I take your point that skews can happen, but it seems a bit suspicious to me that desire to be effective and altruistic should be so heavily skewed towards white dudes.
Edit: I previous said “straight white dudes” but removed the “straight”. See below.
This reminds me of a pattern I see in social justice movements, which goes something like this: We are observing some kind of gender or race-based disparity, with a variety of different hypotheses for why it might be occurring. Some people think discrimination is the most likely hypothesis. Other people have other hypotheses. The people who think discrimination is the most likely hypotheses see the people suggesting other hypotheses and loudly decry those people as discriminatory. Those people get quieter. The gender or race-based disparity persists. The only hypothesis that anyone is allowed to talk about is the discrimination one. So it’s more clear than ever that discrimination is the only possible explanation. Given this clarity, the people pushing the discrimination hypothesis have the mandate to decry milder and milder instances of discrimination. Eventually, the community undergoes a schism over the issue of whether to be hypersensitive to mild instances of discrimination or not.
The Google memo Kelly references is a good case study. Kelly implies that the author is an “outright asshole”. I assume she makes this judgement solely based on the author’s willingness to explore hypotheses besides the discrimination one—in terms of communication style, it’s clear that the author takes pains to be as civil as possible.
The question for me is: How long do we need to test out the discrimination hypothesis before it’s disconfirmed? If it’s been 5 years since anyone talked about any hypothesis besides the discrimination one, and the disparity still persists, are we allowed to consider the possibility that the discrimination hypothesis is incorrect? What if it’s been 10 years?
Realistically, if there’s a disparity, there’s probably a combination of several things going on. So, how can we capture the low-hanging fruit from fighting discrimination without putting ourselves on a path towards a schism?
The histories of many forms of prejudice are histories biological essentialism and biological determinism. Even if such claims are now made out of a “willingness to explore” alternative hypotheses despite this long history of precisely being an unwillingness to explore the much newer hypothesis of prejudice, they tend to be over-simplistic, as in the memo, and tend to have the effect—if not also the intention—of dismissing the other, newer hypothesis of prejudice, which is robustly supported by data that the memo’s author fails to include.
That’s not to say it’s a black and white matter of total biological similarity or total culturally-imposed disparities and prejudice. That’s what the author of the memo implies, and I disagree. The evidence that prejudice is a major problem that is holding people back is substantial nonetheless.
Some of his suggestions for ways to reduce the gender gap are worth considering, and charitably he’s not exceptionally prejudiced and is able to analyze information that has found its way to him, but is just very poorly informed and has no willingness to explore the alternative explanation of prejudice. At most charitable this still enables that prejudice.
Given the extent of my knowledge, which is just the words in the memo, I can agree he’s not an outright asshole, and I should have phrased my side note about this example of zero tolerance with a heavy hand differently. It may even be a poor example, as I would say corrective action should have been taken in his case before he was fired if it wasn’t, which I don’t know about either way.
Thanks for the reply, Kelly, and I’m sorry you’re getting downvoted. I really appreciate your willingness to be charitable and admit your mistakes, and I will strive to emulate your example.
That’s not to say it’s a black and white matter of total biological similarity or total culturally-imposed disparities and prejudice. That’s what the author of the memo implies
Hm, that’s not how I read it. For example, in the first sentence, he says he doesn’t deny that sexism exists. Later, he writes: “Of course, men and women experience bias, tech, and the workplace differently and we should be cognizant of this...” My interpretation is that Google already has a ton of discussion of the impact of sexism, bias, etc. and Damore wanted to fill in the other side of the story, so he didn’t bother to repeat stuff that everyone already agrees on. Maybe that was a mistake in retrospect.
I agree that that qualification suggests his view on the contribution of biology to the gender gap is weaker than his otherwise definitive framings suggest. [Edit: Sentence here removed because I’m too tired and my thoughts are not in order, will get sleep before responding to any more comments. Replacement: He’s still presenting it as a black-and-white issue if he’s only presenting one side.]
