Hacker News Reader

Comment by raganwald | Hacker News Reader

Probably incorrect.

And I could just as easily posit the reverse. Say we have one woman and nine men. If we add a woman, does it really double the likelihood of a harassment suit? Why would it? Either one of the nine men harasses women, or he doesn't, why would adding a second women change his behaviour?

On the other hand, if we halve the proportion of women by hiring ten more men, I'd say you may have doubled the chances of having a harasser on your staff.

I guess it all comes down to whether you walk around thinking that women cause sexual harassment lawsuits or men cause sexual harassment lawsuits. I think it's the men doing the harassing, but whoo-boy are there a lot of apologists out there who believe that women are making this stuff up.

View on HN

I don't completely understand your argument here, on a purely intellectual level...

In a very simple model, going from 0 women to 1 woman should increase the number of harassment suits, right? Then going from 1 woman to 2 women should have a similar effect, because all naturally occurring functions are differentiable :-) Intuitively it seems to me that the amount of reported harassment should be proportional to how many opportunities for harassment exist, or (under a different theory) how many opportunities for false claims exist, which is the same number by another name.

So it's a little strange to hear you say that anyone who disagrees with you must believe in nasty things like "harassment is made up by women". Do you know how hard it is to figure out causality from observational data? For any possible trend you could find in the data, like "reported harassment scales as the number of women times the square root of the number of men", one could probably make up a just-so story that blames men, and a different just-so story that blames women.

I think the onus is on you and those who agree with you to discard all this "intuitively" and other hand-wavy stuff and come up with some hard evidence and/or citations, or back down and say, "we're making this shit up because it suits our biases."

I wasn't born yesterday. If someone makes a claim without a shred of evidence, it's perfectly legitimate to say, "Well, here's another alternate explanation that's just as unproven, so clearly yours has nothing special going for it." I gave you a Flying Spaghetti Monster, and you're telling me that intuitively, God has a long white beard, not a saucer.

I'm perfectly ok with your saying that you choose to believe something without a shred of evidence, to take it as faith.

But that won't change the fact that simply making shit up, asking if it's "wrong," and then saying that you'll stick to it unless someone else proves your unsubstantiated prejudices are mistaken is bunkum.

That thinking isn't thinking.

His intuitions are simply basic mathematical intuitions ("natural functions are differentiable") and have nothing whatsoever to do with any prejudices. He isn't "making shit up", he just isn't justifying it with as much mathematical rigor as I did.

It really does not help discussions like this to simply assume and accuse those you disagree with are prejudiced and unthinking.

Sorry, what he and you are doing is starting with assumptions you are making up and then saying GIVEN this shit I'm making up THEN given these mathematical intuitions I cherry-pick as simple demonstrations of the conclusions I've drawn THEN the conclusion I drew before I back-filled my reasoning holds.

Again, without even the barest attempt to go out and gather some empirical data. That kind of talk belongs amongst consultants pedalling methodology snake-oil. And yes I am being dismissive of your so-called arguments.

Around here we regularly make fun of the "social sciences" for their lack of rigor. Except, it appears, when we want to throw our prejudices around. Shame on you all for treating Hacker News like it's Reddit.

Which prejudices?

I made three assumptions. One was social distance (a harasser will harass only one of the K < N people he has met). The other was your assumption that a harasser always harasses. The third was your assumption that all harassment allegations are true.

I assume that the social distance assumption is the one you believe to be prejudiced?

I'm also curious why you believe your post is any different. You also posited a mathematical model (exactly like mine, except without the social distance), and hinted that those you disagree with must be sexist. Is it merely the existence of an ad-hominem attack that makes your post more valid?

You wish to be clinical about this? Consider.

Your statement essentially assumes linearity, when such certainly isn't the case. Look at other industries, harassment lawsuits do not rise in tandem with number of women. This is obvious, one can grow the number of women in an industry far more quickly than harassers. Since these harassers cannot well parallelize their activity, the strongest explanatory variable will be the number of potential harassers. Raganwald's point. We are taking it as a given that most men will not provoke things till disciplinary actions are required and most women are not lying about them.

Your post is also highly simplifying - assumes trivial reducibility, homogeneity and ignores the effect of interactions to create emergent modalities. Observe that Not all women will escalate to lawsuits or even disciplinary action first. Lawsuits are a sign of an endemic problem, lack of support - symptom of horrible rot. With more women the support structures would be such that these problems could be nipped in the bud early. Indeed it's very possible that beyond a certain threshold, the number of women in an organization shifts company culture.

These two points show why it is not a simple matter of 500% more lawsuits.

If we add a woman, does it really double the likelihood of a harassment suit? Why would it? Either one of the nine men harasses women, or he doesn't, why would adding a second women change his behaviour?

He might not harass all women. He might only harass women in his general vicinity, either a social, work related or geographic.

Lets consider only correct accusations, and further suppose a given individual only interacts with K coworkers (for K < N, with N the company size). Harassment will occur if one of the K coworkers a given harasser H interacts with is a woman (assuming he always harasses). The probability of that not occurring is (1-P(coworker is woman))^K. The number of harassers is beta x (1-P(coworker is woman)) x N, with beta the probability of an individual man being a harasser. So the probability of a no correct harassment accusations is (1-P(woman))^(K x beta x (1-P(woman)) x N).

Provided P(coworker is woman) is small (as is the case in most tech companies), this number decreases as P(woman) increases. I.e., more women raises the probability of a harassment lawsuit.

And this is all in a model which assumes women never ever make this stuff up.

Notice the use of your word "might?" That's a clue that we're bandying around unsupported guesses. As in, "well, it might be that this, or it might be that that."

There's nothing wrong with that kind of talk, provided it is followed by, "And here's how we would go out and discover what the actual truth is." That is Science. Stopping with "might" is fine, provided you don't call it anything other than speculation.

I'm not calling what I said anything other than speculation. I put it there to show that if you're going to make stuff up, you can arrive at any conclusion you like. So I needn't defend my suggestion. I agree it is indefensible, just as the one you're supporting is indefensible.

It all comes down to either saying "I don't know and that's fine with me because I have faith," or "I don't know, and I'm going to find out what is actually true."

Take your pick, but start with "I don't know."

Let me quote the part of your post I'm disagreeing with:

I took a model very similar to yours, in which women never make stuff up. The only factor I added was a (in my opinion quite reasonable) assumption that you can't harass the people in the company you've never met.

I.e., I showed one can disagree with you even while believing that women never ever make stuff up.

If you want to call all beliefs based on mathematical models rather than some frequentist statistical estimate "faith", fine. I don't agree that mathematical models are useless, I think they can give us excellent guesses about the world and we should use them until we have something better (particularly when a variety of different models give similar conclusions). But I suppose we just have radically different views on epistemiology.

But in that case, why are you claiming that kudzu_bob's claim is "probably incorrect"? How could you possibly know this? Absent some frequentist study proving this is the case, why would you believe that "it's men doing the harassing"?

Meanwhile, on Planet Earth, as the number of women in the workforce increases, so does the number of sexual harassment lawsuits.

Correlation is not causation, my friend. I can't believe I have to say this on Hacker News. My God, what happened to us?

Reported cases of harassment are up, because of:

1. it is no longer socially acceptable to harass? 2. we have removed barriers to reporting harassment?

Maybe there are more women in the workforce because we made it safer for them to work and we removed their dependence on a husband for sustenance.

Let's do a little thinking about this. I feel like I'm listening to a conservative call-in show.

Reply on news.ycombinator.com