I have never been mistaken for male in person or on the phone, ever. Additionally, people who identify me as male (or choose to express their uncertainty with male pronouns) on the Internet aren’t typically doing so because there’s positive evidence to that effect; they’re guessing based on my location (“the Internet” or the specific site), which amounts to careless, casual stereotyping and rankles horribly. If people tended to only identify me as male after I dropped a casual reference to an ex-girlfriend without mentioning in the same context that I’m bi, that would bother me less, albeit still some, because it would be a reasonable update to make on the basis of information I’d provided beyond simply having wandered into an area that they suppose to be the province of males.
I know what I said about unicorns above, and I think that’s still relevant, but I disagree with your characterization of the gender misidentification as “stereotyping”.
Given that there are more men than women on Internet discussion sites, and especially on Less Wrong, wouldn’t it be reasonable to guess that any given poster is male, unless there’s evidence to the contrary ? By analogy, if I knew that a bag contained 75 black marbles and 25 white marbles, why shouldn’t I guess that a random marble, that I pulled out of the bag without looking, is black ?
I’m not sure how the looking would work in practice. I would feel incredibly creepy if, every time I wanted to quote someone’s blog post, I had to first contact the poster and inquire about his/her/etc. gender. Conversely, assuming I ever posted anything of consequence (unlikely, I know), I’d feel uncomfortable if someone asked me, “hey, I liked your blog post and I want to respond to it; BTW, what is your gender ?”.
I am not upset if people write in or around their uncertainty about my gender. “He/his” does not do either, but “(s)he” or “ey” or “they” or “Alicorn” or “the OP” or whatever would be all fine and no contact would be necessary.
I’m not sure how the looking would work in practice.
She mentioned that a simple google search would have done the job. Granted, the gender-inquiry step often bypasses one’s consciousness completely, and it happened to me here and elsewhere on more than one occasion.
This link is not stable. Google uses filter bubbles. I don’t have the same first results as you do.
In fact, the first two results for me point to LessWrong directly, the third to an MLP fan wiki, and the fourth to a random news article that apparently misspelled “unicorn”.
“alicorn gender site:lesswrong.com″ with the date restriction of before Feb 24 2010 (when she posted a question about correcting her gender in someone’s blog) gives me a pretty unambiguous second hit.
But you have to know that the person who uses the nickname ‘alicorn’ has posted something about her gender.
The word ‘alicorn’ itself doesn’t seem associated with anything femmine, other than the ‘unicorns are girly’ stereotype which is itself far from obvious.
(...) other than the ‘unicorns are girly’ stereotype which is itself far from obvious.
You mean the monstrous, superpowered godlike entity of human-level intelligence that purely selfishly rewards with mystical life-enhancing divine gifts those that save it, and those who would threaten it find themselves and all their relatives and descendents forever cursed, including any innocent offspring five generations removed from a single unicorn-threatening ancestor?
The first time I knew I’d probably encounter a unicorn in a game of D&D, I started rolling my next character.
All in all, I think the “unicorns are girly” stereotype isn’t all that widespread outside of certain typical US populations. For most populations, I’d figure the question of unicorn genderness never even occurs in the first place—unicorns are just one of those many “mythical creature” thinghies. Then again, I’m from a rather young generation and I have an extended family that is rather high standards in terms of gender cultural programming and social expectations (or prevention thereof).
All in all, I think the “unicorns are girly” stereotype isn’t all that widespread outside of certain typical US populations. For most populations, I’d figure the question of unicorn genderness never even occurs in the first place—unicorns are just one of those many “mythical creature” thinghies.
In whatever population I am part of (not a US one), it isn’t the beast itself that is considered female, but rather that females are more likely to be associated with it. Probably a selection effect because they are slightly less likely to be impaled on sight.
Perhaps the unicorn suffers from a similar problem as the angel. When I hear the word “Angel” I think “Enormously powerful, ruthless, highly masculine yet somewhat pretty enforcer that is quite likely to slaughter you on sight”. I don’t think “scantily clad girl with harp”. Unicorns are somewhat analogous albeit being territorial beasts rather than henchman.
I associate the word “Angel” with an eldritch inhuman monstrosity, the very sight of which will drive you mad, if you’re lucky. It is a messenger of an inscrutable divine omni-power, and it only ever carries one message: annihilation.
It is a messenger of an inscrutable divine omni-power, and it only ever carries one message: annihilation.