Google may have had that conversation on prejudice going, but he is very oversimplistic and offers the essentialist view as so definitive that his solutions are the right ones, that Google is the “biased” party for talking about prejudice, and that it isn’t worth even mentioning that evidence demonstrating a bias against women exists (if he even knows or believes that), not to mention that the evidence for the real-world effect of prejudice is far more vast and robust than his evidence for biological causes. And he does all this when the essentialist view has been so dominant and people are only talking so much about prejudice because they’re trying to overcome the essentialist thinking that so inhibits people. (Sure, there are differences, but there are even more misconceptions, as well as oversimplistic and deterministic assumptions about what real differences mean.)
[Edit for clarification and additional analysis: In a context of prejudice, presenting stereotypes is a delicate matter even if you think them sufficiently biologically valid and are content to make simplistic inferences about their real-world effects. Doing so without acknowledgement of the prejudices people experience which line up with these stereotypes and which harm them serves to reinforce those stereotypes and prejudices.]
So it’s not an appropriate way to contribute to the conversation—at best it’s reacting to perceived overshooting by retreating to a flawed status quo.
Can I suggest that the Damore issue be parked? Even though it is currently producing a high quality, civil conversation, I worry that talking about such a highly polarised topic is somewhat risky as you never know who might join the thread.
I think there’s a bit of an empathy gap in this community. When people are angry for what seems to be no reason, a good first step is to ask whether you’ve done something that made them feel unsafe/humiliated/demeaned/etc, even if that wasn’t your intention. It doesn’t take a lot of imagination to see how unsolicited exploration of “other hypotheses” (cough cough) for racial and gender disparities could be very distressing for the people who are being discussed as if they’re not there.
I actually think we should discuss other hypotheses.
Firstly, “other hypotheses” includes all kinds of inoffensive explanations like the primary cause of a difference being:
Broader society has instilled certain social norms in people, as opposed to it being anything specific about this group
Founder effects—A guy gets a few of his mates to start the group, they rope in their mates, ect.
That the message happens to resonate among groups of people that are currently disproportionately one gender (ie. programmers)
But going further than this, I don’t think we should limit discussion of different intrinsic preferences either, especially if someone makes an argument that is dependent on this being false.
Where do we draw the line? Is intrinsic abilities an acceptable topic of casual discussion? Do you think it would be humiliating for people who are being discussed as having less intrinsic ability?
I think it depends on the particular space. The rationality community should aim to have everything open to discussion because that is its purpose. The EA community should minimise these discussions in that they are rarely necessarily and quite often a distraction. In most groups I’ve been in, social norms can prevent the need for formal rules though.
It doesn’t take a lot of imagination to see how unsolicited exploration of “other hypotheses” (cough cough) for racial and gender disparities could be very distressing for the people who are being discussed as if they’re not there.
Oh, I totally agree, and I don’t think we should explore them. [I edited my comment in an attempt to clarify this.]
But you don’t want discrimination hypotheses to be discussed either? I guess that could be an acceptable compromise, to not debate the causes of disparities but at the same time focus on improving diversity in recruitment.
Yeah. I’m also in favor of trying to grab low-hanging fruit from addressing discrimination, as long as we don’t get overzealous. But in terms of trying to make our demographics completely representative… there are already a lot of groups trying and failing to do that, sometimes in a way that crashes & burns spectacularly, so I would rather hang back and wait for a model that seems workable/reliable before aiming that high.
FWIW, I’m sympathetic to the google guy. However, it’s not clear to me this case in the same. It might be, but I’d want someone to give me a series of reasons, backed by evidence, before we conclude “oh, it turns out affluent white males are just a lot more moral than everyone else and there’s nothing to explain here”.
“oh, it turns out affluent white males are just a lot more moral than everyone else and there’s nothing to explain here”
Do you think it is possible that EA could be majority white affluent male because programmers, philosophers, mathematicians, ect. are disproportionately white affluent male and EA has become good at recruiting these specific audiences?
I think that’s a huge part of the reason why we overrepresent people the demographics we do. But offloading responsibility onto part of the pipeline below us isn’t sufficient, least of all when we can source from other pipelines.
I don’t actually believe that affluent white males are a lot more moral than everyone else, but anyway, let’s put aside the question of whether such evidence exists for a moment and ask: if such evidence did exist, would it be sensible for us to discuss it? My answer is no. I would rather take a compromise position of addressing clear cases of discrimination, being mildly worried about mild cases, and letting sleeping dogs lie.
