Tuesday, 8 December 2015

If You Really Cared About Effective Altruism You'd Give All Your Money To Feminist Frequency

I'm interested in the effective altruism movement. I think it is on the whole a good thing. Even if you disagrees with the hardcore universalist utilitarian principles upon which mainstream EA is based, the general idea that we should evaluate charities based upon how much good they do seems to me to be solid. All charity may be good, but some charities are better than your local donkey sanctuary.

I don't personally give to effective altruist charities, or any charities at all, excluding the occasional pound for a remembrance poppy. As a long term and current student, I am not exactly flushed with cash. I think those that pledge for the future when they're not able to actually give are making a very empty gesture, see the number of students currently signed up here. In the future, however, I will consider it.

My interest in EA, however is mainly in its structure and the space it occupies in the social marketplace. The current movement is heavily influenced by people who are, for want of a better word, total shitlords. The obsession with data, auditing your own beliefs, and treating human life and happiness as a simple variable in your spreadsheet to be maximised is not one that is held by someone with a below average autistic spectrum quotient. They enjoy stories about failed well meaning projects, such as roundabout water pumps, and are determined to do better than the dumb do-gooders who currently direct the world's aid money. They also revel in bullet biting on tough moral questions, in ways that make many people feel very uncomfortable. See, for example, EA demi-God Peter Singer's comments on infanticide and abortion. These people are my ingroup.

Many of these same people are also interested in other questions that are simply far beyond the current Overton window of public debate. The fringes of the rationalist blogosphere has long spent time discussing questions about AI risk, and many prominent EAs strongly endorse donating money to organisations devoted to assessing such risks. While fascinating, these questions are probably not the most important thing we need to focus on right now, they have few immediate benefits for the world and are always going to have little mainstream appeal.

Opposing these slightly odd types are more traditional liberals. These people are equally concerned with improving general global welfare, but hold slightly more conventional moral views, and are uncomfortable with the more esoteric questions their opposite numbers enjoy. The Less Wrongers see EA as an all encompassing new moral paradigm, with all the somewhat cultish behaviour that implies. The liberals see EA as a method of refining their moral convictions and implementing them in a more efficient way.

I'm unsure if these two groups can continue to work together much longer, the movement is starting to reach the critical stage at which it is large enough to potentially split. The faultline here looks very similar to that which killed the broad atheist/skeptic coalition about five years ago with atheism plus, and ultimately led to the eternal blood war of gamergate. Nerds and herbivores are just similar enough to be natural enemies.

The break is a little more interesting than simple kulturkampf though, because of the incentive gradients involved. Let's say I'm an important figure in the effective altruism movement, perhaps working for Giving What We Can or a similar organisation. My noble goal is quite simply to improve human welfare through charitable giving. In order to do this, I can attempt to refine the effectiveness of existing charity, and I can try to recruit as many people as possible to donate as much as possible in this way.

But what if goals conflict? What if by focusing rigorously on the most effective things, I start confronting questions that will piss off one side of the see-saw. Talk too much about AI risk, and I'll alienate the normos, talk too little about it and I'll bore the spergs. This conflict seems to me to be unresolvable. Eventually I'm going to have to pick a side.

From a recruitment perspective, the growth area seems to me to be on the liberal side. There are simply far more of them. If we can convince every smart university graduate who had an Obama poster on their wall in 08 and tweets in solidarity with #blacklivesmatter and Caitlyn Jenner to give 10% of their income to charity, we could do a lot of good. On the other hand there simply aren't that many movement rationalists out there. If we can convince the entire user base of Less Wrong to give 10% of their income to charity, we can do a lot less good. Our nerds, therefore, need to fall by the wayside, their obsession with AI risk is nutty, and they put off recruitment at the other end. The most effective altruistic strategy for effective altruism is to focus on recruiting young liberals to the cause.

The incentives here, however, become somewhat strange. What if rigorously focusing on schistosomiasis, two of the top four recommended charities by evaluators Give Well, isn't optimal? Even if every marginal dollar spent on SCI is best, if it harms my recruitment through being a bit gross and generally non-sexy, it has to go. Giving money to alright, but not fantastic, charities may be a good thing if it encourages more people to sign up. A million people giving 10% of their income to an 8/10 charity is much much better than a thousand people giving to a 10/10 one in terms of global human welfare.

