It’s interesting to imagine a hypothetical. Suppose a company chose an ad targeting segment with perfect 50/50 historical probability of a viewer being of either traditional male / female gender identity given they are in the segment.
Now suppose through some weird fluke, the ads are actually only delivered to people of one single gender classification.
Has the company done everything required of them by trying to target an equal mix along this particular characteristic dimension that society attaches a legal criteria to? Or has the company failed to do what is required because they merely tried to target 50/50 but failed to guarantee that it actually happened?
Then flip it around. Suppose a company targets an ad segment that is 100% male because they are actively discriminating against women. But through some fluke of ad delivery they actually show the ad to a 50/50 mix. Are they free of committing a crime because nobody ended up being discriminated against? Or have they committed a crime merely by trying to target a certain subset of the population with a disapproved set of characteristics?
To me it begs a question of whether the company’s legal obligation occurs at the level of defining who to try to target, at the level of who is actually targeted (even if it is executed by a third party), or both?
A further interesting question is that if this applies to one type of resource (jobs) does it also apply to others like plain old goods and services?
Suppose a sporting goods company knows that only 100% males have ever bought their merchandise or expresssed willingness to buy in market research.
Is the company legally obligated to still show ads to women?
If not, what is the underlying criteria that supposedly separates job listings from product listings?
Another interesting question is whether accidental correlation with mix of the targeted audience also results in blame.
What if you target a set of people with behavioral characteristics A, B and C. None of them has any obvious connection to gender, sexual orientation, etc., but it turns out they are heavily correlated and (honestly) nobody knew ahead of time (maybe even the correlation is seasonal and prior data could not have shown it). Does this make the company responsible?