The difficulty for movements against discrimination (between humans) in a lot of modern society lies in that definition of what constitutes “clear” discrimination. For instance, people don’t say explicitly discriminatory things as much as they used to, but they still hold discriminatory beliefs that make them e.g. mistrust, discredit and undervalue others, and we can for the most part only assess e.g. hiring bias by looking at whole samples, not at any one individual.
I don’t think we should police thoughts, only actions.
We don’t make it a crime to fantasize about killing someone—you only become a criminal when you act on those thoughts. This illustrates a useful and widely applied principle of our legal system. The willingness of some diversity advocates to disregard this principle is a good example of diversity advocates getting overzealous about diversity and sacrificing other values, as I complain about in this comment.
Furthermore, I don’t think condemning people for having beliefs we don’t want is an effective way to change those beliefs—a varietyofresearch seems to indicate this doesn’t work (though, I generally don’t put too much stock in social psychology research, which includes those links, and I’m also not a good paper scrutinizer).
The problem is that those thoughts, as I noted, become actions, just actions we can usually only see as systematic trends. Just because someone does not say “women are incompetent” does not mean they aren’t underestimating women’s competence and e.g. hiring them less than he should. Taking action on this just requires a more systematic approach than explicit discrimination does.
I agree that in terms of what works, just pointing out bias doesn’t seem to help and can even backfire, as I mentioned, which is why I provided a list of other possible solutions.
The problem is that those thoughts… become actions… we can usually only see as systematic trends. Just because someone does not say “women are incompetent” does not mean they aren’t underestimating women’s competence and e.g. hiring them less than he should.
The flip side of it being hard to discern whether people have bad thoughts and act biasedly except by drawing inferences from broader patterns is that it’s also hard to discern whether people actually do have bad thoughts and acted biasedly from those broader patterns. (c.f. the many fields where women dominate men in terms of prevalence and performance, as well as EAs many other demographic biases which don’t receive the same treatment e.g. a 14:1 left-right bias, and a 4:1 20-35:any age over 35 bias).
“I assume she makes this judgement solely based on the author’s willingness to explore hypotheses besides the discrimination one”—This seems like a very uncharitable assumption to make. I can easily think of multiple other reasons why she might consider him an asshole.
I think there are some varied skews here. It seems that we do well on representation of trans people generally and queer women relative to the total number of women, but not on queer men relative to the number of men. I think there are probably more political queer men (rejection of gay/straight binary sort of thing) than in most communities, but not many men who regularly sleep or seek to sleep with men. I know I was in the community for years before meeting one.
So yes, I think the skew is toward straight, white dudes, and I’ll say I do find the machismo off-putting even as a fairly straight-passing gay man.
Hmm. Maybe EA is more inclusively representated on the sexual dimension. I’d hadn’t really noticed this either way, and more typed this out of habit. I stick by there being an oddly high number of white dudes though.
If you aren’t going to defend the claim made in the original comment, I would suggest that it would be good practise to edit the word “straight” out of the comment. There are a lot of Cached Thoughts on both sides of the debate and I would like to encourage people to break out of them.
“I take your point that skews can happen, but it seems a bit suspicious to me that desire to be effective and altruistic should be so heavily skewed towards straight, white dudes.”
(1) Where did “straight” come into this picture? The author says that EAs are well-represented on sexual diversity (and maybe even overrepresented on some fairly atypical sexual orientations), and my comment (and the data I used) had nothing to say about sexual orientation?
(2) “”“it seems a bit suspicious to me that desire to be effective and altruistic should be so heavily skewed towards straight, white dudes”””
I didn’t say that desire to be effective and altruistic is heavily skewed toward men. I just said that membership in a specific community, or readership of a specific website, and things like that, can have significant gender skews, and that is not atypical. The audience for a specific community, like the effective altruist community, can be far smaller than the set of people with desire to be effective and altruistic.
For instance, if a fashion website has a 90% female audience (a not atypical number), that is not a claim that the “desire to look good” is that heavily skewed toward female. It means that the specific things that website caters to, the way it has marketed itself, etc. have resulted in it getting a female audience. Men could also desire to look good, albeit in ways that are very different from those catered to by that fashion website (or more broadly by the majority of present-day fashion websites).
