Thanks for this post. It’s brave, thorough, fair, and well-researched—a breath of fresh air compared to 99% of internet discussion on this topic.
People seemed to appreciate it when I laid out my points of disagreement with the last post of this sort, so I’m going to try doing the same for this post. Feel free to let me know (including via PM) if you think it’s a bad idea for me to do this.
Like the last post, I think this post could benefit from less uncritical acceptance of social science research. That said, my sources aren’t any better. So my comment is just an attempt to present a coherent worldview—take it with a grain of salt.
I think this post underrates the degree to which effective altruists are likely to be unrepresentative of the population at large. In particular, all the EAs I’ve met are really smart. And all the discussion I see in online EA communities is really intelligent. I don’t think there’s any IQ data that backs up these observations directly. But the 2015 EA survey found that Less Wrong was the most popular way to discover EA, and survey data on Less Wrong users seems to indicate that the average IQ on Less Wrong is around 140. My impression is that Less Wrong is responsible for a lot of “founder effects” in the EA movement (the fraction of EAs from LW was even higher in the 2014 EA survey), and that being a magnet for high-IQ people goes along with this.
Why does this matter? Because the sexual behavior of high IQ people seems to be much different than the sexual behavior of the population at large. Check out this graph from this blog post. High IQ male teens are about 3x more likely to be virgins than average, and high IQ female teens are about 5x more likely.
Why is this? There are a few possible explanations. Personally, I suspect it’s a combination of smart people prioritizing other things, and having lower libidos. (Consider that only 20% of female graduate students at MIT masturbate, compared to 70% of the female population at large—source.)
You quote a study which found that frottage offenders targeted an average of nine hundred people each. That’s definitely an interesting stat, but I think you have to be extra careful when extrapolating from “institutionalized sex offenders… mandated to receive specialized treatment” to EAs, especially given that all the research I’ve heard about indicates that prisoner IQs tend to be well below average.
Furthermore, I would expect that sex offenders who actually get imprisoned are much more blatant than average. So that’s a reason why it could be unsafe to extrapolate data from imprisoned rapists to your “average” undetected rapist.
That all said, I believe you when you say there have been cases of sexual assault in EA, and I think you have good ideas about how to stop it. Here are some of my thoughts. In some cases, I’ll be emphasizing points you already made, because our worldviews overlap a fair amount.
I think that sexual assault prevention presents a different challenge relative to a lot of other causes EAs are interested in. We are used to seeing causes as bottlenecked on money or talent, but sexual assault strikes me as a cause bottlenecked on experimentation and tribalism. It’s a topic that’s difficult to have a two-sided conversation about due to the current political environment. People tend to retreat in to bubbles lead by the most motivated individuals, who are often extremists. Sometimes the extremists suggest interventions which may actually hurt their cause, because they don’t have good awareness of how people in other bubbles are thinking.
I was really impressed by how aware you seem to be of these issues. But, I think it’s important to recognize that your willingness to write a ~30 page essay on this topic is an indicator that sexual assault is something you’re unusually sensitive to. (That doesn’t mean you shouldn’t work on it; I’m glad it’s something you feel motivated to work on. I’m just saying you should keep this in mind.) Furthermore, I think it’s interesting to contrast your suicide stats with the fact that sexually harassed employees are only 1.63 times more likely to have turnover intentions. See also this article about women in Washington DC who prioritize their career over making sexual harassment allegations.
This isn’t meant to excuse sexual assault or sexual harassment. It’s meant to indicate that sexual harassment, sexual assault, and rape are offenses of much different magnitude. Similar to how shoving someone, punching someone so hard they get knocked out, and murdering someone are offenses of much different magnitude. Unfortunately, it seems to me that a lot of feminist discourse serves to lump these offenses of different magnitude together using terms like “rape culture”. Equivalently, imagine someone shoved another person during a bar dispute, and we condemned their action as indicative of “murder culture”.
Why does this matter? As you state in this post, our legal system doesn’t work that well for sex offenses. Our legal system works on the presumption of innocence and tries to gather evidence that a person committed a crime. But for sex offenses, there typically isn’t much evidence to gather.
Because the legal system doesn’t work that well, we have to fall back on the court of public opinion. My issue with a term like “rape culture” is that it pushes us closer to a world where a person receives the maximal sentence in the court of public opinion even for the most minor offense.
Why is this a world we don’t want to be in?
The punishment doesn’t fit the crime. This undermines the legitimacy of the effort to fight sex offenses.
In the traditional legal system, a person can commit a minor offense, pay their debt to society, and stop thinking of themselves as an offender. But if a person who commits a minor offense has no chance of re-integrating with society, that creates an incentive to attack the legitimacy of the entire system. See above.
It frames a minor sex offense in a way that’s likely to cause greater psychological harm. See also Lila’s comment.
Outrage is like antibiotics: The more it’s overused, the less well it works.
However, even if many women tolerate sexual aggressiveness well, I suspect the EA movement has an unusually high density of women who don’t. As I mentioned, high-IQ women seem to have much less sexual desire, and high-IQ people in general seem to be more anxious. So you may not be all that unrepresentative relative to the _EA_ population.
This ties in to the whole Red Pill/pickup topic. There’s a lot I could say about that, but one major point is that it is targeted at average-IQ women who like going to bars—sorority girls who say things like:
Don’t ask for permission for a first kiss. Dear lord, I hate that nonsene. Things are great and then whoa, I’ve got a pushover on my hands. No, you can’t kiss me, pussy. Goodbye.
Personally, I know of approximately 0 sorority girls in EA. And more broadly, I don’t see attitudes like this expressed frequently by women in the EA movement.
I could write another long comment on my issues with Red Pill ideas, and why I think feminist writing fails to engage productively with them. But I don’t have a ton of confidence in my views, and I’m also not sure how many guys who read that stuff will end up reading my comment. So for now I’ll just make a few points directed at those guys:
In the same way reversed stupidity is not intelligence, reversed censorship is not intelligence. Just because feminists furiously condemn red pill ideas using arguments that are bad does not mean that red pill ideas are correct.
In general, the epistemics of red pill discussion sites aren’t that great. Ideas are selected for based on whether they spread virally within the echo chamber rather than whether they’re factually supported. Reading something frequently enough can cause you to believe it’s true even if it isn’t. (This is how propaganda posters work.) If you want an evidence-based dating guide, prominent evolutionary psychologist (and effective altruist) Geoffrey Miller has a book and a website.
If sexual gratification is what you seek, there’s a lot of data showing that it’s easier for a man to get laid in environments with more women than men. Effective Altruism is not such an environment.
On “minor” sex offences vs. rape: even if the 900 acts per frottage figure is wrong, I suspect that acts like groping and frotteurism are far more common than rape. Because frottage offences are so much easier to commit, I believe that frottage offences are likely to be several times as common as rape. I would even go so far as to bet that we can agree on this point.
I didn’t find any information on whether frottage causes psychological trauma. My own observations tell me that frottage offences do cause psychological trauma, but it’s usually less intense.
