To prohibit generalizations about gender without overwhelming hard data is usually to in effect silence the topic. We are all very interested in gender, and many of us have made interesting and relevant observations about the gender we see around us, but few of us have much in the way of overwhelming hard data. This post seems to be making generalizations about gender aspects of LW posts and comments without itself offering overwhelming hard data—why hold this meta gender discussion to a lower standard?
We already hold discussions of politics to a higher standard—I see the reproof ‘politics is the mind-killer’ relatively often. And this without any particular post arguing that we’re so hideously biased about politics that we need to hold ourselves to a much higher standard than on just about any other topic. And given that the long and very well-documented history of sexism and discrimination against women suggests that enormous masses can be completely wrong for long periods of time (both us and the ancients can’t be right about women), we already have arguments that we specifically are massively biased about gender issues and should hold ourselves to unusually high standards.
Or, if the relevant comments Alicorn cited were about blacks, I don’t think anyone here would even question the need for a higher standard. We all understand intuitively the appeal of racism, its long, hateful, and entrenched history, and that if we’re going to make arguments like blacks are stupider, we’d better have damn good evidence—and merely anecdotal evidence like we see in the cited comments, which boil down to ‘in my experience’ and ‘according to my armchair theorizing’, will cut absolutely no mustard.
Indeed. I, for one, found myself genuinely surprised by the last word of RobinZ’s introduction and had a reaction similar to that of Hofstadter upon finding the answer to the surgeon riddle.
RobinZ illustrates a good point about race-consciousness, though. I was in the Boy Scouts myself, worked at 2 camps, and have seen some demographic data, and the simple fact is: the Boy Scouts are as white as sour cream. It’s not just that there are/were more white Americans than blacks, it’s that whites participate at a vastly higher rate. From a Bayesian perspective, shouldn’t we be surprised to learn that RobinZ is both black and a Boy Scout?
(The Hofstadter example isn’t good for this point; Bayesianly, I think there are many more female surgeons than there are reincarnated-train-wreck-victim-surgeons, so thinking about the latter before the former is just biased and stupid.)
I hate to confess this, but I got stuck on a similar problem, in which the solution was “The secretary is the boy’s father.” (I kept thinking of divorces and such.)
I would question imposing a much higher standard of evidence, e.g. overwhelming hard evidence, for discussions about blacks; that would also basically prohibit discussing such topics.
Gender bias is not some objective feature of writing; it is determined and defined by our perceptions. Men seem extremely likely to perceive it differently from women (see almost every sexual harassment lawsuit). Alicorn has never been super outspoken on this issue and has never been confrontational about it. As one of the most active female readers of this site, if she perceives a gender bias in many posts/comments here that she believes warrants a top-level post, that seems to me to be very strong evidence that something is wrong. That fact combined with the data she did provide seem to me quite convincing that this issue is at least worth thinking about.
Also, if I understand her correctly, her objection is not to making generalizations without overwhelming hard data, it is to making generalizations without the humility appropriate to generalizations not supported by overwhelmingly hard data. If some little study fits a pre-existing belief about ev-psych and how the genders work, posters shouldn’t be super-certain that it’s correct simply because it conforms to their pre-existing beliefs. After all, being human, they may well dismiss a similar study showing the exact opposite effect without second thought.
many of us have made interesting and relevant observations about the gender we see around us,
...observations of a sort which are, in my experience, quite likely tobe extemely inaccurate. Since we’re doing anecdotal observations, I’ve observed that people’s thoughts on gender differences, when not backed up with hard data, are maybe 90% likely to be seriously off base. For whatever reason, these sorts of perceptions seem to be subject to extreme effect from all the usual biases to the extent that I don’t even trust people around here to have reasonable intuitions on the matter.
Also, I’d like to note that the post here included nigh-Yudkowskian levels of cross-linking to other material on LW. When we’re talking about “conversation norms on LW”, how is that not solid data?
Also, I’d like to note that the post here included nigh-Yudkowskian levels of cross-linking to other material on LW. When we’re talking about “conversation norms on LW”, how is that not solid data?