To recruit even more people, we need to focus on building the reputation of effective altruism. The most effectively altruistic thing to do is to publicly abandon the founding principles of the movement. Lots of liberals seem to care about sexism in video games. This is an incredibly unimportant issue compared to global poverty, but if GWWC chuck a tenner a month at Feminist Frequency, and this generates enough good publicity to give us another hundred liberal 10% pledgers, maybe it's a good thing? Throwing off the connotations of brogrammers and nerds may be a really important component of mass growth of the cause. Maybe we should actually pledge money to donkey sanctuaries if it helps us get a few more donkey lovers on board, provided there's always a marginal increase in the schistosomiasis money?

This, of course, assumes that the people behind effective altruism are truly committed to its founding principles. My proposal may well be the best way of helping the most people, and if so they should seriously consider it, but it has other drawbacks. The biggest of these is that maybe there is more to the motivations behind EA than meets the eye, a point made by Sam Bowman when I ran this past him last night. Effective altruism sends out several social signals, it is a costly and therefore credible signal of virtue and wealth, and it is more generally associated with high IQ and status. The level of donation leads to the former, the process and targets are associated with the latter. Distinguishing themselves from regular slapdash philanthropists may therefore be important for the latter function. They want to be better than the average charity donor, in terms of money, sure, but just as importantly in terms of information. Effective altruism may be a positional good, if it were just giving money away, many may lose interest. Giving money to Anita Sarkeesian is silly, I wouldn't want to be associated with it, and I doubt most EAs would either. It shows you're a low status idiothole. What's more, the more average people you recruit into the movement, the noisier the signal is. Once everyone is an EA, its no longer cool.

So we're left with a paradox. If EAs really cared about being effectively altruistic, they would do all they could to improve recruitment, this may include sending out the right signals by throwing money at Feminist Frequency. But if what they really care about is signalling that they are better than that, they won't.

Saturday, 14 November 2015

This Is What Disney Princesses Would Look Like With Ambiguous Genitalia

The TERF wars have been raging for a few decades, but have become especially prevalent in the last couple of years. They really are a fantastic toxoplasma, and provide endless fodder for vicious online conflict, mostly between women with plenty of intrasexual competition[1] for kicks. In the blue corner are the TERFs, the Trans Exclusionary Radical Feminists. Led by the entire staff of the New Statesman, Germaine Greer and notorious B&Q lesbian Julie Bindel. In the red corner are the sexy young pro-trans intersectionalistas, recently, Paris Lees has come to the fore as a leader, but they tend to move in packs, led by semi-anonymous social media users rather than prominent mainstream figures. So far it's pretty much been a bloodbath, with the reds winning, but there are often signs that the TERFs may gain ground.

What I think they need right now is for a man to turn up with some real science... This post will not be about trans people especially, though there is plenty of interesting research into that area, and I may cover it in a future post. I'm instead going to guide you as simply as I can through the process of sexual development, how this can go wrong, and the degrees of intersexuality this can lead to. [2]

As always, Richard Dawkins is the King of Twitter and right about everything.

The question of who is and who isn’t a woman is not a scientific one. The map is not the territory, and people can draw their borders all over however they like. However, this doesn't mean that there isn't a stonking great mountain in the middle of it. My aim is to identify as many mountains as I can, in order that it may either clarify or cloud your map, but hopefully in a way that makes the territory clearer. Though this obviously can’t answer the question of who is or isn’t a woman or a man, it provides several interesting margin cases which may help clarify your thinking on the issue. 


When drawing their borders, the TERFs tend to put emphasis on possession of biological characteristics, such as XX chromosomes or a uterus. Simultaneously, they put a lot of emphasis on lived and learned experience of being a woman in society, and tend to disagree strongly that behaviour and psychology has much of a biological basis. In this way, the very idea of a trans woman is a contradiction.

When the intersectionalistas use the word woman, they include in that set anyone who chooses to be in it. They tend to disparage people who talk about anatomy, and put a lot more emphasis on the question of personal identity. They equally emphasise environmental factors in developing our behaviour, but disagree that a male upbringing precludes someone from being a ‘real woman’.