So, if we were to compare the average rapist with the average groper, we might find that psychological trauma caused by the groper adds up. This might compare with the amount of trauma caused by a rapist. It might even exceed that amount. Maybe it does not even come close. I have no good way to tell since I did not find any studies on the amount of trauma caused by frottage offences.
Another important thing to note is that each time someone experiences psychological trauma, their sensitivity to future trauma increases. (I suspect I learned this from “The Body Remembers” but I’m not sure.)
So, it might be that if someone is groped ten times, by the tenth time, the experience is as bad as a rape. Maybe it would take a hundred times to get to that point. Maybe three. If I had to shoot from the hip, I’d give it a Fermi estimate between 10-100 times, but I really am not sure. Maybe sensitivity would never increase so much that one frottage offence would be as traumatic as a rape.
Also, and this is really important: the amount of trauma caused by a frottage offence varies greatly depending on context.
For a comparison: I was targeted by an acquaintance in a coat closet at a night club, by a co-worker on a job, and by a popular person at a party. I never had to see the guy from the night club again, so I just avoided the night club and felt fine. The co-worker caused me more problems. I became concerned about things happening to me in workplaces. That was context.
The popular person was far worse than either of those. I was sleep deprived due to jet lag, which a friend and I estimated amplified the trauma by about 3x. Also, this person is higher status than I am and tried smearing me. That situation was several times as traumatic as any of my previous experiences. I still have nightmares about this person sometimes.
Say you have a high status frottage offender, who is very prolific, who targets people who have just flown in and have jet lag, and who likes smearing his victims afterward. I think that a particularly destructive frottage offender of this description could easily do as much damage as a rapist.
Given my level of uncertainty about the psychological effects of frottage, and given the likelihood that frottage is far more common than rape, I am concerned about it. Therefore it was included. Perhaps I need to do a better job of explaining to everyone why I am so concerned about frottage offenders.
These are all good points. I find it totally plausible that some individuals are responsible for many assaults. I think this is a problem we should address. And I’m really sorry to hear about your experiences.
University student have an IQ of around 113. If we, as you suggest, use LessWrong as a proxy for the EA community the average EA IQ might be as high as 139. There’s still a difference there, just not as large as if the 6% figure was drawn from the population at large.
Effect of IQ on sexual behaviour is lower than that study suggests
That graph is unfortunately misleading, it’s from before the authors of the source paper (Halpern et. al, 2000) controlled for age, physical maturity and mother’s education. After taking these things into account the relative odds (compared to IQ 100) of a high IQ male teen being a non-virgin is 0.62 rather than 0.32 – quite a lot higher!
It’s worth being aware that Halpern et. al were studying adolescents between 7th and 12th grade. IQ scores don’t stabilise until later in life so there are a lot of questions left to be asked on how well this translates to the adult population. Perhaps the discrepancy here is due to high IQ teens being those who have developed their executive function faster than others?
Then there’s the question of how being more likely to be a virgin as a teen affects the odds of perpetrating sexual violence as an adult. For example: it might be that odds of perpetrating sexual violence are correlated with sexual frustration in teen years, in which case having a higher IQ might indirectly increase the odds of becoming a rapist!
(Also:
High IQ male teens are about 3x more likely to be virgins than average, and high IQ female teens are about 5x more likely.
I think you might have misread the graph? The graph you linked to is relative of odds of having had intercourse: so a male teen with average IQ is 3x more likely to have had sex, 5x for a female teen. As that graph is of relative odds so without a base rate there’s no way to to establish from those figures how much more likely a high IQ teen is to be a virgin than an average IQ one.)
On punishment and stigma: I think it would be better for everyone if we found a cure for sexual violence and persuaded all the offenders to use it. That said, I have no idea what the rest of the world will choose to do. I suggested two options which can scale globally, sting operations and researching a cure. I cannot leave out an option if it might succeed because that would be a failure of honesty, and because I think using either option is much better than letting them run amok.
I did note that finding a cure for sex offenders is probably more cost effective because most of the cost will be paid by the sex offenders themselves when they pay for their prescription and because I think it will scale better. Also, a cure can prevent offences from happening in the first place if people use it early on. Justice can only happen after harm has been done. A prevention method would get at more of the problem for this reason, and therefore has a chance to be more effective.
I think a cure is both more fair to the taxpayers who didn’t cause this problem and shouldn’t have to pay to fix it, and more fair toward offenders and non-offending paraphilia sufferers who did not choose to have their paraphilias—especially if they would choose to be cured if a cure were available.
I definitely relate to the desire for justice. I suspect that it reduces psychological trauma if justice is swift—like in that study about self defence which shows people experienced less trauma if they fought back, even if they didn’t win. I have occasionally had an opportunity to tell off an offender immediately after an offence, and that does seem to reduce the harm to me.
I don’t know if justice improves the survivor’s health days or weeks or years after the fact, but I suspect justice does help if it’s swift enough.
I think the desire for justice is healthy, though it can easily become twisted and go wayward. There are countless examples of a desire for justice going horribly wrong throughout history like the Salem witch hunts and the Spanish Inquisition. I try to avoid the sort of careless thinking which seems to be behind this sort of twisted behavior.
If people choose to use the justice approach, I don’t really know where people should draw the line, honestly, and this is because the whole mess is so complicated. Here are a few things I am sure of, in case sharing my perspective may be useful in some way:
If we have a credible reason to believe an offender has reformed and the risk they pose is average, there is no reason to insist that they accept a criminal label like “rapist”. There is no reason to pursue somebody who is of average risk when there are high risk people running amok. If they did the hard work to change, and to provide credible evidence that they are actually low risk, I would give them the basic acceptance they are seeking. (For instance: I would treat them like a human being when I see them around, though I’d be unlikely to invite them to my next sleep over.) That said, it is super hard for people to believe someone has reformed. I’m not sure if there is any type of evidence that is of high enough quality that it can be used for this purpose. Each individual person will make their own decisions about whether to trust someone who claims to have reformed, and some people will have much higher standards of evidence than others. At best, acceptance after reformation would be imperfect and it might be highly controversial.
If someone has not reformed, and we have every reason to believe they pose a risk, then we aught to take precautions, whether or not they were punished. This is sad, but punishment doesn’t actually cure sex offenders. Please check their recidivism rate. I know some people want to believe in “paying a debt to society” or “doing your time” but this is different from lowering risk. If someone raped someone, and they did nothing to reform, I don’t want them around me, and I think this is perfectly reasonable. Other people may have done some stuff to the rapist, and other people may have beliefs about repaying debts, but if other people did not cure the rapist, the rapist still poses a risk. I will protect myself, period.
I agree with all of this. Perhaps I’m too quick to extrapolate from my own experiences—I know that I’ve accidentally creeped out women in the past, and I always feel really bad about it afterwards, but this could be a bad mental model of the typical case.
Yeah. There are a lot of different people using a lot of different definitions of sex offender. There are definitions like the undetected rapist study I linked which sticks to such obvious and stereotypical behaviors that it leaves out at least half of the ways one can do obvious bad things (Example: why didn’t they ask about spiking drinks?). Then there are people who advocate asking for explicit consent for everything every time, starting with kissing. In practice, most of the people in my experience use things like context and body language to communicate about kissing rather than verbal consent. I have no idea how to resolve this mess of definitions. I guess people need to tell each other what consent philosophy they want to use in addition to stuff like sexual orientation. Maybe we need a norm of advertising our consent philosophy in prominent places the same way we do with gender, marital status and orientation.
