Ask a Female Engineer: Thoughts on the Google Memo

Note: I find it interesting/disturbing/sad/telling that I've been sitting here for a long time contemplating if I should even submit this message since I use my real name here. The fact that we, as a society, have come to a point where we are afraid to even have this discussion really makes me sad. I respect every one of my colleagues deeply, male and female alike. The idea that someone could twist my words and paint me as a misogynist is beyond troubling.

This seems to be the common argument against diversity programs, but it strikes me as statistically true only if we assume a very even distribution between STEM graduates and prospective employers. Given tech's reputation as being relatively hostile to women, a company could theoretically find ways to advertise to prospective women employees that their internal culture was more welcoming of them -- that, in fact, they wouldn't be subject to the sexism that the female engineers in the linked article all said that they routinely face. This doesn't require the company to have different hiring standards between genders, or to pay women more. It does require them to change their recruiting practices in ways that acknowledge they may have to make specific outreach to women and other underrepresented minorities, but that doesn't strike me as having to be inherently discriminatory.

Thank you. I have been noticing this for the past year, both here and in almost all other online forums. And it's not just the ability to troll or make a comment that you know might push a few buttons.

It's questioning whether it's even safe to post a logical argument against any of the narratives deemed sacred these days by the left.

The fact that the worst offenders in this new witch hunt are the same ones who have massive amounts of data on all of us is terrifying.

Choosing to recruit only from pools where the proportion of women is greater than 20% isn't quite the same as lowering the bar. The bar is at the same height for men and women, just with a skewed population. You're artificially excluding pools of people who would make good candidates, but since both women and men can bypass your outreach efforts by going straight to you, you're not refusing to hire anyone who is both qualified and motivated enough to apply directly.

I think proponents of the google diversity programs are arguing that they do this. I'm not sure whether they do. I think the real situation might be a hodgepodge of systemic factors and biases in both directions that sum up to something unpredictable, plus a few largely ineffectual diversity programs, and a massive question mark around why there are so few female CS grads in the first place (biology! sexism! gender roles! c64 ad campaigns! inertia!). Nevertheless, it's a bit more complicated than just lowering the bar for women.

these are the results:

"there are 48840 males, we will pick only 25000(51.187551187551186%)"

"we will pick all females to represent the company reaching out to them"

"let's say the company is going to hire 5000"

"hiring based on competence and taking females when equal"

"results:"

"male: number: 3505 percentage: 70.1% average score: 123.81256204767604"

"female: number: 1495 percentage: 29.9% average score: 123.75346343448992"

"if we force the 50% ratio"

"the average male score: 126.26036797470225"

"the average female score: 119.60577230318559"

so forcing a 50% ratio does indeed lower the bar. data for males and females were generated using the same function so arguments about biological factors are not even needed.

the code:https://jsbin.com/nogujuqewe/1/edit?html,console,output

Or you solicit more resumes from the 20%. Or you randomly throw out some resumes from the 80%. Neither of those will move the bar.

"Choosing to recruit only from pools where the proportion of white people is greater than 95% isn't quite the same as lowering the bar."

I don't think it would get much sympathy but it's an equivalent with race substituted for sex (both are protected and it's illegal to discriminate based on them).

>>Nevertheless, it's a bit more complicated than just lowering the bar for women.

If you own a pub and want only white waitresses so you only invite white women for interviews you can do that without lowering the bar as well. Still you are discriminating even if you put elaborate system out there which magically result in only (to make the point stronger, substitute with a ratio like 90-10 or 80-20 to make the situation equivalent) applications from white women at the end.

My claim is a much narrower one, that you can hire a disproportionate amount of female developers without lowering the bar if you bias your incoming hires. It can be simultaneously true that Google's diversity policies are harmful to quality (because they restrict where Google hires from) while their female developers are as qualified as their male developers (because they came from the same place and meet the same standards).

If more companies are doing that then it's impossible to sustain without lowering the bar. If only you are doing that there is no point because then others will hire more men (as there is more qualified men left proportionally as you took bigger % of qualified women).

I am saying that the policy of "we don't lower the bar, we just look more into avenues to hire more women specifically" is somewhere between pointless and dishonest (dishonest as in created to hide the discrimination based on sex).

EDIT: As to affirmative action: I agree it's not the place for debating ethics of it. I am saying that affirmative action = lowering the bar either directly or indirectly and there is no way around that fact (at least industry wise, you can maybe sustain it locally if you are ok with others skewing their ratio in the other direction).

This rests on the assumption that hiring from a given pool exhausts it. It seems intuitive that hiring students from a university or bootcamp would have the opposite effect, as would hiring students from a particular academic background, since unemployment/pay metrics and prestige would drive more students there.

dishonest as in created to hide the discrimination based on sex

Since the clearly stated goal of affirmative action is to hire less of a majority group, it seems more likely that such a policy would be created to prevent imposter syndrome and "my male co-workers think I'm incompetent because of all the diversity hires" syndrome. With such a policy, nobody is a diversity hire.

So it implies lowering the bar unless it doesn't.

Companies can put more effort into finding woman candidates without caring whether the whole industry does so. If some companies bias toward women (without lowering the bar), and some companies don't bias, then the overall effect is that qualified women can get hired instantly, and more of them might be encouraged to enter the industry.

The difficulty is that "this discussion" can be - and usually is - conducted in a way that is harmful to women, either on a broad scale (specious arguments / failure to understand systemic bias) or an individual scale (wrecking someone's day / making a formerly welcoming environment feel hostile).

Those real consequences are on the line every time someone hits "Post" in this sort of discussion, and are a really good reason for any thoughtful person to pause and contemplate before doing so... perhaps do some additional self-education, or take the time to pose genuinely explorative-questions rather than rhetorical-questions or flat-out conclusions. If more people did that, I think you'd eventually see a lot less fiery refutation and much better discourse.

"If the numbers are true and the gender distribution in STEM graduates is 80/20, and you are intent on increasing your number of female employees, you have two choices. You either lower the bar on the 20% or you raise the bar on the 80%. In the end, the net effect is the same. The employees from the 20% group had a granted advantage against the 80% group."

This logic assumes that the 80% and 20% are functionally equivalent? (Which can be so if there's, eg, no systemic bias, but seems rather less likely when such is present.)

Another option would be to realize the the 20% already had to overcome substantial hurdles to get where they are, and to factor that into your decision-making.

Its so easy to make up crap about how its impossible to fix the issue without (made-up strawman). How about turning our intelligence toward useful comments? E.g. you could look harder for truly qualified candidates from among the 20%?