Angels carry at least two types of message: Annihilation and threats of annihilation if compliance with arbitrary demands is not immediate! Sometimes they are also scouts come to investigate whether said annihilation is necessary. Tip: if large flawlessly beautiful men walk up to your city don’t try to gang rape them. Offering your daughters up to the would-be rapists as a compromise is frowned upon but not penalized.
One of my most salient associations with unicorns is dangerous men. One of my friends was a social worker, and he found that nearly every time he saw the lodgings of a male serial rapist or other such severely disturbed male, they were decorated with unicorn posters.
Hah, yeah, that’s exactly the kind of usage I would come up with if I had to pick something unicorns would be a symbol for.
My image of unicorns as incredibly monstrous, scary supernatural creatures first came from the question: “Okay, it’s a white horse with a long, pointy, sometimes serrated or with screw-like sharp spiral edges, horn, but… what the hell do they use that horn for?”
they’re guessing based on my location (“the Internet” or the specific site), which amounts to careless, casual stereotyping and rankles horribly
My prior probability that someone is male is about 50%, knowing that they read Less Wrong amounts to about +10 dB of evidence that they are male (as of the last survey), so the posterior probability that someone is male given that they read Less Wrong is about 90%. How does Bayesian updating amount to careless, casual stereotyping?
I completely agree. Well, it doesn’t rankle for me in the same way because I probably post a lot less on the Internet than you do, and thus get a lot fewer assumptions. (Also, I kind of like the thought of people not knowing my gender.) But I completely agree that the Internet, and especially sites like LessWrong, is assumed to be populated by males.
Additionally, people who identify me as male (or choose to express their uncertainty with male pronouns) on the Internet aren’t typically doing so because there’s positive evidence to that effect...
Have you considered that your writing style might be unusual for a woman? Even based on a small sample of writing that has no obvious clues, it’s usually possible to guess the author’s sex much better than chance. You write in a very technical matter-of-fact style, with long, complex, and yet very precisely constructed sentences, and take unusual care to avoid ambiguities and unstated implications. (You’d probably be a great textbook writer.) Whatever the reason for this state of affairs might be, people who write like that are overwhelmingly men.
Also, why not simply use a female name if you’re bothered by this?
Perhaps I should stress that it’s not that you write in a typical masculine style. Rather, you write in a style that’s altogether unusual, and the minority of people who write like that are predominantly men. So it does constitute some evidence, unless I’m completely mistaken about the facts of the matter (and I pretty confident I’m not).
Regarding the typical male/female style, it’s hard to give a simple description. It’s an intuitive impression that’s not amenable to detailed introspection. Somehow a given text usually sounds more natural in male voice than female or vice versa, unless perhaps it’s a completely dry technical discussion, and while far from being 100% reliable, these guesses are also far better than chance. As for those clues that can be analyzed explicitly, I’m not sure if it would be a good idea to get into that topic, since it’s mostly about (statistically accurate) sex-stereotypes, which is clearly a hot-button issue.
It’s been my experience that writing style isn’t especially gendered. I used to think I could tell, but I can’t actually guess accurately based on writing style alone. (Topic choice, sure.) And, of course, women have successfully written under male pseudonyms many times. Lots of behaviors are gendered, but there’s psychological evidence that people are biased towards seeing gender differences in everything, and I think the “female writing style” is one of those supposed gender differences that doesn’t actually exist. If you want, though, we can see if you can guess the gender of the authors of a few writing samples where the choice of topic doesn’t give it away.
I initially thought Alicorn was male too, but that’s because she has a genderless username and writes on a majority-male site. I’ve been mistaken for a guy on the internet, when I thought my username was plenty girly, but, you know, I was on the internet so the priors are skewed.
I used to think I could tell, but I can’t actually guess accurately based on writing style alone. (Topic choice, sure.)
Writing style can’t be strictly separated from the choice of topic (or rather sub-topics addressed when writing on a given topic), and some of the most powerful clues come exactly from where these things blend into each other. Moreover, in interactive back-and-forth writing on forums and blogs, typical male and female behaviors and attitudes often quickly become apparent, just like in a live conversation, and are clearly detectable in writing.
Someone has already posted a link to a paper whose authors claim to have found measurable statistical differences between male and female styles, but I’m not sure how much (if at all) the usual human intuition relies on those specific clues.
But in any case, I don’t see where exactly you disagree with my above diagnosis, given that it’s discussing what I believe to be a fairly extreme and clear-cut case. Do you think Alicorn’s style doesn’t have the characteristics I described, or that such writing isn’t statistically likely to come from men?