So the one hand, we have the TERFs, who I think are wrong about many things. I hope that this post may help to clarify some of the ways they are wrong, by showing the world is a lot more complicated than men and women, there will be men with uteruses, women with penises and everything inbetween, without even addressing about what most people think of as trans people. I also want to challenge the idea that gender is learned or performative, and show some evidence that it has a strong innate basis, regardless of experience in society.

On the other hand, we have the intersectionalistas. Their ‘genderbread person’ has a lot to it that is worthwhile, showing that there are several different axis we can study, and that these do not always perfectly line up. Their ‘make no assumptions’ mantra, however, while a good rule for some social situations, is completely anti-knowledge, and ignores the way that these dimensions usually covary in predictable ways based on biology and human development. Frequently people will claim that human sexuality has nothing to do with your gender, but it correlates 95% with it. The vast majority of women are androphile, the vast majority of men are gynephile[3] and there are reasons this happens which are mostly little to do with socialisation.


Sex differentiation begins with chromosomes. In almost all cases, XX is female and XY is male, but this isn’t the entire story. The default development pattern of the human foetus is essentially female (hence why men have nipples). In order to develop as male, additional factors which inhibit female and promote male development are necessary. The key is not the Y chromosome per se, but a gene on it imaginatively called the sex-determining region Y (SRY), which is responsible for differentiating the hitherto bipotential gonads into a testes, whereas otherwise it will develop into an ovary. Basic chromosomal abnormalities can shed some light on this process.

Turner syndrome is X0. I.e. The person has only one X chromosome, and so lacks the SRY and does not develop male characteristics, but is missing the other X necessary for normal female development.[4] Among other medical problems that are not relevant to this post, people with Turner’s syndrome have poorly developed ovaries, and are so almost always infertile. They do not go through puberty naturally, so don’t properly develop secondary female sex characteristics, such as breasts and larger hips, and do not menstruate. Typically they will be treated with estrogen replacement therapy, beginning when puberty should have occurred or at diagnosis. There are cases in which women with Turner’s syndrome have become pregnant, but this is difficult, and requires IVF using a donor egg.

People with Turner’s syndrome generally consider themselves to be women, and it seems to me that almost everyone in the world would agree with this. If you take the definition of woman as meaning ‘XX chromosomes’, however, they would not be included. Equally, if we say that a woman is someone who menstruates as an adult, they are left out of the category. Similar to trans women, but unlike most other women, they use hormones medically to achieve female secondary sex characteristics. They identify as women, and are treated by those around them as women for the duration of their lives, so may be considered under threat of sexism. They seem no more likely to be lesbian than the general population. They have a vagina and a uterus, not a penis and scrotum, but they have a mass of undifferentiated tissue, rather than real functional ovaries. Some of these definitions here both match up to our natural intuitions, some don’t seem to.


Individuals with Klinefelter’s syndrome have an extra chromosome, they are XXY[6]. They generally look male (due to the presence of SRY), but often have small penis and testes, less body hair, less muscle mass, poor fertility and gynaecomastia (male breast growth/moobs) stemming from lower testosterone levels than average. These all fit with a general pattern of slight feminisation, though people with Klinefelter’s are taller than average men, which does not. They do not seem to be more likely to be gay than the general population, but this may be complicated by a generally lower sex drive. People with Klinefelter’s will often take testosterone supplements in order to counteract some of these problems, in a way that may seem similar to female to male transsexuals.

Again, almost everyone would normally consider someone with Klinefelter’s to be male, but their chromosomes don’t quite match up, and they have several more feminine physical features. The majority consider themselves to be men, though I have found some weak evidence of a very slight increase in prevalence of Klinefelter’s among male to female transsexuals, with several case reports on a literature search, and a rate of 3 in 251 in a sample who were tested, which is well above the expected rate, but hardly conclusive.[6]


There are another two relevant chromosomal abnormalities, with are in many ways the reverse of one another. In Swyer Syndrome, the individual has XY chromosomes, but our old friend the SRY is non-functioning, and so the testes do not develop, becoming non-functioning streak gonads. A vagina and uterus does form, and they can become pregnant using donor eggs. The overall picture is similar to Turner’s, puberty will not occur without medical intervention to provide hormones, though in this case they tend to only present upon failure to reach puberty, whereas Turner’s syndrome patients have more other problems which can show up earlier.

