6

Kathy_Forth comments on An Exploration of Sexual Violence Reduction for Effective Altruism Potential - Effective Altruism Forum

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (133)

You are viewing a single comment's thread. Show more comments above.

Comment author: xccf 13 November 2017 10:57:42AM *  20 points [-]

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.

Comment author: Kathy_Forth 13 November 2017 11:03:42PM *  0 points [-]

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.

Comment author: xccf 14 November 2017 02:14:20AM 1 point [-]

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.

Comment author: Kathy_Forth 14 November 2017 02:34:20AM *  1 point [-]

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!

Comment author: xccf 14 November 2017 09:55:04AM *  4 points [-]

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!

Cool!

Comment author: Kathy_Forth 15 November 2017 01:23:47AM *  2 points [-]

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.