Your suspicions provide a small amount of Bayesian evidence, but could you explain why you believe none of the alternate explanations that have been proposed seem satisfactory?
I take your point that skews can happen, but it seems a bit suspicious to me that desire to be effective and altruistic should be so heavily skewed towards white dudes.
Edit: I previous said “straight white dudes” but removed the “straight”. See below.
This reminds me of a pattern I see in social justice movements, which goes something like this: We are observing some kind of gender or race-based disparity, with a variety of different hypotheses for why it might be occurring. Some people think discrimination is the most likely hypothesis. Other people have other hypotheses. The people who think discrimination is the most likely hypotheses see the people suggesting other hypotheses and loudly decry those people as discriminatory. Those people get quieter. The gender or race-based disparity persists. The only hypothesis that anyone is allowed to talk about is the discrimination one. So it’s more clear than ever that discrimination is the only possible explanation. Given this clarity, the people pushing the discrimination hypothesis have the mandate to decry milder and milder instances of discrimination. Eventually, the community undergoes a schism over the issue of whether to be hypersensitive to mild instances of discrimination or not.
The Google memo Kelly references is a good case study. Kelly implies that the author is an “outright asshole”. I assume she makes this judgement solely based on the author’s willingness to explore hypotheses besides the discrimination one—in terms of communication style, it’s clear that the author takes pains to be as civil as possible.
The question for me is: How long do we need to test out the discrimination hypothesis before it’s disconfirmed? If it’s been 5 years since anyone talked about any hypothesis besides the discrimination one, and the disparity still persists, are we allowed to consider the possibility that the discrimination hypothesis is incorrect? What if it’s been 10 years?
Realistically, if there’s a disparity, there’s probably a combination of several things going on. So, how can we capture the low-hanging fruit from fighting discrimination without putting ourselves on a path towards a schism?
The histories of many forms of prejudice are histories biological essentialism and biological determinism. Even if such claims are now made out of a “willingness to explore” alternative hypotheses despite this long history of precisely being an unwillingness to explore the much newer hypothesis of prejudice, they tend to be over-simplistic, as in the memo, and tend to have the effect—if not also the intention—of dismissing the other, newer hypothesis of prejudice, which is robustly supported by data that the memo’s author fails to include.
That’s not to say it’s a black and white matter of total biological similarity or total culturally-imposed disparities and prejudice. That’s what the author of the memo implies, and I disagree. The evidence that prejudice is a major problem that is holding people back is substantial nonetheless.
Some of his suggestions for ways to reduce the gender gap are worth considering, and charitably he’s not exceptionally prejudiced and is able to analyze information that has found its way to him, but is just very poorly informed and has no willingness to explore the alternative explanation of prejudice. At most charitable this still enables that prejudice.
Given the extent of my knowledge, which is just the words in the memo, I can agree he’s not an outright asshole, and I should have phrased my side note about this example of zero tolerance with a heavy hand differently. It may even be a poor example, as I would say corrective action should have been taken in his case before he was fired if it wasn’t, which I don’t know about either way.
Thanks for the reply, Kelly, and I’m sorry you’re getting downvoted. I really appreciate your willingness to be charitable and admit your mistakes, and I will strive to emulate your example.
Hm, that’s not how I read it. For example, in the first sentence, he says he doesn’t deny that sexism exists. Later, he writes: “Of course, men and women experience bias, tech, and the workplace differently and we should be cognizant of this...” My interpretation is that Google already has a ton of discussion of the impact of sexism, bias, etc. and Damore wanted to fill in the other side of the story, so he didn’t bother to repeat stuff that everyone already agrees on. Maybe that was a mistake in retrospect.
I agree that that qualification suggests his view on the contribution of biology to the gender gap is weaker than his otherwise definitive framings suggest. [Edit: Sentence here removed because I’m too tired and my thoughts are not in order, will get sleep before responding to any more comments. Replacement: He’s still presenting it as a black-and-white issue if he’s only presenting one side.]
Google may have had that conversation on prejudice going, but he is very oversimplistic and offers the essentialist view as so definitive that his solutions are the right ones, that Google is the “biased” party for talking about prejudice, and that it isn’t worth even mentioning that evidence demonstrating a bias against women exists (if he even knows or believes that), not to mention that the evidence for the real-world effect of prejudice is far more vast and robust than his evidence for biological causes. And he does all this when the essentialist view has been so dominant and people are only talking so much about prejudice because they’re trying to overcome the essentialist thinking that so inhibits people. (Sure, there are differences, but there are even more misconceptions, as well as oversimplistic and deterministic assumptions about what real differences mean.)
