I'm trying to explain that gender targeting is imprecise.

When I say the hit rate of ads targeting women is 50%, I mean that of the number of people who see ads that were targeted at women, only 50% of them were women. Targeting men works a little bit more consistently, by usually about 10%, but I don't know why.

That's industry-wide. I don't know if Facebook does better or not, but I have no reason to think so. Even if they did, it probably doesn't matter that much.

I'd expect Facebook to be better because they ask what your gender is and many people will have explicitly told them.

50% of viewers were women on ads targeted to women? How much better than chance is that when you take into account there are more men than women on facebook? It seems low given how well they can guess less tangible things like sexual orientation...

Which is why ad targeting based on gender is perfectly fine.

well, in this case it isn't perfectly fine because it shouldn't be described as "Gender. Target ads to women, men or people of all genders." if there is only 50/50 chance to display ad to woman. Isn't it just ridiculous and straight lie to potential advertisement buyer?

Ad buyers are very sophisticated spenders and know the score.

If they don't, they're going to get their money taken anyway and a lot more savagely than a mistargeted campaign hinged on gender.

I really have to drive this point home. Selecting gender as a target on Facebook isn't actually selecting gender and everybody involved knows it. It's a much simpler way to describe a much more complex set of decisions.

What's confusing me here is that this is what you seem to be saying is happening:

1. Facebook knows my gender.

2. Facebook tell advertisers they will select/not select me based on my gender.

3. Facebook, despite knowing my gender, completely disregard that information and try and guess it with exceptionally poor results.

I was speaking more about the industry as a whole originally. Facebook knows what you report as your gender but that's not a very good attribute to target off of. If I were doing e-commerce, for example, I would never use Facebook's implementation of that target (if they're going based off of what's reported to Facebook) because it would completely exclude gift purchases across genders. It would leave a ton of money on the table and be a very stupid thing to do.

My entire argument is that splitting hairs over this is irrelevant because any advertisers who rely on the accuracy of that attribute are stupid and pissing money away to begin with.

We are very specifically on a topic about what facebook does, and what I was originally replying to you about was that I think your general findings are not particularly relevant when looking at facebook. You said this story is totally overblown because gender targeting isn't normally very accurate, but here it absolutely will be because a large number of people will accurately set this value on their own profile.

> I would never use Facebook's implementation of that target (if they're going based off of what's reported to Facebook) because it would completely exclude gift purchases across genders. It would leave a ton of money on the table and be a very stupid thing to do.

But we're not talking about gifts, we're talking about jobs.

> because any advertisers who rely on the accuracy of that attribute are stupid

But why would it be inaccurate for adverts on facebook for logged in facebook users?

So let's think this thing through for a second. This will be best read in a Louis Rossmann-esque manner.

Hypothetical me: An HR staffer at a company with a headcount large enough to need an HR department. I get to post job listings and spend advertising dollars for those listings on Facebook.

Hypothetical my manager: Female. As ~75% of HR managers _globally_ are.

Oh yeah, hypothetical me again: Also female; depending on country and department rank, your HR staffer is 70-90% likely to be female. (It's the outlier, but in your country, new HR hires are 96% women)

The argument: Hypothetical me is going to buy ads, that my manager and probably my mostly-female coworkers can see, where I checked a box that says "only show this job listing to men". In an overwhelming social climate pushing for inclusivity and equal pay. In a work culture where these efforts are being _led_ by HR departments.

On what planet do think that this actually occurs? It is almost guaranteed that _at least_ one woman is involved in the decision making process or can audit it. Where are these companies with entirely sexist hiring processes that have significant advertising dollars to spend and where those dollars spent are going to make a significant impact in whether a man gets that job over a women in that industry and at that company?

I'll wait.

You've been arguing up until now that you can't target adverts to a particular gender. Do you now agree it seems you can?