Then, there’s the fact that most guys are not hit on by other guys, and have not seen what the range of behavior looks like. A lot of them are surprised that it’s very common for me to be asked things like whether I want to make out, whether I want to go home with him. I am coy rather than fast (referring to the distinctions Dawkins makes), so I can’t really understand this but it doesn’t bother me. I tend to assume those men are seeking fast women rather than relationships. Otherwise, I have no judgment. However, if some guy comes up and tells me to smile, my radar beeps and I want to recoil. Why? Because every guy who has ever said that to me has harassed the heck out of me afterward. I’ve been conditioned to hate it.
Plus, there’s this weird variety in pickup lit which includes everything from perfectly healthy self-confidence tips to explicit instructions to commit sex offences.
A lot of guys I know don’t have any idea what’s normal. Some of them are terrified of trying at all or have given up. It’s very sad.
I’m not fully aware of the male experience of this bizarre minefield of information. I might see only the tip of the ice berg. I can tell that it is a very confusing thing.
I really want to do something about this. I would benefit if all the men in my social network had a solid understanding of healthy boundaries. I think they would feel a lot less lost if they had that, too.
Maybe we need a norm of advertising our consent philosophy in prominent places the same way we do with gender, marital status and orientation.
Here is an interesting thread in the slatestarcodex subreddit where some possible problems with this idea were discussed (ctrl-f “public symbol”).
The simplest way to do this would be on the basis of whether a woman wears revealing clothes. Unfortunately, there’s something of a taboo among feminists regarding this approach, because if you say anything about how a woman wore revealing clothes, feminists round that off to victim blaming and condemn you on that basis.
What are your thoughts on what needs to happen?
Thanks for asking! Some brainstorming:
I like your centralized reporting idea. Julia Wise says she has served as a contact point, and CFAR recently added a community disputes council which does this, among other things. (If you’re unable to see the post I linked, you can get in touch with the CFAR people here.) Getting all these people to make entries in a shared database seems good. (BTW, how many people reached out to you since you made this post and shared your anonymous feedback link?)
Try to understand why people don’t want to report sexual assaults. Would they be willing to report them in the context of the EA survey or some other anonymous survey? But also, would allowing for anonymous accusations this way present its own set of problems? Maybe if someone gets more than one anonymous accusation from different sources, then they should get quietly kicked out of the community (“quietly”=in a way that doesn’t harm their reputation, because these anonymous accusations are unverifiable—so we are kicking them out in order to mitigate risk, not because we think they should be condemned).
Instead of doing expensive social norm experimentation ourselves, maybe survey the anthropology literature and try to understand whether some societies have lower rates of rape/sexual assault than others, and what those societies have in common culturally.
I personally don’t really think the EA community should be seen as a place for dating/relationships—the same way workplaces are not seen this way. So I’m happy for the EA community standards to trade off decreased relationship formation in favor of increased safety. If you really want to date other EAs, there’s always reciprocity.io.
However, in society at large, I think it’d be good for feminists and women who create incentives for sexually aggressive behavior—e.g. 50 Shades of Grey fans—to dialogue more and come to some kind of compromise position. Right now the 50 Shades fans are nodding along with the feminists and then quietly rewarding guys who push boundaries.
I also think it’d be good for men to get positive examples of behavior that women consider OK or even appreciated, instead of only hearing about behaviors we shouldn’t do.
I get the impression feminists think they can solve sexual assault by making the threshold for what constitutes sexual assault ever lower. I think this hurts their credibility. Maybe there are some guys who take the feminist guideline and always adjust it some distance towards being more aggressive, but then there are other guys who want a decent-sized safety buffer between what we do and what gets you condemned. Learning that you are a monster if you have any sexual desire for women can also cause suicidality.
So maybe it’d be useful if there was a group that tried to put together a set of common-sense rules by surveying a representative population of women about what bothered them, and having a group of reasonable and fairly randomly sampled men & women engage in dialogue (with an equal number of both men and women in the group, and without penalizing people for saying uncomfortable stuff and expressing “Red Pill” sentiments—maybe by making it an anonymous online discussion or something. BTW in general, I think solving tribalism and helping people who disagree on these issues to engage productively would be good, e.g. you and Marcus in this thread). Then you could release your recommendations, and if someone violated them, it would hopefully create less drama to take action against the violator (because a broader set of people would be able to agree that the guidelines were reasonable and action is therefore justified if someone violates them). And if the level of drama decreased, then maybe women would be more willing to make accusations, and then we’d be able to get rid of the real bad apples more effectively.
The point is that if accusers are always being seen as part of the feminist camp, which lumps them in with some people saying some pretty crazy and unreasonable stuff, then that’s going to make them unpopular whenever they make an accusation, which is going to lead fewer women to be willing to make accusations. So another thing that might work is for more accusers to be explicit about which parts of feminist discourse they disagree with. If they seem like reasonable people, maybe their accusations will be taken more seriously, less drama will be generated, and making an accusation won’t make you unpopular.
I’m also not sure the comparison is fair. Workplaces have specific regulations about how people can interact; an EA meetup is not your workplace and attendees probably shouldn’t be held to the same standard of conduct.
I also know of a number of happy EA relationships. I think it would be a shame if we decreed that they were off-limits.
I know about the replication crisis, I’ve read “Statistics Done Wrong” and I’ve read some Ioannidis. Perhaps I was too subtle, my way of addressing these concerns was to load up on as many review articles and meta-analyses as I could find, in all the areas where there was enough research for me to do so. In other areas, I looked for as many studies as I could find and included them all.
This is not perfect either. Ioannidis has warned about some specific vulnerabilities in meta-analyses and review articles. There isn’t something perfect for me to do. I could have chosen to do nothing because the research is flawed. I decided that the subject is too important to ignore and I made the best of it.
Thanks for this post. It’s brave, thorough, fair, and well-researched—a breath of fresh air compared to 99% of internet discussion on this topic.
I have seen several responses saying things like this, but in reality, the research in this article goes only as far as collecting the standard feminist narrative on sexual assault, which is not original and can be found in many places if you are familiar with this subject. The only thing that’s new is attempting to marry this perspective to EA, despite the methodology being highly partisan and significantly different from EA methodology.
Among the problems:
uncritically taking feminist sexual assault prevalence research at face value, without addressing the many methodological criticisms
mixing and matching studies with very different populations and methods to maximize the perception of male perpetration and female victimhood
failing to address the debate on false accusations
failing to discover any of the complicating lines of research, such as the token resistance research, and the research on self-justification and unreliability of memory
discussing drastic and hasty interventions like stings and medicalization
accusing an entire community of males of containing hundreds of rapists based on poorly-executed studies on a totally different population
While the article is being criticized for being too long, in some ways, it’s actually too short to support the extraordinary claims it is making, which are perhaps not fully recognized as extraordinary due to feminist research not being held to the same standards as other fields, and all sorts of glaring errors being normalized.