If you want, though, we can see if you can guess the gender of the authors of a few writing samples where the choice of topic doesn’t give it away.
If you have some samples ready, I’d be curious to give it a try.
Two writing samples, one male and one female, from humanities journals:
First:
Machiavelli’s advocacy of force and fraud in the conduct of politics is the key teaching that has secured his reputation as “Machiavellian” and that has led to the conception of “The Prince” as the first document in the Western tradition to lay bare the dark, demonic underside of civic humanism. But this interpretation overlooks the degree to which a politics of intense competition and personal rivalry inhabits the humanist vision from antiquity, producing an ethics of expediency and a rhetoric of imposture that seeks to mask its alertness to advantage behind the guise of integrity and service. This vision is nowhere more apparent than in Cicero’s “De Oratore,” which exerted a powerful influence on the Italian humanists of the quattrocentro in whose direct descent Machiavelli stands. Deception, to put it simply, is an acknowledged and vital element in civic humanism long before “The Prince.” The difference is that Cicero typically couches it in a sacrificial rhetoric that is euphemistically inflected while Machiavelli opts for a hard-edged rhetoric of administrative efficiency to make his case. But the stylistic differences, important as they are, should not mask the essential affinity between the Machiavellian doctrine of princely fraud and the Ciceronian ethics of gentlemanly dissimulation.
Second:
This article analyzes the intellectual content of civic humanism in the specific context of Medici power, asking the question: what ideological role did civic humanism play in Medicean Florence? It argues that there is no contradiction between the ideals of civic humanism and support for the Medici regime. On the contrary, civic humanism could be used to justify and legitimate Medici power. The article analyzes the writings of principal humanists such as Leonardo Bruni, Poggio Bracciolini, and Francesco Filelfo, showing that Hans Baron’s republican “civic humanism” was compatible with different constitutional forms and different distributions of power.
Two excerpts from New York Times articles, one male and one female:
First:
Under newly fortified Republican control, many state governments started the year pledging forceful action to crack down on illegal immigration, saying they would fill a void left by the stalemate in Washington over the issue. Gov. Susana Martinez of New Mexico backed a repeal to a law granting driver’s licenses to illegal immigrants. Now, with some legislatures winding down their sessions, the lack of consensus that has immobilized Congress has shown up in the legislatures as well, and has slowed — but not stopped — the advance of bills to penalize illegal immigrants. No state has passed a law that replicates the one adopted last April in Arizona, which greatly expanded the powers of police officers to question the immigration status of people they stop. Still, immigrant advocates in many states say the debate has clearly shifted in favor of tougher enforcement. They say they have had to fight just to hold the line on immigration issues that they thought were long settled.
Second:
With the referendum over the constitutional amendments that will shape Egypt’s immediate political future just days away, the country’s nascent political forces were squaring off on Sunday, scrambling to influence a choice that leaves many confused.
The Muslim Brotherhood and rump elements of the disbanded governing National Democratic Party, which both stand to gain the most from a rapid rebirth of electoral politics, support the amendments. Arrayed against them is much, but not all, of the remaining political spectrum, centered on the young organizers behind the Tahrir Square demonstrations who fear a yes vote would ultimately rob them of their revolution. Yet everyone agrees on two things. The referendum, which is scheduled for Saturday, will be a milestone and the first one not rigged outright in about 60 years. “Whether we accept the amendments or we reject them, either situation means a page in our history will turn,” said Amr Shubaki, a political analyst at the state-financed Ahram Center for Political and Strategic Studies.
Two excerpts from short stories, one male and one female.
First:
An old man with radiation scars surrounding the chromed half of his face limped down a salt-grass covered dune. Metal armatures creaked under his clothing as he thumped heavily across dry sand to wet, scuffing through the black and white line of the high-tide boundary, where the sharp glitter of cast-up teeth tangled in film-shiny ribbons of kelp. About his feet, small combers glittered in the light of a gibbous moon. Above, the sky was deepest indigo: the stars were breathtakingly bright.
The old man, whose name was Aethelred, fetched up against a large piece of seawrack, perhaps the wooden keel of some long-ago ship, and made a little ceremony of seating himself. He relied heavily on his staff until his bad leg was settled, and then he sighed in relief and leaned back, stretching and spreading his robes out around him.