De La Chapelle
syndrome gives us an XX individual, but who has somehow acquired an SRY on one of their chromosomes due to a problem in the formation of the father’s sperm cells. They are anatomically normal males, with some abnormalities generally similar to those in Klinefelter’s, with the exception of lower than average height. They go through puberty naturally, but will be infertile. Many never discover they are XX at all, until they discover they are unable to conceive.

These perhaps pose more difficult question for answering whether someone is male or female, almost everyone would agree that the former are women and the latter men. The chromosomes. however, start to fall out of the window at this point.


Even if the chromosomes are normal, there is plenty that can go wrong. One particularly fun bit of infighting that occurs in the TERF wars is the fight over whether you can have a male uterus.

The mullerian duct is the anatomical precursor of most female internal sex organs. In the male foetus, anti mullerian hormone is produced, which obliterates the duct before is can develop any further. If this hormone is not produced, Persistent Mullerian Duct Syndrome occurs. Normal male sex development occurs around it, but a small uterus, fallopian tubes and even cervix can end up mixed up inside getting in the way. So here we have someone who is a man, with XY chromosomes, a penis, testicles, plenty of testosterone, fertility, the works, but with a random uterus entirely made up of his own XY cells hanging around and herniating into inconvenient places.

I’m pretty sure BRC’s nemesis wasn’t thinking about this though.


While disorders of the gonads and sex chromosomes are responsible for primary hermaphrodism such as the conditions above, we need to move further along the chain of development to get to the next set. The vast majority of cells in the body differentiate by sex according to the hormones that influence them, produced from the gonads and adrenal glands, rather than because they have XX or XY chromosomes. If you give a normal XY man tons of estrogen, he will grow breasts in spite of his genes. During development, even the growth of a penis or vagina is under the control of hormones, not the chromosomes in the area's cells. Secondary hermaphrodism, therefore, occurs due to endocrine problems.

The most common of these is Complete Androgen Insensitivity Syndrome. This is a mutation in the gene responsible for the androgen receptor, such that the body’s cells do not respond to testosterone. Testes, therefore, initially develop, and internal female reproductive organs do not. The testosterone that the testes produce has no effect, however, and the individual develops a normal female external appearance, down to a vagina which ends in a blind sac, with no cervix or uterus to join onto. Since the normal feedback loops which inhibit testosterone formation are impaired, testosterone levels are actually at the upper end of normal for a man, but this is just aromatised into estrogen naturally by the body, and so has no effect. They enter female puberty without medical intervention, though lack of a uterus means they will never menstruate or be fertile. Patients usually present with a bilateral inguinal herniation of the testes. At puberty the testes try to descend, as they would in a normal male, but have nowhere to go. These are usually removed, which means the patient will require estrogen hormone replacement therapy. Some sources online claim that women with CAIS are stereotypically beautiful, as well as taller than the average woman. This does not seem to be the case, from the photos of patients I have seen they look fairly normal, and their height is not different from the female average.

People with CAIS are typically average women in terms of their psychological and sexual profile. I have found several claims that patients are actually less likely than the average woman to be masculine or gynephilic. Certainly this makes conceptual sense, they’re literally immune to testosterone, but the research into this seems lacking.

With XY chromosomes, testes that have been surgically removed, a blind ending vagina and hormone replacement therapy, CAIS individuals seem to me to similar in a number of important ways to male to female trans women. I imagine rather more people would consider the former as real women than the latter, and it may be useful for the reader to reflect on why that is.

If the mutation in the androgen receptor gene is less severe, the foetus may develop a more mixed intersex condition, ranging from a masculine appearance with micropenis, hypospadis (the penis opening in differing places to normal), up to a more female appearance, with masculine features such as an enlarged clitoris. Individuals with these more intermediate phenotypes generally have a higher rate of transgender identity, and higher rates of androphilia than the general male population. In addition, the more severe the insensitivity, the more feminine the individual's behaviour will be as a child.


Congenital adrenal hyperplasia is a serious condition which can occur in either XY or XX individuals, but for our purposes is only really interesting in the latter. The process is a little complicated, so I’ve tried to simplify it down into steps.
  •     Cortisol is often described as a stress hormone, and is produced by the adrenal glands.
  •     People with CAH have a mutation in one of several genes responsible for cortisol production, and so have lower levels of cortisol than average.
  •     These lower levels induce a feedback loop, and so the pituitary gland produces hormone called ACTH, which normally causes increased cortisol production.
  •     The adrenal gland cannot respond to this ACTH, but instead grows (hyperplasia), producing more testosterone as a result (among other problems).