Here is my detailed rebuttal which is only enough to cover some of those problems, and here is an additional comment on the lack of balance in the original post.
I plan to write a new post about estimating the number of sexually violent people in EA. I have a large number of specific concerns about biasing our estimate. For instance, a lot of people commented or messaged me saying that the estimate was too high, but they didn’t incorporate any of the information which would actually increase the estimate. They only included information that would decrease the estimate. There are a lot of other ways we could go wrong with adjusting this estimate. That’s part of why my estimate is so simple. If I adjust it at all, I could easily be introducing biases.
I will invite the whole EA community to provide specific references that are related. I don’t know that this would decrease the chance of bias. It might increase the chance of bias. However, with such a post, one will at least see how complicated it is, think about whether one’s perspective is biased, and hopefully start compensating for whatever biases are present.
Why do you feel it’s important to have a more accurate guess regarding the number of sexually violent people in EA? I’m in favor of trying to measure the rate of sexual assault using e.g. the EA survey, because that is a metric we can track in order to measure whether things are improving. (Ideally using a question such as “Were you assaulted in the past year?”, so our metric will be responsive year over year.) But it seems to me that time spent refining our guess based on priors would be better spent implementing measures to reduce sexual assault.
I don’t think we can have an accurate idea of how much sexual assault is happening in EA without a separate high-quality survey. This is because there are so many definitions of sexual violence which contradict one another, that to ensure an accurate picture of what’s going on, we’d have to wrangle with definitions for a long time—and we’d end up asking a set of questions, not just one question.
I’d love to see a yearly undetected sex offender survey given to both men and women regarding how much sexual violence they committed against EAs in the last year, and a yearly sexual violence survey given to both men and women to ask how much sexual violence they received from EAs in the last year. If they added this to the yearly survey that would be awesome!
Then we’d have a way to track progress, and that’s important. The survey would have to be designed very carefully from the beginning though.
The reason I want to write a separate article about the number of sex offenders in EA is because it appears quite controversial. If we can get closer to having a consensus on sexual violence related matters, I think this will make us more effective at reducing it. The purpose of the article is not to create a more accurate number. I’m not even sure that’s possible. The purpose of the article is to address the controversy, explore the complexities, and encourage people to compensate for the various biases that may be interfering.
Edit: I’ve set up a collaboration with the yearly EA survey team!
I don’t think we can have an accurate idea of how much sexual assault is happening in EA without a separate high-quality survey. This is because there are so many definitions of sexual violence which contradict one another, that to ensure an accurate picture of what’s going on, we’d have to wrangle with definitions for a long time—and we’d end up asking a set of questions, not just one question.
I’d love to see a yearly undetected sex offender survey given to both men and women regarding how much sexual violence they committed against EAs in the last year, and a yearly sexual violence survey given to both men and women to ask how much sexual violence they received from EAs in the last year. If they added this to the yearly survey that would be awesome!
I think there’s a tradeoff here. If this is created as a second, separate survey, there will likely be selection effects in who chooses to take it. I expect people who are more concerned about the problem of sexual assault (such as people who have been sexually assaulted) will be more likely to complete a survey that’s specifically about sexual assault. Given these selection effects, I suspect it’s best to settle on a relatively brief measure and include it in the main survey.
Brainstorming on what to include in that measure:
One idea is to just ask people “were you sexually assaulted” and let them use their own definition. After all, our goal is to reduce psychological trauma. If someone’s experience met some technical definition of sexual assault, but it didn’t bother them very much, maybe it’s not something we need to worry about.
In a memo that has now been signed by about 70 institute members and advisers, including Judge Gertner, readers have been asked to consider the following scenario: “Person A and Person B are on a date and walking down the street. Person A, feeling romantically and sexually attracted, timidly reaches out to hold B’s hand and feels a thrill as their hands touch. Person B does nothing, but six months later files a criminal complaint. Person A is guilty of ‘Criminal Sexual Contact’ under proposed Section 213.6(3)(a).”
Far-fetched? Not as the draft is written. The hypothetical crime cobbles together two of the draft’s key concepts. The first is affirmative consent. The second is an enlarged definition of criminal sexual contact that would include the touching of any body part, clothed or unclothed, with sexual gratification in mind. As the authors of the model law explain: “Any kind of contact may qualify. There are no limits on either the body part touched or the manner in which it is touched.” So if Person B neither invites nor rebukes a sexual advance, then anything that happens afterward is illegal. “With passivity expressly disallowed as consent,” the memo says, “the initiator quickly runs up a string of offenses with increasingly more severe penalties to be listed touch by touch and kiss by kiss in the criminal complaint.”
Source. I don’t think using a broad technical definition like this would be very useful, but a narrow technical definition of rape seems like it could be pretty useful to measure.
This blog post makes the case for vague rules like “don’t be a jerk” and “don’t be creepy”. Maybe that could make a good survey question: “Did you get creeped out by another EA in the past year? How creeped out were you on a scale of 1 to 10? Here’s a rubric.” I actually think a measure like this could be less controversial than trying to precisely define sexual assault. Hopefully even the most fraternity brother-ish of EAs can recognize the case for not creeping chicks out. (Similarly, having a central registry that tells people things like “a lot of people are getting creeped out by you” seems like it could maybe work better than trying to define what exactly constitutes “assault”—it frames the problem as something you’d like to become aware of and fix, like having body odor, as opposed to grounds for ostracization? Of course ostracization is in fact justified in some cases—I’m just thinking aloud here.)
I suppose one issue with these measures is that they will fluctuate depending on the presence/absence of highly sensitive people in the movement. Overall, I’m much more comfortable using measures like these as indicators for what we should prioritize internally vs an overall measure of the moral worth of the movement. In other words, maybe we should not make them public? I don’t know.
It might also be interesting to include a measure of how many women in EA would like men in EA to be more direct and sexually assertive with them—see this comment.
The reason I want to write a separate article about the number of sex offenders in EA is because it appears quite controversial. If we can get closer to having a consensus on sexual violence related matters, I think this will make us more effective at reducing it. The purpose of the article is not to create a more accurate number. I’m not even sure that’s possible. The purpose of the article is to address the controversy, explore the complexities, and encourage people to compensate for the various biases that may be interfering.
Well, as a man in EA, I don’t like the idea of people thinking of me as a possible sex offender—especially if I’m not, in fact, a sex offender. And whenever you try to estimate how X men in EA are sex offenders, no matter what X you tell people, you’ve framed things in a way that is gonna make people see me as a possible sex offender. So maybe that’s why you got some pushback on that statement.
I’m happy for us to do a survey to measure sex offenses, because that will give us a way to actually measure and fight the problem. I know that any nonzero number that survey finds is going to reflect poorly on me, a man in EA, even if the number is much lower than we’d expect on base rates, because of the framing effects I discussed. However, I am willing to pay that cost because I care about addressing the problem of sexual assault. But I think trying to make an estimate based on prior information will just stir people up.
Edit: I’ve set up a collaboration with the yearly EA survey team!
