“ You are saying it as if it was a good thing. You think there should be more of this?”

I’m stating it as a fact. You are the one adding qualifiers.

In UE there is GDPR to limit companies using your data but in practice, people sing away the right to limit the use for their data whenever they download an app, so it is useless. I’m afraid that my opinion or yours on data privacy won’t change that.

On the other hand, excess of information or trash data is useless for them. Which is the reason they pay data engineers to ensure data quality.

And it is known that third party companies are doing studies by crossing DB.

“ I guess you are from a “humanities background”? Because you seem failing to understand the difference between “percentage of women who are interested in these jobs is much smaller than percentage of men” and “women are not interested”.”

I think it is you who fail to understand that when you say “ So you want those women who have no interest in jobs in “male-dominated” areas to read ads they can’t possibly be interested in in order to get “get percentages right”?” imply that no women are interested. Which is a false premise that you tried to defend. In fact, if your filter criteria is “men”, you can send the ad to men who are not interested. So interest groups that the person follow are a much better criteria to predict interest than sex. It is quite simple to understand and the reason that labels and tag where introduced on post/FB groups. It makes easier to filter by that criteria.

Interestingly enough, you are not bothered by the men who are not interested in the ad that would received it anyway. It only becomes a problem if a woman who is not interested get it. Nice strawman. XD

Again, if, let's say a 10% of men are interested in comics, while 5% of women are interested, men would be 66% approximately, the majority, but if your filter criteria is “men”, you would fail 9 out of 10 times and send the ad to people who are not interested. If instead you filter by people who follow marvel FB account, you would send the ad to more people who are interested. Same logic apply to job adds. So yes, I do understand the idea about targeting ads to the right people and also know that CV, interest and skills are way better filters to find out good candidates.

I think it is you who might want to read about different techniques to identify relevant variables when doing an analysis. Because yes, it is that simple to use tags and labels to make a “select” and filter by them.

“ (at least until they figure out how to get detail information on every individual and some kind of system to be able to process all that data — using ML or whatever)”

hahaha, good trolling. That is already old technology. Considering IAs record our voice and that information is classified for further analysis, I would say that soon they would be sending offers to automatically fill out orders to shops. I mean, Amazon already tried that with the button, but logic dictates that they would try that with smart fridges being the selling point that you “would never go out of milk and that it would control that you don’t have spoiled food”. And same for other things that you consume regularly. It is a matter of time that providers offer to automatically send you those things.

“ And you think that if it was that easy the advertisers would not be using this information? Are you saying that advertisers not using such information on purpose?”

I’m saying that they have unconscious bias. And that unconscious bias cost money to companies. Since it cost money, companies like google try to reduce it.

They don’t do it on purpose, they are just like you, people who make assumptions that aren’t backed up by data or results.

“ As for FB — I personally don’t remember much jobs advertised. Most of what I get is obviously based on my previous browsing to shopping sites — I leave it to you to figure out how FB manage to enable this ads without them actually not knowing themselves what goods I looked on, say, Amazon.”

Because you don’t belong to the target demographic which as the original article said, excludes people above certain age too. They are only is interested on young white males, if you are “old” (and by usual criteria you might be considered old at 35), you would be excluded too. As I said they have an archetype in mind and discard anyone who won’t fit the image they have of a good candidate. For some time people joked that candidates who wear a suit instead of a hoodie were discarded by looks. That is not rational, neither effective.

FB use cookies like anybody else(which make also easy to work with the raw data).

Now, I’ll let you make a second guess about my “background” which should be irrelevant anyway since what should be relevant for a self proclaimed sceptical is the data and arguments, not who is making them…

Again, is there any amount of data and logic that would make you question you initial assumption about women not being interested? I mean, the very fact that they bother suing and spending time, effort and money on this should be an indication of interest, but if you still had doubts you could just google for 30 seconds to find out if there are women and people of color interested in those things… And yes, it is that easy so leave it to you to figure out why they didn’t since it is not because they lack the appropriate tools and knowledge.