This can occur to varying extent, but the general picture is an XX child who has received a massive overdose of testosterone. In the most extreme cases, the internal ovaries and uterus form normally, but the external genitalia can be very masculine. The clitoris can enlarge to penis size, with the urethra even opening from it. The labial folds can even appear as a scrotal sac (though obviously with no testes inside). How to raise such children is a controversial area, and depends very much on how masculine or feminine they appear at birth, but for the most part all but the most severe cases are initially assigned and raised as women.

Trigger warning on the following two pictures I guess, showing a megaclitoris, and a hypospadic penis with bifid scrotum. You've probably already scrolled down though...

We therefore have a situation where an XX individual, who is usually raised as a woman, has something approaching a ‘female penis’, entirely naturally. Alternatively, they may be deemed sufficiently virilised to be raised as a man, in which case we have someone who is considered by most people to be male, and yet has XX chromosomes, ovaries, and a uterus hidden inside.

In terms of their psychology, the literature is often complicated by not fully taking into account the severity of the virilisation and testosterone dosage. However, the overall picture is relatively clear. With regards their sexuality, around a third of women with CAH are at least partially gynephilic, compared to around 5% of the general population, implying a strong role for prenatal androgen exposure in developing a sexual attraction towards women. In terms of gender identity, there is a significant increase in the numbers who are severely conflicted, and who later transition to live as males, with reports of rates of between 1 in 20 and 1 in 3 (compared to around one in ten thousand in the general population) as well as generally a more male identification, even if severe transgender feelings are not present. As children, girls with CAH have more masculine behaviour and gender roles, playing in a more conventionally boyish way. XX patients with CAH may become more feminine across their lifespan, perhaps due to socialisation, or perhaps due to estrogen supplementation and treatment of their excess testosterone. The most severe cases are still more masculine in their characteristics than average women. 


Polycystic ovary syndrome is not really an intersex condition by any definition, but it does cause increased testosterone levels in women, and so has some of the same features as CAH. It is also very very common, so relatively easy to collect data for. Women with PCOS report more masculine behaviour as children, though this does not seem to have persisted into adulthood. There are also much higher rates in female to male transgender people. In this sample, 58% of FTM patients had PCOS, even prior to any hormonal intervention, compared to a population base rate in the region of 10%. Lesbians seem more likely to suffer from polycystic ovaries than do androphilic women, (there is some controversy over this, here the difference was present, but non-significant, and on a small sample).


Our final intersex condition is 5-alpha-reductase deficiency, better known as 'guevedoces' (meaning 'penis at twelve'). Those with it lack an enzyme necessary for converting testosterone into its more potent form in peripheral tissue. At birth, they appear normally female, but at puberty, when testosterone levels rise massively, they develop into men, testes descend, and their clitoris enlarges into something approaching a penis. Crucially, however, they had male typical levels of testosterone affecting the brain while in the womb.

Studying guevedoces is complicated by it being so rare, and so focused among small groups of people. Since there are barely any cases in the UK or USA, like for like comparisons ignoring unique cultural and genetic factors of the study population are difficult to make. Guevedoces is accepted as a relatively normal transition. The vast majority successfully make the transition from female to male at puberty, seem to have gynephilic attraction, and parents often say their children seemed more boyish prior to puberty.


I want to finish by talking about some evidence which can get us a better handle on how flexible or fixed gender identity, behavior, and sexuality is. In order to separate environmental and genetic influences on these things, ideally we'd take a bunch of girls and a bunch of boys and randomly assign them to be brought up as either their own gender, or the opposite. If gender, sexuality and behaviour were an environmental, learned feature, we would expect them to follow the group they are assigned to. If it were more biological, we would expect this to go rather badly.