Second:
Kalak rounded a rocky stone ridge and stumbled to a stop before the body of a dying thunderclast. The enormous stone beast lay on its side, riblike protrusions from its chest broken and cracked. The monstrosity was vaguely skeletal in shape, with unnaturally long limbs that sprouted from granite shoulders. The eyes were deep red spots on the arrowhead face, as if created by a fire burning deep within the stone. They faded. Even after all these centuries, seeing a thunderclast up close made Kalak shiver. The beast’s hand was as long as a man was tall. He’d been killed by hands like those before, and it hadn’t been pleasant. Of course, dying rarely was.
The humanities articles both use the same formal dry academic style, from which it’s hard to say anything. The first one is much worse in terms of long-winded verbiage, but that doesn’t say much. If I had to guess, I would toss a coin.
The Times articles are similarly written in a routine journalistic style (it reads like a telegram with some cliche phrases cut and pasted between the words), so again it’s hard to say anything. (Also, from what I know, news articles are heavily edited and it’s questionable how much individual style they preserve.) If I really had to guess, I would say the first is more likely to be female by an epsilon. Google confirms this is correct, but I admit I wouldn’t bet any money on it. (On the other hand, I wouldn’t be surprised if an experienced newspaper editor could guess much better.)
As for the short stories excerpts, well, that’s some bad prose. The first one sounds to me a bit more feminine; I’d say it’s something like a 60-40 guess. Googling these paragraphs, I see that I guessed right; admittedly, I wouldn’t have bet very much money on this one either.
Do you think Alicorn’s style doesn’t have the characteristics I described, or that such writing isn’t statistically likely to come from men?
Well, she thinks explicitly and abstractly, like most people here, and I suppose that could be more common in men, but I don’t think I’ve noticed anything especially male or female in her prose. I didn’t notice an unusual lack or predominance of pronouns. (Actually I think Alicorn, more than most LessWrongers, tends to illustrate ideas with anecdotes about individual people, whether real or hypothetical. So that would mean more pronouns—but then again, Eliezer has the same habit, and I don’t know if that means you’d consider his writing feminine.)
I’ve been collecting examples of Eliezer being mistaken for female; so far I’ve got six, plus two people uncertain. (Someone suggested that it’s because of his name, but I don’t remember why.)
Numerous cases in Methods of Rationality, especially during the early days. It’s as if they had priors suggesting that most Harry Potter fanfiction authors were female.
For what it is worth Alicorn’s writing style always resolved to female written for me. And the name seemed even more female—along the lines of “Alison”. My intuition possibly focuses on somewhat different features of communication when making the distinctions.
Also, why not simply use a female name if you’re bothered by this?
Being bothered is not usually about avoiding the negative stimulus.
Whatever the reason for this state of affairs might be, people who write like that are overwhelmingly men.
This may be (although I’d like to see solid data before assuming so). But I also suspect that being on a rationality blog acts a filter for the sorts of people who DON’T write like that.
Have you considered that your writing style might be unusual for a woman? Even based on a small sample of writing that has no obvious clues, it’s usually possible to guess the author’s sex much better than chance. You write in a very technical matter-of-fact style, with long, complex, and yet very precisely constructed sentences, and take unusual care to avoid ambiguities and unstated implications. (You’d probably be a great textbook writer.) Whatever the reason for this state of affairs might be, people who write like that are overwhelmingly men.
That just says ‘well educated and highly intelligent’ to me. Now, such people tend to be more commonly male than female, but given that someone posts on Less Wrong I don’t think that writing style is further evidence for them being male.
I’m not sure which shadows which here. It doesn’t seem like writing in this style will cause someone to post on LessWrong, while conversely it seems much more likely that someone who has been posting on LessWrong for a while will adopt this writing style. Thus, ISTM that the writing style overshadows posting-on-LessWrong more than the other way around.
Given that I have the text itself, learning that it was posted on LessWrong I wouldn’t infer much from it, since regardless of gender they aren’t obviously more likely to write this way on LessWrong if they’re a regular user, whilst if I only know that someone is posting on LessWrong then learning that they write in this manner also gives me all the other non-LessWrong writer data.
Not sure if I’m really being clear. Basically, in my model the causal chain goes the other way around.
My priors before I started paying attention at all (i.e. before the first time it came up in a conversation I was part of, on the internet, that someone was uncomfortable / sad / whatever that they were being referred to as the wrong gender—which, for record, was a post-op man genetically female who still had a few culturally-programmed female-expected behaviors) were around about .4 female to .6 male for any random person I meet and discuss with on the Internet.
