> Why do we place the blame and assign malice of intent to those who have little control over the constraints?
Because 1) there is a feedback loop that does show that having role models at the end of the pipeline does help the entire pipeline. 2) It is easier than restructuring the entire pipeline than admitting that it is a societal problem rather than big bad Google being sexist (I'm not saying Google isn't sexist).
This is kind of a pattern in problems that don't have smoking guns. We like to pretend that things are first order problems (problem causes an effect, directly) when most problems are pretty high order and complex. First order problems are usually easier to solve because they have a clear smoking gun. You take care of the bad person that fired the gun and the problem is solved. I actually suspect that this is an evolutionary problem, being that most of our early problems could be approximated by first or second order and thus increase our survival rate. I'm also pretty convinced that we've solved most of those now, and thus that line of thinking doesn't work for modern problems.
But for part 1), if you have the flu, tissues help. They don't cure the flu, but they help you from preventing the spread and alleviate the symptoms. So if you take my preposition about first order problems, I don't think it is hard to see why people would get stuck at that stage. Essentially "we did something and it had a positive effect. So let's keep doing that thing!" That's not bad logic. You admit that they do have __a little__ control over the constraints, so it isn't surprising that they have __a little__ impact. The failure is not continuing to look for more potential solutions and stagnating. Just need to do some PCA, find the other factors, and push on those (as well as what we've already found to have positive effects). But this is much easier said than done, and I think that is part of the problem. Difficult problems are tautologically difficult, but we as a society like to pretend there are simple solutions (come on, admit it, you do this too).
So why do we do this? Well, how many people do you know break down complex problems into multiple components? I know very few, and even then they don't consistently do it.
> If we put actual performance metrics and pay on the line for achieving these physically unattainable goals, everyone would be fired.
I'd argue that they would just start cheating. Which I think that there is evidence that this is happening (some being given in this article).
Because 1) there is a feedback loop that does show that having role models at the end of the pipeline does help the entire pipeline. 2) It is easier than restructuring the entire pipeline than admitting that it is a societal problem rather than big bad Google being sexist (I'm not saying Google isn't sexist).
This is kind of a pattern in problems that don't have smoking guns. We like to pretend that things are first order problems (problem causes an effect, directly) when most problems are pretty high order and complex. First order problems are usually easier to solve because they have a clear smoking gun. You take care of the bad person that fired the gun and the problem is solved. I actually suspect that this is an evolutionary problem, being that most of our early problems could be approximated by first or second order and thus increase our survival rate. I'm also pretty convinced that we've solved most of those now, and thus that line of thinking doesn't work for modern problems.
But for part 1), if you have the flu, tissues help. They don't cure the flu, but they help you from preventing the spread and alleviate the symptoms. So if you take my preposition about first order problems, I don't think it is hard to see why people would get stuck at that stage. Essentially "we did something and it had a positive effect. So let's keep doing that thing!" That's not bad logic. You admit that they do have __a little__ control over the constraints, so it isn't surprising that they have __a little__ impact. The failure is not continuing to look for more potential solutions and stagnating. Just need to do some PCA, find the other factors, and push on those (as well as what we've already found to have positive effects). But this is much easier said than done, and I think that is part of the problem. Difficult problems are tautologically difficult, but we as a society like to pretend there are simple solutions (come on, admit it, you do this too).
So why do we do this? Well, how many people do you know break down complex problems into multiple components? I know very few, and even then they don't consistently do it.
> If we put actual performance metrics and pay on the line for achieving these physically unattainable goals, everyone would be fired.
I'd argue that they would just start cheating. Which I think that there is evidence that this is happening (some being given in this article).