No ethics committee would ever approve this, but there are a few natural experiments, and I'm not just talking about weirdos like this, and this from The Last Psychiatrist (PBUH). Most of these cases involve male children being raised as girls, either due to medical incompetence in cutting off their genitals, or due to a condition called cloacal extrophy. The outcomes for such patients were on the whole very bad. Doctors at the time were convinced socialisation could overcome nature, the most famous case being that of David Reimer, who identified as male from childhood, decided to live as a man, but tragically took his own life. More systematically, the picture is similar, all the female-reared children seem more markedly male in behaviour, and many ultimately declare themselves male of their own accord. Almost all have gynephilic sexual orientations. Overall, it seems more than half will transition back to living at their male birth gender, in spite of the obvious anatomical and social obstacles to doing so. Prenatal hormones can't be the entire story, but they're a massive part of it, and likely far more important than socialisation. Thankfully the practice of assigning male children as female in these situations is now no longer in fashion. Those assigned males have far better outcomes, even without a decent todger.


So what have we learned? Primarily, we have discovered that chromosomes are usually simple, but often messy, and they're not an especially good way of distinguishing men and women. Things like this are silly, in an overwhelming majority of cases, XX is female and XY is male, and that is a reasonable statement to make, but there are always a few exceptions. More than this, there probably isn't a single biological characteristic that is either necessary or sufficient for us to declare anyone a man or woman. When hormones and chromosomes conflict at full force, the hormones usually win. Hormones, especially in the womb during early development, are also almost certainly more important than chromosomes in determining people's behaviour, identity and sexuality [7]. Both gender roles and sexuality have strong hormonal components, and so while they may be in some sense social constructs, there is an underlying biological determination. This underlying biological determination isn't the entire story, environment plays a role, but some people are likely a lot more malleable than others. Anyone trying to force anyone to be anything they're not (even if they're trying to force them to not conform to stereotype) is probably going to be in for a rough ride. 

To the intersectionalistas, your gender isn't just an abstract social constuct, the way people respond to stimuli is just as much a product of hormones, genes and environment as your genitals are , and it's not as simple as saying sex and gender have nothing to do with one another.

To the TERFs, there are sex difference across all domains, including psychology and the brain. Though these are heavily bimodal, they're often not totally binary. These differences, though they may have a social learning component, are to a large degree innate, and there probably isn't much we can do to change them.


[1] Links are to relevant sources and papers. Some are open access, unfortunately most are not. Check my privilege. If you don't trust me, or simply want to learn more, I can provide pdfs on request, as long as you don't ask for them all, that'll take forever. Unfortunately many of these conditions are rare, so sample sizes are often small on these papers.

[2] Fittingly, the definition of intersex, which conditions it encompasses, and therefore prevalence, is somewhat controversial. Again, however, this is just a debate about categorisation and borders, rather than biology. For the sake of interest and completeness, I will talk about several disorders that are not always considered intersex. Please do not take this as representing any particular position on the appropriateness of the label in all cases.


I think the terms heterosexual and homosexual do a lot to hide the differences between how men and women behave. For that reason, and also because obviously we are dealing in complicated territory with regards the subject, I will try to refer to attraction as ‘androphile’ (attracted to men) and ‘gynephile’ (attracted to women.

[4] The X chromosome is much larger than the Y chromosome. For the most part only one X chromosome is active in a cell at any one time, so that there is not an excess of X chromosome genes expressed in women compared to men, which would cause problems. However, a number of genes are still expressed on the inactive chromosome, these are mainly genes that are also present on the Y chromosome, and so are not expressed in excess in women compared to men. If inactivation was normally total, Turner’s and Klinefelter’s would make no difference and cause no problems.

[5] XXX females and XYY males also exist, but these people have few notable clinical features and it is rarely discovered. The often quoted fact that XYY men are more violent is very unlikely to be true. There are no 0Y men. A fertilised egg with that genotype would not survive.

[6] If we take this 3/251 figure as reliable, we can add it to a general population estimate of 1/10,000 MTF trans and 1/1000 prevalence of Klinefelter’s to work out a rough estimate of about 1 in 800 XXY people being trans. This is more than a tenfold increase compared to the general population, though the vast majority of transwomen are not XXY and so it doesn’t explain a great deal of that.

[7] Declamare points out that, though incidences of gender nonconformity and trans identities may be more common in intersex people, on the whole most intersex people have a clear and normal gender identity one way or the other, mostly matching their hormonal environment in utero. This is worth emphasising if it is not already clear.