Even that .4 seems rather high compared to the base stats I’ve seen since then for some populations, but there’s apparently some factor which makes me more likely to engage and enjoy discussions and interactions over the web with women, for some reason I don’t yet understand.
However, since then, I’ve had to update downwards. Even with my abnormal encounter rates (e.g. meeting 30% women in communities that are 3% women overall), on average I still only expect and observe that I “befriend” (or otherwise engage and interact with more actively with) women only one in five times of such people, i.e. the other four are men. This if I only include so-called “normal” men and women, because I also end up meeting abnormally high numbers of transgenders, asexuals, queers, and other nonstandard genders.
On top of that, out of the women that I do tend to interact with (which are already at less than 0.2 expected rate), only one in two cases I’ll end up having to refer to them before their gender becomes “revealed” in some manner (sometimes because of an obvious nickname). Of the half where I do, nowadays I use gender-neutral format, but before I started doing so, only one in four (well, three out of thirteen to the best of my memory, in total) got slighted/offended/whatever that I used male pronouns.
Which basically ends up with there being an approximately 2% expected probability that for any person I start interacting with I might use a male pronoun for a woman that will be affected by it, unless there is a high amount of women who get hurt from wrong pronoun usage but never reveal this. This is including my current rate of female:male encounter ratio, which I’ve already confirmed is abnormal from other stats (e.g. I’ve eliminated cases like women just befriending more people than men or similar situations).
I would expect that most people have their expectations somewhere around this or even rarer, so I can’t fault them for using male pronouns on the Internet by default without first having to fault them for thousands of other, much worse things.
For bias evaluation purposes: I have been identified and referred to with female pronouns at least twice on the Internet, and wasn’t offended (in one case, I was actually flattered, for various contextual reasons). In the past week, excluding work colleagues and LessWrong, I’ve maintained more meaningful Internet-based interactions with three regular men, two women, and one genetically-male unclear-someone who hasn’t quite yet resolved his own personal gender-identity yet and would probably fall in some gray area between “queer” and “pre-op transgender”.
I have never been mistaken for male in person or on the phone, ever. Additionally, people who identify me as male (or choose to express their uncertainty with male pronouns) on the Internet aren’t typically doing so because there’s positive evidence to that effect; they’re guessing based on my location (“the Internet” or the specific site), which amounts to careless, casual stereotyping and rankles horribly. If people tended to only identify me as male after I dropped a casual reference to an ex-girlfriend without mentioning in the same context that I’m bi, that would bother me less, albeit still some, because it would be a reasonable update to make on the basis of information I’d provided beyond simply having wandered into an area that they suppose to be the province of males.
I know what I said about unicorns above, and I think that’s still relevant, but I disagree with your characterization of the gender misidentification as “stereotyping”.
Given that there are more men than women on Internet discussion sites, and especially on Less Wrong, wouldn’t it be reasonable to guess that any given poster is male, unless there’s evidence to the contrary ? By analogy, if I knew that a bag contained 75 black marbles and 25 white marbles, why shouldn’t I guess that a random marble, that I pulled out of the bag without looking, is black ?
Only if you are unable to actually look and check the color. Which was Alicorn’s whole point.
I’m not sure how the looking would work in practice. I would feel incredibly creepy if, every time I wanted to quote someone’s blog post, I had to first contact the poster and inquire about his/her/etc. gender. Conversely, assuming I ever posted anything of consequence (unlikely, I know), I’d feel uncomfortable if someone asked me, “hey, I liked your blog post and I want to respond to it; BTW, what is your gender ?”.
But perhaps my reaction is atypical ?
I am not upset if people write in or around their uncertainty about my gender. “He/his” does not do either, but “(s)he” or “ey” or “they” or “Alicorn” or “the OP” or whatever would be all fine and no contact would be necessary.
She mentioned that a simple google search would have done the job. Granted, the gender-inquiry step often bypasses one’s consciousness completely, and it happened to me here and elsewhere on more than one occasion.
Actually a simple google search yields this: http://www.google.it/search?q=alicorn
The first result is Wikipedia, the second and the third are My Little Pony stuff, and they even mention a male alicorn.
This link is not stable. Google uses filter bubbles. I don’t have the same first results as you do.
In fact, the first two results for me point to LessWrong directly, the third to an MLP fan wiki, and the fourth to a random news article that apparently misspelled “unicorn”.
Interesting.