The evidence presented is a number of anecdotes from LW conversation. A fully analysis of LW would need to categorize different types of offending comments, discuss their frequency and what role they play in LW discussion. Even better would be to identify who does them, etc.
Although I do find it plausible that LW should enact a policy of altering present discussions of gender seems I certainly will not say the evidence presented is “overwhelming”.
I said “solid data” in this case, not “overwhelming evidence”. Alicorn was probably being a bit overly strong in phrasing; even a modest amount of evidence would go a long way.
Robin is, of course, being disingenuous here, because even that is more evidence than many of his observations carry.
Seconding SoullessAutomaton’s reply. Also making an analogy to discussions of race.
In addition: “overwhelming hard data” isn’t too high a standard when prejudice and other cultural factors cloud the picture. I don’t think it’s too big a stretch to suppose that such factors are present in this case.
(Finally: a stronger emphasis on hard data in the meta-gender discussion might not be a bad thing, but remaining silent when a problem of this kind exists is a decidedly bad thing.)
To prohibit generalizations about gender without overwhelming hard data is usually to in effect silence the topic.
I think the concern is that a lot of these generalizations aren’t being made through a good-faith attempt to unbiasedly order one’s observations about the world. A lot of people see these arguments and have an (arguably often justified) prior that the individuals who make them are biased and/or bigoted. I realize that it can be frustrating to be told that you’re being criticized because your arguments resemble those made by morally-reprehensible people, but.… it’s often not unjustified for people to come to the table with those assumptions.
You also have the less-defensible argument sometimes being made that we shouldn’t make these theories lightly because they often lead to cryptosexism. That probably won’t fly in a rationalist discussion community, but it does in many other communities where the “social consequences” of one’s speech are supposed to be a serious factor in its moral evaluation.
You also have the less-defensible argument sometimes being made that we shouldn’t make these theories lightly because they often lead to cryptosexism. That probably won’t fly in a rationalist discussion community, but it does in many other communities where the “social consequences” of one’s speech are supposed to be a serious factor in its moral evaluation.
Why is it necessarily more rational to disregard “social consequences”? There’s plenty of objective evidence that calling attention to such issues can in fact be self-fulfilling prophecies, cf. cognitive priming, stereotype threat, &c.
It is of course valuable to be able to discuss ideas freely, but my patience wears thin very quickly when the evidence for such theories is far weaker than the evidence that the theories are harmful.
Why is it necessarily more rational to disregard “social consequences”?
I think it’s not irrational per se, just that it probably wouldn’t fly in this community as a substantive consideration in whether an argument should or should be presented here. Usually it’s considered eminently rude (but not strictly false) to say that the members of your own ingroup are too dumb/biased to discuss a given topic fairly.
I suppose I could also try to bootstrap this into an argument for a strong presumption against restricting speech due to its expected “social consequences” in general, but I think my original points suffice.
Usually it’s considered eminently rude (but not strictly false) to say that the members of your own ingroup are too dumb/biased to discuss a given topic fairly.
Isn’t social acceptance of saying rude but not false things exactly what you’re arguing in favor of?
In general I do carry a strong presumption against restricting speech. But I have a lot of prior experience that, for “gender difference observations not backed by data”, the value of the speech approaches nil in the average case, and is only marginally better on LW, so counterarguments carry a lot more relative weight.
Isn’t social acceptance of saying rude but not false things exactly what you’re arguing in favor of?
I don’t think so, unless you’re implying that the armchair theorizing in this community is always rude. I’d prefer to presume that not to be the case unless there’s evidence otherwise… and I conceded in my top-level reply to this thread that there sometimes is (in my view.)
To add to SoullessAutomaton’s response, the accusation is that the topic is already not being discussed fairly. And I think one could follow the spirit of this post without actually leaving out any ideas, but merely employing some considerate phrasing.
If people are going to make generalizations about the gender they see around them, I’m not going to insist on hard data, but I want to see the soft data. What did you actually see? Was your sample limited to a particular culture? How many instances are you basing your generalization on? Do you notice data which points away from your generalization? How large do you think the effects are?
If men and women have different perceptions of gender bias in writing, which seems overwhelmingly likely, the fact that a woman (who is by no means confrontational or overly outspoken on this topic) feels this issue is worth a top-level post seems like strong evidence that e should worry about it.