What I was envisioning was a whole section within the survey where multiple questions about sexual violence are asked. For whatever reason, I described this using the word “separate”. That’s not actually what I was trying to suggest. I agree that if the questions are separated, there will probably be some bias.
If we use a definition that is vague, a lot of people will ignore the survey results. They’ll assume that a lot of what was reported is stuff they wouldn’t agree is a sexual assault. Therefore, specific definitions are needed. Ideally, I would like to see a set of specific definitions that a lot of people agree are sexual assault, and that cover a broad range of types.
To make sure the questions are relevant to the goals, I think there should be questions about things like whether the sexual harassment resulted in psychological harm, suicidal behavior, or intentions to leave the workplace or movement. I’d also like to see questions about whether sexual assaults are happening at work, EA events, etc. Depending on how well anonymized the survey is, we may or may not get answers to these sorts of questions.
Without knowing the limit to the number of questions we can add, there’s no point in discussing what should be asked. We would just waste time optimizing for the wrong trade off between detail and brevity. Also, it would be good to get some perspectives from people who do research in related areas. I’m going to hold off on investing time into planning until I have had a collaboration with the survey team.
Thanks for this post. It’s brave, thorough, fair, and well-researched—a breath of fresh air compared to 99% of internet discussion on this topic.
People seemed to appreciate it when I laid out my points of disagreement with the last post of this sort, so I’m going to try doing the same for this post. Feel free to let me know (including via PM) if you think it’s a bad idea for me to do this.
Like the last post, I think this post could benefit from less uncritical acceptance of social science research. That said, my sources aren’t any better. So my comment is just an attempt to present a coherent worldview—take it with a grain of salt.
I think this post underrates the degree to which effective altruists are likely to be unrepresentative of the population at large. In particular, all the EAs I’ve met are really smart. And all the discussion I see in online EA communities is really intelligent. I don’t think there’s any IQ data that backs up these observations directly. But the 2015 EA survey found that Less Wrong was the most popular way to discover EA, and survey data on Less Wrong users seems to indicate that the average IQ on Less Wrong is around 140. My impression is that Less Wrong is responsible for a lot of “founder effects” in the EA movement (the fraction of EAs from LW was even higher in the 2014 EA survey), and that being a magnet for high-IQ people goes along with this.
Why does this matter? Because the sexual behavior of high IQ people seems to be much different than the sexual behavior of the population at large. Check out this graph from this blog post. High IQ male teens are about 3x more likely to be virgins than average, and high IQ female teens are about 5x more likely.
Why is this? There are a few possible explanations. Personally, I suspect it’s a combination of smart people prioritizing other things, and having lower libidos. (Consider that only 20% of female graduate students at MIT masturbate, compared to 70% of the female population at large—source.)
You quote a study which found that frottage offenders targeted an average of nine hundred people each. That’s definitely an interesting stat, but I think you have to be extra careful when extrapolating from “institutionalized sex offenders… mandated to receive specialized treatment” to EAs, especially given that all the research I’ve heard about indicates that prisoner IQs tend to be well below average.
Furthermore, I would expect that sex offenders who actually get imprisoned are much more blatant than average. So that’s a reason why it could be unsafe to extrapolate data from imprisoned rapists to your “average” undetected rapist.
That all said, I believe you when you say there have been cases of sexual assault in EA, and I think you have good ideas about how to stop it. Here are some of my thoughts. In some cases, I’ll be emphasizing points you already made, because our worldviews overlap a fair amount.
I think that sexual assault prevention presents a different challenge relative to a lot of other causes EAs are interested in. We are used to seeing causes as bottlenecked on money or talent, but sexual assault strikes me as a cause bottlenecked on experimentation and tribalism. It’s a topic that’s difficult to have a two-sided conversation about due to the current political environment. People tend to retreat in to bubbles lead by the most motivated individuals, who are often extremists. Sometimes the extremists suggest interventions which may actually hurt their cause, because they don’t have good awareness of how people in other bubbles are thinking.
I was really impressed by how aware you seem to be of these issues. But, I think it’s important to recognize that your willingness to write a ~30 page essay on this topic is an indicator that sexual assault is something you’re unusually sensitive to. (That doesn’t mean you shouldn’t work on it; I’m glad it’s something you feel motivated to work on. I’m just saying you should keep this in mind.) Furthermore, I think it’s interesting to contrast your suicide stats with the fact that sexually harassed employees are only 1.63 times more likely to have turnover intentions. See also this article about women in Washington DC who prioritize their career over making sexual harassment allegations.
This isn’t meant to excuse sexual assault or sexual harassment. It’s meant to indicate that sexual harassment, sexual assault, and rape are offenses of much different magnitude. Similar to how shoving someone, punching someone so hard they get knocked out, and murdering someone are offenses of much different magnitude. Unfortunately, it seems to me that a lot of feminist discourse serves to lump these offenses of different magnitude together using terms like “rape culture”. Equivalently, imagine someone shoved another person during a bar dispute, and we condemned their action as indicative of “murder culture”.
Why does this matter? As you state in this post, our legal system doesn’t work that well for sex offenses. Our legal system works on the presumption of innocence and tries to gather evidence that a person committed a crime. But for sex offenses, there typically isn’t much evidence to gather.
Because the legal system doesn’t work that well, we have to fall back on the court of public opinion. My issue with a term like “rape culture” is that it pushes us closer to a world where a person receives the maximal sentence in the court of public opinion even for the most minor offense.
Why is this a world we don’t want to be in?
The punishment doesn’t fit the crime. This undermines the legitimacy of the effort to fight sex offenses.
In the traditional legal system, a person can commit a minor offense, pay their debt to society, and stop thinking of themselves as an offender. But if a person who commits a minor offense has no chance of re-integrating with society, that creates an incentive to attack the legitimacy of the entire system. See above.
It frames a minor sex offense in a way that’s likely to cause greater psychological harm. See also Lila’s comment.
Outrage is like antibiotics: The more it’s overused, the less well it works.
However, even if many women tolerate sexual aggressiveness well, I suspect the EA movement has an unusually high density of women who don’t. As I mentioned, high-IQ women seem to have much less sexual desire, and high-IQ people in general seem to be more anxious. So you may not be all that unrepresentative relative to the _EA_ population.
This ties in to the whole Red Pill/pickup topic. There’s a lot I could say about that, but one major point is that it is targeted at average-IQ women who like going to bars—sorority girls who say things like:
Personally, I know of approximately 0 sorority girls in EA. And more broadly, I don’t see attitudes like this expressed frequently by women in the EA movement.
I could write another long comment on my issues with Red Pill ideas, and why I think feminist writing fails to engage productively with them. But I don’t have a ton of confidence in my views, and I’m also not sure how many guys who read that stuff will end up reading my comment. So for now I’ll just make a few points directed at those guys:
In the same way reversed stupidity is not intelligence, reversed censorship is not intelligence. Just because feminists furiously condemn red pill ideas using arguments that are bad does not mean that red pill ideas are correct.
In general, the epistemics of red pill discussion sites aren’t that great. Ideas are selected for based on whether they spread virally within the echo chamber rather than whether they’re factually supported. Reading something frequently enough can cause you to believe it’s true even if it isn’t. (This is how propaganda posters work.) If you want an evidence-based dating guide, prominent evolutionary psychologist (and effective altruist) Geoffrey Miller has a book and a website.