“alicorn gender site:lesswrong.com″ with the date restriction of before Feb 24 2010 (when she posted a question about correcting her gender in someone’s blog) gives me a pretty unambiguous second hit.
But you have to know that the person who uses the nickname ‘alicorn’ has posted something about her gender.
The word ‘alicorn’ itself doesn’t seem associated with anything femmine, other than the ‘unicorns are girly’ stereotype which is itself far from obvious.
You mean the monstrous, superpowered godlike entity of human-level intelligence that purely selfishly rewards with mystical life-enhancing divine gifts those that save it, and those who would threaten it find themselves and all their relatives and descendents forever cursed, including any innocent offspring five generations removed from a single unicorn-threatening ancestor?
The first time I knew I’d probably encounter a unicorn in a game of D&D, I started rolling my next character.
All in all, I think the “unicorns are girly” stereotype isn’t all that widespread outside of certain typical US populations. For most populations, I’d figure the question of unicorn genderness never even occurs in the first place—unicorns are just one of those many “mythical creature” thinghies. Then again, I’m from a rather young generation and I have an extended family that is rather high standards in terms of gender cultural programming and social expectations (or prevention thereof).
In whatever population I am part of (not a US one), it isn’t the beast itself that is considered female, but rather that females are more likely to be associated with it. Probably a selection effect because they are slightly less likely to be impaled on sight.
Perhaps the unicorn suffers from a similar problem as the angel. When I hear the word “Angel” I think “Enormously powerful, ruthless, highly masculine yet somewhat pretty enforcer that is quite likely to slaughter you on sight”. I don’t think “scantily clad girl with harp”. Unicorns are somewhat analogous albeit being territorial beasts rather than henchman.
I associate the word “Angel” with an eldritch inhuman monstrosity, the very sight of which will drive you mad, if you’re lucky. It is a messenger of an inscrutable divine omni-power, and it only ever carries one message: annihilation.
Angels carry at least two types of message: Annihilation and threats of annihilation if compliance with arbitrary demands is not immediate! Sometimes they are also scouts come to investigate whether said annihilation is necessary. Tip: if large flawlessly beautiful men walk up to your city don’t try to gang rape them. Offering your daughters up to the would-be rapists as a compromise is frowned upon but not penalized.
In one case, the message was “You’re pregnant”. Then, later, a whole chorus of angels gave a concert in celebration of that child’s birth.
Which is not to detract from your point! The very first thing those angels said, to Mary and the shepherds both, was “Fear not!”
This set of posts made my day.
It seems to be encouraged, not frowned upon. And offering your daughter and the wife of the intended (non-angelic) victim is super-righteous.
Um, no.
[source]
“The sex of angels” is an Italian idiom for an irrelevant question, much like “how many angels can dance on the head of a pin” is in English.
One of my most salient associations with unicorns is dangerous men. One of my friends was a social worker, and he found that nearly every time he saw the lodgings of a male serial rapist or other such severely disturbed male, they were decorated with unicorn posters.
...huh.
Be right back. If anybody needs me, I’ll be reevaluating everything I thought I knew about My Little Pony.
Hah, yeah, that’s exactly the kind of usage I would come up with if I had to pick something unicorns would be a symbol for.
My image of unicorns as incredibly monstrous, scary supernatural creatures first came from the question: “Okay, it’s a white horse with a long, pointy, sometimes serrated or with screw-like sharp spiral edges, horn, but… what the hell do they use that horn for?”
Stabbing lions. And ill-informed hunters.
More like 89 black marbles, 8 white marbles, and 3 striped/grey/transparent/other-colour marbles. But still, I don’t usually use gendered pronouns unless I’m >95% sure of someone’s gender (from the venue, their username, and what I’ve read by them so far).
Given that the prior for male is 89%, iit doesn’t seem it would take lots of evidence to reach 95% posterior probability that somebody is male.
My prior probability that someone is male is about 50%, knowing that they read Less Wrong amounts to about +10 dB of evidence that they are male (as of the last survey), so the posterior probability that someone is male given that they read Less Wrong is about 90%. How does Bayesian updating amount to careless, casual stereotyping?
I completely agree. Well, it doesn’t rankle for me in the same way because I probably post a lot less on the Internet than you do, and thus get a lot fewer assumptions. (Also, I kind of like the thought of people not knowing my gender.) But I completely agree that the Internet, and especially sites like LessWrong, is assumed to be populated by males.