To prohibit generalizations about gender without overwhelming hard data is usually to in effect silence the topic. We are all very interested in gender, and many of us have made interesting and relevant observations about the gender we see around us, but few of us have much in the way of overwhelming hard data. This post seems to be making generalizations about gender aspects of LW posts and comments without itself offering overwhelming hard data—why hold this meta gender discussion to a lower standard?
We already hold discussions of politics to a higher standard—I see the reproof ‘politics is the mind-killer’ relatively often. And this without any particular post arguing that we’re so hideously biased about politics that we need to hold ourselves to a much higher standard than on just about any other topic. And given that the long and very well-documented history of sexism and discrimination against women suggests that enormous masses can be completely wrong for long periods of time (both us and the ancients can’t be right about women), we already have arguments that we specifically are massively biased about gender issues and should hold ourselves to unusually high standards.
Or, if the relevant comments Alicorn cited were about blacks, I don’t think anyone here would even question the need for a higher standard. We all understand intuitively the appeal of racism, its long, hateful, and entrenched history, and that if we’re going to make arguments like blacks are stupider, we’d better have damn good evidence—and merely anecdotal evidence like we see in the cited comments, which boil down to ‘in my experience’ and ‘according to my armchair theorizing’, will cut absolutely no mustard.
Indeed. I, for one, found myself genuinely surprised by the last word of RobinZ’s introduction and had a reaction similar to that of Hofstadter upon finding the answer to the surgeon riddle.
Hah! I almost didn’t include that word—now I’m glad I did.
RobinZ illustrates a good point about race-consciousness, though. I was in the Boy Scouts myself, worked at 2 camps, and have seen some demographic data, and the simple fact is: the Boy Scouts are as white as sour cream. It’s not just that there are/were more white Americans than blacks, it’s that whites participate at a vastly higher rate. From a Bayesian perspective, shouldn’t we be surprised to learn that RobinZ is both black and a Boy Scout?
(The Hofstadter example isn’t good for this point; Bayesianly, I think there are many more female surgeons than there are reincarnated-train-wreck-victim-surgeons, so thinking about the latter before the former is just biased and stupid.)
I hate to confess this, but I got stuck on a similar problem, in which the solution was “The secretary is the boy’s father.” (I kept thinking of divorces and such.)
So yeah.
I would question imposing a much higher standard of evidence, e.g. overwhelming hard evidence, for discussions about blacks; that would also basically prohibit discussing such topics.
But arguments that aren’t merely about, but which run down the well-worn grooves of racist quack science, those would need overwhelming hard evidence.
Gender bias is not some objective feature of writing; it is determined and defined by our perceptions. Men seem extremely likely to perceive it differently from women (see almost every sexual harassment lawsuit). Alicorn has never been super outspoken on this issue and has never been confrontational about it. As one of the most active female readers of this site, if she perceives a gender bias in many posts/comments here that she believes warrants a top-level post, that seems to me to be very strong evidence that something is wrong. That fact combined with the data she did provide seem to me quite convincing that this issue is at least worth thinking about.
Also, if I understand her correctly, her objection is not to making generalizations without overwhelming hard data, it is to making generalizations without the humility appropriate to generalizations not supported by overwhelmingly hard data. If some little study fits a pre-existing belief about ev-psych and how the genders work, posters shouldn’t be super-certain that it’s correct simply because it conforms to their pre-existing beliefs. After all, being human, they may well dismiss a similar study showing the exact opposite effect without second thought.
...observations of a sort which are, in my experience, quite likely to be extemely inaccurate. Since we’re doing anecdotal observations, I’ve observed that people’s thoughts on gender differences, when not backed up with hard data, are maybe 90% likely to be seriously off base. For whatever reason, these sorts of perceptions seem to be subject to extreme effect from all the usual biases to the extent that I don’t even trust people around here to have reasonable intuitions on the matter.
Also, I’d like to note that the post here included nigh-Yudkowskian levels of cross-linking to other material on LW. When we’re talking about “conversation norms on LW”, how is that not solid data?
