're rather missing the point, however. I don't see how it makes sense to object to the phrase 'mostly attributable' when that's a premise of the hypothetical. Let's look at the original comment in context:

E.g. I've often heard it claimed that the difference in average pay between women and men is mostly attributable to differences in ambition and time voluntarily spent at home with children. I haven't looked at the matter enough to know if this is true. But if it is, then denying any population-level differences between men and women seems harmful, because it implies that something that actually has an innocuous explanation is because of discrimination.

That is, IF [the difference is mostly attributable to something innocuous], THEN [denying population-level differences seems harmful]. That's all that was said. Kaj_Sotala never claimed the innocuous explanation was true.

Editeditedit: I apologize for my horrible social skills.

When asked "who" would do something, asking for a definition of who is an interesting move.

I do not really understand your questions. <--serious

Can you define 'who' <--funny

'them' 'whose' and 'it'? <--serious

Would, compensate, benefit, is, and for I get. <--funny

Point the first - Now I'm confused. Is it that

This question:

Who would compensate them? Whose benefit is it for?

is an excellent question that I actually do not want to answer,

or is it that "I do not really understand your questions."? Or did my explanation allow you to understand that you didn't want to answer, or...

Point the second - Hypothetically, if this:

the difference in average pay between women and men is mostly attributable to differences in ambition and time voluntarily spent at home with children.

is true, then gender pay inequities do have an innocuous explanation- namely, the above. Kaj_Sotala made no claims beyond that, certainly not to the extent of claiming the above statement is true in the real world.

This leads me to believe your point is not valid or valuable criticism. If you think I'm wrong, could you explain why?