If sexual gratification is what you seek, there’s a lot of data showing that it’s easier for a man to get laid in environments with more women than men. Effective Altruism is not such an environment.
I agree with this for the most part, but let’s not exclude people from EA who, like me, are low-IQ and high-libido.
On “minor” sex offences vs. rape: even if the 900 acts per frottage figure is wrong, I suspect that acts like groping and frotteurism are far more common than rape. Because frottage offences are so much easier to commit, I believe that frottage offences are likely to be several times as common as rape. I would even go so far as to bet that we can agree on this point.
I didn’t find any information on whether frottage causes psychological trauma. My own observations tell me that frottage offences do cause psychological trauma, but it’s usually less intense.
So, if we were to compare the average rapist with the average groper, we might find that psychological trauma caused by the groper adds up. This might compare with the amount of trauma caused by a rapist. It might even exceed that amount. Maybe it does not even come close. I have no good way to tell since I did not find any studies on the amount of trauma caused by frottage offences.
Another important thing to note is that each time someone experiences psychological trauma, their sensitivity to future trauma increases. (I suspect I learned this from “The Body Remembers” but I’m not sure.)
So, it might be that if someone is groped ten times, by the tenth time, the experience is as bad as a rape. Maybe it would take a hundred times to get to that point. Maybe three. If I had to shoot from the hip, I’d give it a Fermi estimate between 10-100 times, but I really am not sure. Maybe sensitivity would never increase so much that one frottage offence would be as traumatic as a rape.
Also, and this is really important: the amount of trauma caused by a frottage offence varies greatly depending on context.
For a comparison: I was targeted by an acquaintance in a coat closet at a night club, by a co-worker on a job, and by a popular person at a party. I never had to see the guy from the night club again, so I just avoided the night club and felt fine. The co-worker caused me more problems. I became concerned about things happening to me in workplaces. That was context.
The popular person was far worse than either of those. I was sleep deprived due to jet lag, which a friend and I estimated amplified the trauma by about 3x. Also, this person is higher status than I am and tried smearing me. That situation was several times as traumatic as any of my previous experiences. I still have nightmares about this person sometimes.
Say you have a high status frottage offender, who is very prolific, who targets people who have just flown in and have jet lag, and who likes smearing his victims afterward. I think that a particularly destructive frottage offender of this description could easily do as much damage as a rapist.
Given my level of uncertainty about the psychological effects of frottage, and given the likelihood that frottage is far more common than rape, I am concerned about it. Therefore it was included. Perhaps I need to do a better job of explaining to everyone why I am so concerned about frottage offenders.
These are all good points. I find it totally plausible that some individuals are responsible for many assaults. I think this is a problem we should address. And I’m really sorry to hear about your experiences.
You make an interesting point here that I thought was worth digging into a little bit.
Difference in IQ between the self-reported rapists and the the community is lower than you might think
The 6% self-reported rapist figure used in the above is from a paper that surveys university students.
University student have an IQ of around 113. If we, as you suggest, use LessWrong as a proxy for the EA community the average EA IQ might be as high as 139. There’s still a difference there, just not as large as if the 6% figure was drawn from the population at large.
Effect of IQ on sexual behaviour is lower than that study suggests
That graph is unfortunately misleading, it’s from before the authors of the source paper (Halpern et. al, 2000) controlled for age, physical maturity and mother’s education. After taking these things into account the relative odds (compared to IQ 100) of a high IQ male teen being a non-virgin is 0.62 rather than 0.32 – quite a lot higher!
(The paper itself is behind a paywall so for reference I’ve uploaded the controlled graph for male teens to Imagebucket.)
It’s worth being aware that Halpern et. al were studying adolescents between 7th and 12th grade. IQ scores don’t stabilise until later in life so there are a lot of questions left to be asked on how well this translates to the adult population. Perhaps the discrepancy here is due to high IQ teens being those who have developed their executive function faster than others?
Then there’s the question of how being more likely to be a virgin as a teen affects the odds of perpetrating sexual violence as an adult. For example: it might be that odds of perpetrating sexual violence are correlated with sexual frustration in teen years, in which case having a higher IQ might indirectly increase the odds of becoming a rapist!
(Also:
I think you might have misread the graph? The graph you linked to is relative of odds of having had intercourse: so a male teen with average IQ is 3x more likely to have had sex, 5x for a female teen. As that graph is of relative odds so without a base rate there’s no way to to establish from those figures how much more likely a high IQ teen is to be a virgin than an average IQ one.)
On punishment and stigma: I think it would be better for everyone if we found a cure for sexual violence and persuaded all the offenders to use it. That said, I have no idea what the rest of the world will choose to do. I suggested two options which can scale globally, sting operations and researching a cure. I cannot leave out an option if it might succeed because that would be a failure of honesty, and because I think using either option is much better than letting them run amok.
I did note that finding a cure for sex offenders is probably more cost effective because most of the cost will be paid by the sex offenders themselves when they pay for their prescription and because I think it will scale better. Also, a cure can prevent offences from happening in the first place if people use it early on. Justice can only happen after harm has been done. A prevention method would get at more of the problem for this reason, and therefore has a chance to be more effective.
I think a cure is both more fair to the taxpayers who didn’t cause this problem and shouldn’t have to pay to fix it, and more fair toward offenders and non-offending paraphilia sufferers who did not choose to have their paraphilias—especially if they would choose to be cured if a cure were available.
I definitely relate to the desire for justice. I suspect that it reduces psychological trauma if justice is swift—like in that study about self defence which shows people experienced less trauma if they fought back, even if they didn’t win. I have occasionally had an opportunity to tell off an offender immediately after an offence, and that does seem to reduce the harm to me.
I don’t know if justice improves the survivor’s health days or weeks or years after the fact, but I suspect justice does help if it’s swift enough.
I think the desire for justice is healthy, though it can easily become twisted and go wayward. There are countless examples of a desire for justice going horribly wrong throughout history like the Salem witch hunts and the Spanish Inquisition. I try to avoid the sort of careless thinking which seems to be behind this sort of twisted behavior.
If people choose to use the justice approach, I don’t really know where people should draw the line, honestly, and this is because the whole mess is so complicated. Here are a few things I am sure of, in case sharing my perspective may be useful in some way:
If we have a credible reason to believe an offender has reformed and the risk they pose is average, there is no reason to insist that they accept a criminal label like “rapist”. There is no reason to pursue somebody who is of average risk when there are high risk people running amok. If they did the hard work to change, and to provide credible evidence that they are actually low risk, I would give them the basic acceptance they are seeking. (For instance: I would treat them like a human being when I see them around, though I’d be unlikely to invite them to my next sleep over.) That said, it is super hard for people to believe someone has reformed. I’m not sure if there is any type of evidence that is of high enough quality that it can be used for this purpose. Each individual person will make their own decisions about whether to trust someone who claims to have reformed, and some people will have much higher standards of evidence than others. At best, acceptance after reformation would be imperfect and it might be highly controversial.