The evidence presented is a number of anecdotes from LW conversation. A fully analysis of LW would need to categorize different types of offending comments, discuss their frequency and what role they play in LW discussion. Even better would be to identify who does them, etc.
Although I do find it plausible that LW should enact a policy of altering present discussions of gender seems I certainly will not say the evidence presented is “overwhelming”.
I said “solid data” in this case, not “overwhelming evidence”. Alicorn was probably being a bit overly strong in phrasing; even a modest amount of evidence would go a long way.
Robin is, of course, being disingenuous here, because even that is more evidence than many of his observations carry.
Seconding SoullessAutomaton’s reply. Also making an analogy to discussions of race.
In addition: “overwhelming hard data” isn’t too high a standard when prejudice and other cultural factors cloud the picture. I don’t think it’s too big a stretch to suppose that such factors are present in this case.
(Finally: a stronger emphasis on hard data in the meta-gender discussion might not be a bad thing, but remaining silent when a problem of this kind exists is a decidedly bad thing.)
To prohibit generalizations about gender without overwhelming hard data is usually to in effect silence the topic.
I think the concern is that a lot of these generalizations aren’t being made through a good-faith attempt to unbiasedly order one’s observations about the world. A lot of people see these arguments and have an (arguably often justified) prior that the individuals who make them are biased and/or bigoted. I realize that it can be frustrating to be told that you’re being criticized because your arguments resemble those made by morally-reprehensible people, but.… it’s often not unjustified for people to come to the table with those assumptions.
You also have the less-defensible argument sometimes being made that we shouldn’t make these theories lightly because they often lead to cryptosexism. That probably won’t fly in a rationalist discussion community, but it does in many other communities where the “social consequences” of one’s speech are supposed to be a serious factor in its moral evaluation.
Why is it necessarily more rational to disregard “social consequences”? There’s plenty of objective evidence that calling attention to such issues can in fact be self-fulfilling prophecies, cf. cognitive priming, stereotype threat, &c.
It is of course valuable to be able to discuss ideas freely, but my patience wears thin very quickly when the evidence for such theories is far weaker than the evidence that the theories are harmful.
I think it’s not irrational per se, just that it probably wouldn’t fly in this community as a substantive consideration in whether an argument should or should be presented here. Usually it’s considered eminently rude (but not strictly false) to say that the members of your own ingroup are too dumb/biased to discuss a given topic fairly.
I suppose I could also try to bootstrap this into an argument for a strong presumption against restricting speech due to its expected “social consequences” in general, but I think my original points suffice.
Isn’t social acceptance of saying rude but not false things exactly what you’re arguing in favor of?
In general I do carry a strong presumption against restricting speech. But I have a lot of prior experience that, for “gender difference observations not backed by data”, the value of the speech approaches nil in the average case, and is only marginally better on LW, so counterarguments carry a lot more relative weight.
I don’t think so, unless you’re implying that the armchair theorizing in this community is always rude. I’d prefer to presume that not to be the case unless there’s evidence otherwise… and I conceded in my top-level reply to this thread that there sometimes is (in my view.)
Well, all else equal, speculations about other people based on their intrinsic demographics seems fairly rude to me; your mileage may vary.
I am, as I said, willing to accept rudeness when stating facts, though tact is always appreciated.
To add to SoullessAutomaton’s response, the accusation is that the topic is already not being discussed fairly. And I think one could follow the spirit of this post without actually leaving out any ideas, but merely employing some considerate phrasing.
To start with, generalizations about what sort of gender? Chromosomal? Developmental? Self-identified? Cultural by nurture? Cultural by presentation?
Mostly we only see the latter and assume all the rest.
If people are going to make generalizations about the gender they see around them, I’m not going to insist on hard data, but I want to see the soft data. What did you actually see? Was your sample limited to a particular culture? How many instances are you basing your generalization on? Do you notice data which points away from your generalization? How large do you think the effects are?
If men and women have different perceptions of gender bias in writing, which seems overwhelmingly likely, the fact that a woman (who is by no means confrontational or overly outspoken on this topic) feels this issue is worth a top-level post seems like strong evidence that e should worry about it.