Alicorn:
Have you considered that your writing style might be unusual for a woman? Even based on a small sample of writing that has no obvious clues, it’s usually possible to guess the author’s sex much better than chance. You write in a very technical matter-of-fact style, with long, complex, and yet very precisely constructed sentences, and take unusual care to avoid ambiguities and unstated implications. (You’d probably be a great textbook writer.) Whatever the reason for this state of affairs might be, people who write like that are overwhelmingly men.
Also, why not simply use a female name if you’re bothered by this?
Out of curiosity, what markers do you associate with feminine writing?
Perhaps I should stress that it’s not that you write in a typical masculine style. Rather, you write in a style that’s altogether unusual, and the minority of people who write like that are predominantly men. So it does constitute some evidence, unless I’m completely mistaken about the facts of the matter (and I pretty confident I’m not).
Regarding the typical male/female style, it’s hard to give a simple description. It’s an intuitive impression that’s not amenable to detailed introspection. Somehow a given text usually sounds more natural in male voice than female or vice versa, unless perhaps it’s a completely dry technical discussion, and while far from being 100% reliable, these guesses are also far better than chance. As for those clues that can be analyzed explicitly, I’m not sure if it would be a good idea to get into that topic, since it’s mostly about (statistically accurate) sex-stereotypes, which is clearly a hot-button issue.
It’s been my experience that writing style isn’t especially gendered. I used to think I could tell, but I can’t actually guess accurately based on writing style alone. (Topic choice, sure.) And, of course, women have successfully written under male pseudonyms many times. Lots of behaviors are gendered, but there’s psychological evidence that people are biased towards seeing gender differences in everything, and I think the “female writing style” is one of those supposed gender differences that doesn’t actually exist. If you want, though, we can see if you can guess the gender of the authors of a few writing samples where the choice of topic doesn’t give it away.
I initially thought Alicorn was male too, but that’s because she has a genderless username and writes on a majority-male site. I’ve been mistaken for a guy on the internet, when I thought my username was plenty girly, but, you know, I was on the internet so the priors are skewed.
Well, there’s this.
I did not know about that!
(Quick summary: n-gram analysis shows that women use more pronouns than men, among other distinctions.)
Ok, there does seem to be such a thing as a gender difference in writing style. Even within genres.
Writing style can’t be strictly separated from the choice of topic (or rather sub-topics addressed when writing on a given topic), and some of the most powerful clues come exactly from where these things blend into each other. Moreover, in interactive back-and-forth writing on forums and blogs, typical male and female behaviors and attitudes often quickly become apparent, just like in a live conversation, and are clearly detectable in writing.
Someone has already posted a link to a paper whose authors claim to have found measurable statistical differences between male and female styles, but I’m not sure how much (if at all) the usual human intuition relies on those specific clues.
But in any case, I don’t see where exactly you disagree with my above diagnosis, given that it’s discussing what I believe to be a fairly extreme and clear-cut case. Do you think Alicorn’s style doesn’t have the characteristics I described, or that such writing isn’t statistically likely to come from men?
If you have some samples ready, I’d be curious to give it a try.
Two writing samples, one male and one female, from humanities journals: First:
Second:
Two excerpts from New York Times articles, one male and one female:
First:
Second:
Two excerpts from short stories, one male and one female.
First:
Second:
The humanities articles both use the same formal dry academic style, from which it’s hard to say anything. The first one is much worse in terms of long-winded verbiage, but that doesn’t say much. If I had to guess, I would toss a coin.
The Times articles are similarly written in a routine journalistic style (it reads like a telegram with some cliche phrases cut and pasted between the words), so again it’s hard to say anything. (Also, from what I know, news articles are heavily edited and it’s questionable how much individual style they preserve.) If I really had to guess, I would say the first is more likely to be female by an epsilon. Google confirms this is correct, but I admit I wouldn’t bet any money on it. (On the other hand, I wouldn’t be surprised if an experienced newspaper editor could guess much better.)
As for the short stories excerpts, well, that’s some bad prose. The first one sounds to me a bit more feminine; I’d say it’s something like a 60-40 guess. Googling these paragraphs, I see that I guessed right; admittedly, I wouldn’t have bet very much money on this one either.
Well, she thinks explicitly and abstractly, like most people here, and I suppose that could be more common in men, but I don’t think I’ve noticed anything especially male or female in her prose. I didn’t notice an unusual lack or predominance of pronouns. (Actually I think Alicorn, more than most LessWrongers, tends to illustrate ideas with anecdotes about individual people, whether real or hypothetical. So that would mean more pronouns—but then again, Eliezer has the same habit, and I don’t know if that means you’d consider his writing feminine.)