If someone has not reformed, and we have every reason to believe they pose a risk, then we aught to take precautions, whether or not they were punished. This is sad, but punishment doesn’t actually cure sex offenders. Please check their recidivism rate. I know some people want to believe in “paying a debt to society” or “doing your time” but this is different from lowering risk. If someone raped someone, and they did nothing to reform, I don’t want them around me, and I think this is perfectly reasonable. Other people may have done some stuff to the rapist, and other people may have beliefs about repaying debts, but if other people did not cure the rapist, the rapist still poses a risk. I will protect myself, period.
I agree with all of this. Perhaps I’m too quick to extrapolate from my own experiences—I know that I’ve accidentally creeped out women in the past, and I always feel really bad about it afterwards, but this could be a bad mental model of the typical case.
Yeah. There are a lot of different people using a lot of different definitions of sex offender. There are definitions like the undetected rapist study I linked which sticks to such obvious and stereotypical behaviors that it leaves out at least half of the ways one can do obvious bad things (Example: why didn’t they ask about spiking drinks?). Then there are people who advocate asking for explicit consent for everything every time, starting with kissing. In practice, most of the people in my experience use things like context and body language to communicate about kissing rather than verbal consent. I have no idea how to resolve this mess of definitions. I guess people need to tell each other what consent philosophy they want to use in addition to stuff like sexual orientation. Maybe we need a norm of advertising our consent philosophy in prominent places the same way we do with gender, marital status and orientation.
Then, there’s the fact that most guys are not hit on by other guys, and have not seen what the range of behavior looks like. A lot of them are surprised that it’s very common for me to be asked things like whether I want to make out, whether I want to go home with him. I am coy rather than fast (referring to the distinctions Dawkins makes), so I can’t really understand this but it doesn’t bother me. I tend to assume those men are seeking fast women rather than relationships. Otherwise, I have no judgment. However, if some guy comes up and tells me to smile, my radar beeps and I want to recoil. Why? Because every guy who has ever said that to me has harassed the heck out of me afterward. I’ve been conditioned to hate it.
Plus, there’s this weird variety in pickup lit which includes everything from perfectly healthy self-confidence tips to explicit instructions to commit sex offences.
A lot of guys I know don’t have any idea what’s normal. Some of them are terrified of trying at all or have given up. It’s very sad.
I’m not fully aware of the male experience of this bizarre minefield of information. I might see only the tip of the ice berg. I can tell that it is a very confusing thing.
I really want to do something about this. I would benefit if all the men in my social network had a solid understanding of healthy boundaries. I think they would feel a lot less lost if they had that, too.
What are your thoughts on what needs to happen?
Here is an interesting thread in the slatestarcodex subreddit where some possible problems with this idea were discussed (ctrl-f “public symbol”).
The simplest way to do this would be on the basis of whether a woman wears revealing clothes. Unfortunately, there’s something of a taboo among feminists regarding this approach, because if you say anything about how a woman wore revealing clothes, feminists round that off to victim blaming and condemn you on that basis.
Thanks for asking! Some brainstorming:
I like your centralized reporting idea. Julia Wise says she has served as a contact point, and CFAR recently added a community disputes council which does this, among other things. (If you’re unable to see the post I linked, you can get in touch with the CFAR people here.) Getting all these people to make entries in a shared database seems good. (BTW, how many people reached out to you since you made this post and shared your anonymous feedback link?)
Try to understand why people don’t want to report sexual assaults. Would they be willing to report them in the context of the EA survey or some other anonymous survey? But also, would allowing for anonymous accusations this way present its own set of problems? Maybe if someone gets more than one anonymous accusation from different sources, then they should get quietly kicked out of the community (“quietly”=in a way that doesn’t harm their reputation, because these anonymous accusations are unverifiable—so we are kicking them out in order to mitigate risk, not because we think they should be condemned).
Instead of doing expensive social norm experimentation ourselves, maybe survey the anthropology literature and try to understand whether some societies have lower rates of rape/sexual assault than others, and what those societies have in common culturally.
I personally don’t really think the EA community should be seen as a place for dating/relationships—the same way workplaces are not seen this way. So I’m happy for the EA community standards to trade off decreased relationship formation in favor of increased safety. If you really want to date other EAs, there’s always reciprocity.io.
However, in society at large, I think it’d be good for feminists and women who create incentives for sexually aggressive behavior—e.g. 50 Shades of Grey fans—to dialogue more and come to some kind of compromise position. Right now the 50 Shades fans are nodding along with the feminists and then quietly rewarding guys who push boundaries.
I also think it’d be good for men to get positive examples of behavior that women consider OK or even appreciated, instead of only hearing about behaviors we shouldn’t do.
I get the impression feminists think they can solve sexual assault by making the threshold for what constitutes sexual assault ever lower. I think this hurts their credibility. Maybe there are some guys who take the feminist guideline and always adjust it some distance towards being more aggressive, but then there are other guys who want a decent-sized safety buffer between what we do and what gets you condemned. Learning that you are a monster if you have any sexual desire for women can also cause suicidality.
So maybe it’d be useful if there was a group that tried to put together a set of common-sense rules by surveying a representative population of women about what bothered them, and having a group of reasonable and fairly randomly sampled men & women engage in dialogue (with an equal number of both men and women in the group, and without penalizing people for saying uncomfortable stuff and expressing “Red Pill” sentiments—maybe by making it an anonymous online discussion or something. BTW in general, I think solving tribalism and helping people who disagree on these issues to engage productively would be good, e.g. you and Marcus in this thread). Then you could release your recommendations, and if someone violated them, it would hopefully create less drama to take action against the violator (because a broader set of people would be able to agree that the guidelines were reasonable and action is therefore justified if someone violates them). And if the level of drama decreased, then maybe women would be more willing to make accusations, and then we’d be able to get rid of the real bad apples more effectively.
The point is that if accusers are always being seen as part of the feminist camp, which lumps them in with some people saying some pretty crazy and unreasonable stuff, then that’s going to make them unpopular whenever they make an accusation, which is going to lead fewer women to be willing to make accusations. So another thing that might work is for more accusers to be explicit about which parts of feminist discourse they disagree with. If they seem like reasonable people, maybe their accusations will be taken more seriously, less drama will be generated, and making an accusation won’t make you unpopular.
15% of Americans met their partner/spouse at work, so I’m not sure your claim about workplaces is correct.
I’m also not sure the comparison is fair. Workplaces have specific regulations about how people can interact; an EA meetup is not your workplace and attendees probably shouldn’t be held to the same standard of conduct.
I also know of a number of happy EA relationships. I think it would be a shame if we decreed that they were off-limits.
I know about the replication crisis, I’ve read “Statistics Done Wrong” and I’ve read some Ioannidis. Perhaps I was too subtle, my way of addressing these concerns was to load up on as many review articles and meta-analyses as I could find, in all the areas where there was enough research for me to do so. In other areas, I looked for as many studies as I could find and included them all.
This is not perfect either. Ioannidis has warned about some specific vulnerabilities in meta-analyses and review articles. There isn’t something perfect for me to do. I could have chosen to do nothing because the research is flawed. I decided that the subject is too important to ignore and I made the best of it.