[Edit for clarification and additional analysis: In a context of prejudice, presenting stereotypes is a delicate matter even if you think them sufficiently biologically valid and are content to make simplistic inferences about their real-world effects. Doing so without acknowledgement of the prejudices people experience which line up with these stereotypes and which harm them serves to reinforce those stereotypes and prejudices.]
So it’s not an appropriate way to contribute to the conversation—at best it’s reacting to perceived overshooting by retreating to a flawed status quo.
Can I suggest that the Damore issue be parked? Even though it is currently producing a high quality, civil conversation, I worry that talking about such a highly polarised topic is somewhat risky as you never know who might join the thread.
I think there’s a bit of an empathy gap in this community. When people are angry for what seems to be no reason, a good first step is to ask whether you’ve done something that made them feel unsafe/humiliated/demeaned/etc, even if that wasn’t your intention. It doesn’t take a lot of imagination to see how unsolicited exploration of “other hypotheses” (cough cough) for racial and gender disparities could be very distressing for the people who are being discussed as if they’re not there.
I actually think we should discuss other hypotheses.
Firstly, “other hypotheses” includes all kinds of inoffensive explanations like the primary cause of a difference being:
Broader society has instilled certain social norms in people, as opposed to it being anything specific about this group
Founder effects—A guy gets a few of his mates to start the group, they rope in their mates, ect.
That the message happens to resonate among groups of people that are currently disproportionately one gender (ie. programmers)
But going further than this, I don’t think we should limit discussion of different intrinsic preferences either, especially if someone makes an argument that is dependent on this being false.
I think I’ve noticed a pattern where basically any hypothesis that’s not the discrimination hypothesis gradually leaves the Overton window.
Where do we draw the line? Is intrinsic abilities an acceptable topic of casual discussion? Do you think it would be humiliating for people who are being discussed as having less intrinsic ability?
I think it depends on the particular space. The rationality community should aim to have everything open to discussion because that is its purpose. The EA community should minimise these discussions in that they are rarely necessarily and quite often a distraction. In most groups I’ve been in, social norms can prevent the need for formal rules though.
Oh, I totally agree, and I don’t think we should explore them. [I edited my comment in an attempt to clarify this.]
But you don’t want discrimination hypotheses to be discussed either? I guess that could be an acceptable compromise, to not debate the causes of disparities but at the same time focus on improving diversity in recruitment.
Yeah. I’m also in favor of trying to grab low-hanging fruit from addressing discrimination, as long as we don’t get overzealous. But in terms of trying to make our demographics completely representative… there are already a lot of groups trying and failing to do that, sometimes in a way that crashes & burns spectacularly, so I would rather hang back and wait for a model that seems workable/reliable before aiming that high.
FWIW, I’m sympathetic to the google guy. However, it’s not clear to me this case in the same. It might be, but I’d want someone to give me a series of reasons, backed by evidence, before we conclude “oh, it turns out affluent white males are just a lot more moral than everyone else and there’s nothing to explain here”.
“oh, it turns out affluent white males are just a lot more moral than everyone else and there’s nothing to explain here”
Do you think it is possible that EA could be majority white affluent male because programmers, philosophers, mathematicians, ect. are disproportionately white affluent male and EA has become good at recruiting these specific audiences?
I think that’s a huge part of the reason why we overrepresent people the demographics we do. But offloading responsibility onto part of the pipeline below us isn’t sufficient, least of all when we can source from other pipelines.
Interesting. Hadn’t put these together in my mind. Could well be something here.
I don’t actually believe that affluent white males are a lot more moral than everyone else, but anyway, let’s put aside the question of whether such evidence exists for a moment and ask: if such evidence did exist, would it be sensible for us to discuss it? My answer is no. I would rather take a compromise position of addressing clear cases of discrimination, being mildly worried about mild cases, and letting sleeping dogs lie.