Friday, 14 August 2015


Evan Davis has caused a big hoohah by suggesting that gay men are more promiscuous than straight people.
This sounds like the kind of question we can answer with statistics! Now obviously the statement could be interpreted in different ways, but I've chosen to look at number of sexual partners. Simply looking at the amount of sex various people have is probably inadequate, since most sex presumably happens within relationships, which really isn't what he was talking about. Looking at how casual the sex people have might be worthwhile, but this is a trickier thing to get data on. 

Okcupid tried to answer the question this using their self report data, but this seems inadequate to me. Since you're making a public profile explicitly looking for a date, your answers are about as influenced by social desirability bias as is possible to imagine, as shown elsewhere on their own blog. The Daily Mirror also had a go last year, and got as far as finding out one paper exists, but couldn't be bothered doing it properly, since journalism is pretty hard.

Notes on the data I do have.
  • Natsal is the UK national sexual attitudes survey, it's pretty massive.
  • There are lots of other papers out there covering similar ground, but the ways they present the data is often not directly comparable, and often grouped together at the high end, when we really want medians and means for gay and straight men from the same survey, which we can get from the natsal.
  • For those of you unfamiliar with the terms, mean is the total number of sex partners divided by the number of people, while median is the number of sex partners where half the sample have had more and half the sample have had less. The mean is much higher in all cases here because the data is skewed, a small number of men have a LOT of sex, and this is reflected in the mean but not the median. It's therefore important to report both.
  • Natsal-3 is the most recent, but I can't find the exact data we want for gay men from that, so have gone back a decade to an earlier version to keep the comparisons valid. 
  • In order to get directly comparable data I've had to go for partners in the past five years, rather than lifetime. This shouldn't really make much difference, but the often claimed average of 6 is the median of straight lifetime partners from this dataset.
  • The term 'men who have sex with men' rather than gay is used in an attempt to keep things objective and avoid tricky issues around identity etc.
  • Women typically report fewer partners than men, so it is likely straight men are exaggerating the number of sexual partners they have had, or at least using slightly different definitions of what sex is, in spite of the researchers specifying oral, anal or vaginal intercourse. I've focused on figures for straight men but there are some for women in the linked papers if you're interested.
  • Results aren't broken down by age for MSM, so I'm comparing to all, but the mean age of the two samples is very similar (31.5 for MSM, 31 for non-MSM).
  •  I haven't looked into lesbian sex because I don't know what lesbians do.

Anyway, this is what we get -

Straight men over past 5 years - Mean 3.8 Median 1

MSM over past five years - Mean 24.1 Median 4

We also have this, which is analysing the same dataset, though comes out with slightly different numbers, I'm not sure why, but they're not massively different.

"Men with a homosexual partner during the past 5 years reported a mean (SD) of 26.7 (109.9) sex partners during that time, compared with 4.1 (8.3) among those men who reported having no homosexual partnerships."

The standout problem here, some of you may have noticed, is that because the 'gay' group are defined as men who have sex with men, there are none in that group who are celibate. This obviously skews our results a little. My cack-handed correction for this by excluding all straight men who haven't had sex in the last five years pushes their median up to 2 and the mean up to 4.2. This assumes the number of celibates is similar in both gay and straight men, which probably isn't a reasonable assumption, but I doubt it's incredibly far out and we'd need a huge number of celibate gays in order for this to affect the conclusion. We can pretty safely say that out of those that do have sex, gay men have more sex partners than straight men, and that Evan Davis is pretty right. It might be cruel to generalise, and assuming every gay man has tons and tons of sex with different guys is silly, but there is a notable difference in behaviour on average.

In addition, there does seem to be a far higher rate of nonmonogamy among gay men than among the population as a whole. General estimates of the number of open relationships are tricky to find, but seem to hover around 4% of couples generally while among gay men the number may be as high as around half.

The question, then, is why. Testosterone is associated with a higher sex drive in both men and women, and this must be a large part of the explanation of why men have a higher sex drive than women. Gay men don't seem to have much difference in testosterone levels though, so my guess would be that the higher number of sex partners is probably a result of gay relationships being men squared, rather than anything particular about gay men themselves. It's plausible that tinder as a social technology has allowed some straight men to reveal their preferences to act in a similar way, but I'm generally sceptical of most claims about the rise of hookup culture, and await data from the last couple of years in particular with interest.

As usual, All Stereotypes Are Correct.