I have seen several responses saying things like this, but in reality, the research in this article goes only as far as collecting the standard feminist narrative on sexual assault, which is not original and can be found in many places if you are familiar with this subject. The only thing that’s new is attempting to marry this perspective to EA, despite the methodology being highly partisan and significantly different from EA methodology.
Among the problems:
uncritically taking feminist sexual assault prevalence research at face value, without addressing the many methodological criticisms
mixing and matching studies with very different populations and methods to maximize the perception of male perpetration and female victimhood
failing to address the debate on false accusations
failing to discover any of the complicating lines of research, such as the token resistance research, and the research on self-justification and unreliability of memory
discussing drastic and hasty interventions like stings and medicalization
accusing an entire community of males of containing hundreds of rapists based on poorly-executed studies on a totally different population
While the article is being criticized for being too long, in some ways, it’s actually too short to support the extraordinary claims it is making, which are perhaps not fully recognized as extraordinary due to feminist research not being held to the same standards as other fields, and all sorts of glaring errors being normalized.
Here is my detailed rebuttal which is only enough to cover some of those problems, and here is an additional comment on the lack of balance in the original post.
I plan to write a new post about estimating the number of sexually violent people in EA. I have a large number of specific concerns about biasing our estimate. For instance, a lot of people commented or messaged me saying that the estimate was too high, but they didn’t incorporate any of the information which would actually increase the estimate. They only included information that would decrease the estimate. There are a lot of other ways we could go wrong with adjusting this estimate. That’s part of why my estimate is so simple. If I adjust it at all, I could easily be introducing biases.
I will invite the whole EA community to provide specific references that are related. I don’t know that this would decrease the chance of bias. It might increase the chance of bias. However, with such a post, one will at least see how complicated it is, think about whether one’s perspective is biased, and hopefully start compensating for whatever biases are present.
Why do you feel it’s important to have a more accurate guess regarding the number of sexually violent people in EA? I’m in favor of trying to measure the rate of sexual assault using e.g. the EA survey, because that is a metric we can track in order to measure whether things are improving. (Ideally using a question such as “Were you assaulted in the past year?”, so our metric will be responsive year over year.) But it seems to me that time spent refining our guess based on priors would be better spent implementing measures to reduce sexual assault.
I don’t think we can have an accurate idea of how much sexual assault is happening in EA without a separate high-quality survey. This is because there are so many definitions of sexual violence which contradict one another, that to ensure an accurate picture of what’s going on, we’d have to wrangle with definitions for a long time—and we’d end up asking a set of questions, not just one question.
I’d love to see a yearly undetected sex offender survey given to both men and women regarding how much sexual violence they committed against EAs in the last year, and a yearly sexual violence survey given to both men and women to ask how much sexual violence they received from EAs in the last year. If they added this to the yearly survey that would be awesome!
Then we’d have a way to track progress, and that’s important. The survey would have to be designed very carefully from the beginning though.
The reason I want to write a separate article about the number of sex offenders in EA is because it appears quite controversial. If we can get closer to having a consensus on sexual violence related matters, I think this will make us more effective at reducing it. The purpose of the article is not to create a more accurate number. I’m not even sure that’s possible. The purpose of the article is to address the controversy, explore the complexities, and encourage people to compensate for the various biases that may be interfering.
Edit: I’ve set up a collaboration with the yearly EA survey team!
I think there’s a tradeoff here. If this is created as a second, separate survey, there will likely be selection effects in who chooses to take it. I expect people who are more concerned about the problem of sexual assault (such as people who have been sexually assaulted) will be more likely to complete a survey that’s specifically about sexual assault. Given these selection effects, I suspect it’s best to settle on a relatively brief measure and include it in the main survey.
Brainstorming on what to include in that measure:
One idea is to just ask people “were you sexually assaulted” and let them use their own definition. After all, our goal is to reduce psychological trauma. If someone’s experience met some technical definition of sexual assault, but it didn’t bother them very much, maybe it’s not something we need to worry about.
Source. I don’t think using a broad technical definition like this would be very useful, but a narrow technical definition of rape seems like it could be pretty useful to measure.
This blog post makes the case for vague rules like “don’t be a jerk” and “don’t be creepy”. Maybe that could make a good survey question: “Did you get creeped out by another EA in the past year? How creeped out were you on a scale of 1 to 10? Here’s a rubric.” I actually think a measure like this could be less controversial than trying to precisely define sexual assault. Hopefully even the most fraternity brother-ish of EAs can recognize the case for not creeping chicks out. (Similarly, having a central registry that tells people things like “a lot of people are getting creeped out by you” seems like it could maybe work better than trying to define what exactly constitutes “assault”—it frames the problem as something you’d like to become aware of and fix, like having body odor, as opposed to grounds for ostracization? Of course ostracization is in fact justified in some cases—I’m just thinking aloud here.)
I suppose one issue with these measures is that they will fluctuate depending on the presence/absence of highly sensitive people in the movement. Overall, I’m much more comfortable using measures like these as indicators for what we should prioritize internally vs an overall measure of the moral worth of the movement. In other words, maybe we should not make them public? I don’t know.
It might also be interesting to include a measure of how many women in EA would like men in EA to be more direct and sexually assertive with them—see this comment.
Well, as a man in EA, I don’t like the idea of people thinking of me as a possible sex offender—especially if I’m not, in fact, a sex offender. And whenever you try to estimate how X men in EA are sex offenders, no matter what X you tell people, you’ve framed things in a way that is gonna make people see me as a possible sex offender. So maybe that’s why you got some pushback on that statement.
I’m happy for us to do a survey to measure sex offenses, because that will give us a way to actually measure and fight the problem. I know that any nonzero number that survey finds is going to reflect poorly on me, a man in EA, even if the number is much lower than we’d expect on base rates, because of the framing effects I discussed. However, I am willing to pay that cost because I care about addressing the problem of sexual assault. But I think trying to make an estimate based on prior information will just stir people up.
Cool!
What I was envisioning was a whole section within the survey where multiple questions about sexual violence are asked. For whatever reason, I described this using the word “separate”. That’s not actually what I was trying to suggest. I agree that if the questions are separated, there will probably be some bias.
If we use a definition that is vague, a lot of people will ignore the survey results. They’ll assume that a lot of what was reported is stuff they wouldn’t agree is a sexual assault. Therefore, specific definitions are needed. Ideally, I would like to see a set of specific definitions that a lot of people agree are sexual assault, and that cover a broad range of types.
To make sure the questions are relevant to the goals, I think there should be questions about things like whether the sexual harassment resulted in psychological harm, suicidal behavior, or intentions to leave the workplace or movement. I’d also like to see questions about whether sexual assaults are happening at work, EA events, etc. Depending on how well anonymized the survey is, we may or may not get answers to these sorts of questions.
Without knowing the limit to the number of questions we can add, there’s no point in discussing what should be asked. We would just waste time optimizing for the wrong trade off between detail and brevity. Also, it would be good to get some perspectives from people who do research in related areas. I’m going to hold off on investing time into planning until I have had a collaboration with the survey team.