The difficulty for movements against discrimination (between humans) in a lot of modern society lies in that definition of what constitutes “clear” discrimination. For instance, people don’t say explicitly discriminatory things as much as they used to, but they still hold discriminatory beliefs that make them e.g. mistrust, discredit and undervalue others, and we can for the most part only assess e.g. hiring bias by looking at whole samples, not at any one individual.
I don’t think we should police thoughts, only actions.
We don’t make it a crime to fantasize about killing someone—you only become a criminal when you act on those thoughts. This illustrates a useful and widely applied principle of our legal system. The willingness of some diversity advocates to disregard this principle is a good example of diversity advocates getting overzealous about diversity and sacrificing other values, as I complain about in this comment.
Furthermore, I don’t think condemning people for having beliefs we don’t want is an effective way to change those beliefs—a variety of research seems to indicate this doesn’t work (though, I generally don’t put too much stock in social psychology research, which includes those links, and I’m also not a good paper scrutinizer).
The problem is that those thoughts, as I noted, become actions, just actions we can usually only see as systematic trends. Just because someone does not say “women are incompetent” does not mean they aren’t underestimating women’s competence and e.g. hiring them less than he should. Taking action on this just requires a more systematic approach than explicit discrimination does.
I agree that in terms of what works, just pointing out bias doesn’t seem to help and can even backfire, as I mentioned, which is why I provided a list of other possible solutions.
The flip side of it being hard to discern whether people have bad thoughts and act biasedly except by drawing inferences from broader patterns is that it’s also hard to discern whether people actually do have bad thoughts and acted biasedly from those broader patterns. (c.f. the many fields where women dominate men in terms of prevalence and performance, as well as EAs many other demographic biases which don’t receive the same treatment e.g. a 14:1 left-right bias, and a 4:1 20-35:any age over 35 bias).
“I assume she makes this judgement solely based on the author’s willingness to explore hypotheses besides the discrimination one”—This seems like a very uncharitable assumption to make. I can easily think of multiple other reasons why she might consider him an asshole.
What evidence causes you to think that heterosexuality is overrepresented in EA? That seems backwards to me.
I believe sexuality is a demographic we do well on.
We also have way more trans women than society at large.
I think there are some varied skews here. It seems that we do well on representation of trans people generally and queer women relative to the total number of women, but not on queer men relative to the number of men. I think there are probably more political queer men (rejection of gay/straight binary sort of thing) than in most communities, but not many men who regularly sleep or seek to sleep with men. I know I was in the community for years before meeting one.
So yes, I think the skew is toward straight, white dudes, and I’ll say I do find the machismo off-putting even as a fairly straight-passing gay man.
Hmm. Maybe EA is more inclusively representated on the sexual dimension. I’d hadn’t really noticed this either way, and more typed this out of habit. I stick by there being an oddly high number of white dudes though.
If you aren’t going to defend the claim made in the original comment, I would suggest that it would be good practise to edit the word “straight” out of the comment. There are a lot of Cached Thoughts on both sides of the debate and I would like to encourage people to break out of them.
okay. point taken
“I take your point that skews can happen, but it seems a bit suspicious to me that desire to be effective and altruistic should be so heavily skewed towards straight, white dudes.”
(1) Where did “straight” come into this picture? The author says that EAs are well-represented on sexual diversity (and maybe even overrepresented on some fairly atypical sexual orientations), and my comment (and the data I used) had nothing to say about sexual orientation?
(2) “”“it seems a bit suspicious to me that desire to be effective and altruistic should be so heavily skewed towards straight, white dudes”””
I didn’t say that desire to be effective and altruistic is heavily skewed toward men. I just said that membership in a specific community, or readership of a specific website, and things like that, can have significant gender skews, and that is not atypical. The audience for a specific community, like the effective altruist community, can be far smaller than the set of people with desire to be effective and altruistic.
For instance, if a fashion website has a 90% female audience (a not atypical number), that is not a claim that the “desire to look good” is that heavily skewed toward female. It means that the specific things that website caters to, the way it has marketed itself, etc. have resulted in it getting a female audience. Men could also desire to look good, albeit in ways that are very different from those catered to by that fashion website (or more broadly by the majority of present-day fashion websites).
Your suspicions provide a small amount of Bayesian evidence, but could you explain why you believe none of the alternate explanations that have been proposed seem satisfactory?