I’ve been collecting examples of Eliezer being mistaken for female; so far I’ve got six, plus two people uncertain. (Someone suggested that it’s because of his name, but I don’t remember why.)
Numerous cases in Methods of Rationality, especially during the early days. It’s as if they had priors suggesting that most Harry Potter fanfiction authors were female.
Aren’t they? That’s always been my impression. Although I can think of a lot of exceptions, like you, nonjon, and the guy who wrote Wastelands.
Someone mentioned that his first name could be misread as Eliza.
I didn’t base my conclusion on pronouns at all. Maybe you missed my commend a few turns further up in the thread where I describe it in more detail.
For what it is worth Alicorn’s writing style always resolved to female written for me. And the name seemed even more female—along the lines of “Alison”. My intuition possibly focuses on somewhat different features of communication when making the distinctions.
Being bothered is not usually about avoiding the negative stimulus.
This may be (although I’d like to see solid data before assuming so). But I also suspect that being on a rationality blog acts a filter for the sorts of people who DON’T write like that.
That just says ‘well educated and highly intelligent’ to me. Now, such people tend to be more commonly male than female, but given that someone posts on Less Wrong I don’t think that writing style is further evidence for them being male.
I’m not sure which shadows which here. It doesn’t seem like writing in this style will cause someone to post on LessWrong, while conversely it seems much more likely that someone who has been posting on LessWrong for a while will adopt this writing style. Thus, ISTM that the writing style overshadows posting-on-LessWrong more than the other way around.
Given that I have the text itself, learning that it was posted on LessWrong I wouldn’t infer much from it, since regardless of gender they aren’t obviously more likely to write this way on LessWrong if they’re a regular user, whilst if I only know that someone is posting on LessWrong then learning that they write in this manner also gives me all the other non-LessWrong writer data.
Not sure if I’m really being clear. Basically, in my model the causal chain goes the other way around.
My priors before I started paying attention at all (i.e. before the first time it came up in a conversation I was part of, on the internet, that someone was uncomfortable / sad / whatever that they were being referred to as the wrong gender—which, for record, was a post-op man genetically female who still had a few culturally-programmed female-expected behaviors) were around about .4 female to .6 male for any random person I meet and discuss with on the Internet.
Even that .4 seems rather high compared to the base stats I’ve seen since then for some populations, but there’s apparently some factor which makes me more likely to engage and enjoy discussions and interactions over the web with women, for some reason I don’t yet understand.
However, since then, I’ve had to update downwards. Even with my abnormal encounter rates (e.g. meeting 30% women in communities that are 3% women overall), on average I still only expect and observe that I “befriend” (or otherwise engage and interact with more actively with) women only one in five times of such people, i.e. the other four are men. This if I only include so-called “normal” men and women, because I also end up meeting abnormally high numbers of transgenders, asexuals, queers, and other nonstandard genders.
On top of that, out of the women that I do tend to interact with (which are already at less than 0.2 expected rate), only one in two cases I’ll end up having to refer to them before their gender becomes “revealed” in some manner (sometimes because of an obvious nickname). Of the half where I do, nowadays I use gender-neutral format, but before I started doing so, only one in four (well, three out of thirteen to the best of my memory, in total) got slighted/offended/whatever that I used male pronouns.
Which basically ends up with there being an approximately 2% expected probability that for any person I start interacting with I might use a male pronoun for a woman that will be affected by it, unless there is a high amount of women who get hurt from wrong pronoun usage but never reveal this. This is including my current rate of female:male encounter ratio, which I’ve already confirmed is abnormal from other stats (e.g. I’ve eliminated cases like women just befriending more people than men or similar situations).
I would expect that most people have their expectations somewhere around this or even rarer, so I can’t fault them for using male pronouns on the Internet by default without first having to fault them for thousands of other, much worse things.
For bias evaluation purposes: I have been identified and referred to with female pronouns at least twice on the Internet, and wasn’t offended (in one case, I was actually flattered, for various contextual reasons). In the past week, excluding work colleagues and LessWrong, I’ve maintained more meaningful Internet-based interactions with three regular men, two women, and one genetically-male unclear-someone who hasn’t quite yet resolved his own personal gender-identity yet and would probably fall in some gray area between “queer” and “pre-op transgender”.