We ran the numbers, and there really is a pipeline problem in engineering hiring • Hacker News Board

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#221

Post by weberc2 » Tue, Dec 03, 2019, 10:48 PM UTC

shadowgovt wrote: post #184

TeMPOraL wrote: post #31

Earlier quoted context omitted.

And the point of criticizing the criticism of "it's a pipeline problem" is that, wherever you try to improve the pipeline, you cannot expect to make things better than they are at the input of your pipe segment. Working near the end of the pipe, you'll quickly hit diminishing returns. If the input at your segment is biased 4:1, then you can't improve past it. Sure, if your output is 8:1 then it's bad, but you won't g…

> you cannot expect to make things better than they are at the input of your pipe segment Not in the short-run. But the reason working on the pipeline is valuable is that discrimination includes negative feedback loops. When people who would be entering the pipeline look for role models and success stories and don't find any, they're discouraged from trying. So while there are of course diminishing returns in investi…

I'm sure I'll be downvoted for merely asking, but I'm genuinely curious: what is the evidence for discrimination? Is it simply the null hypothesis (gap, ergo discrimination) or are there studies or something that actually highlight discrimination, and if the latter, how much of the gap is explained by discrimination?

Last I asked this question, I was referred to the orchestra blind audition study (Goldin et al), but I'm pretty sure that was data from 1950s-80s, from a different industry, and didn't actually demonstrate a significant effect when accounting for large error margins.

Downvote me if you like, but at least explain the reasoning for those of us following along at home. In the worst case it's an opportunity to evangelize your point of view.

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#222

Post by uwuhn » Tue, Dec 03, 2019, 10:50 PM UTC

I'm going to preface this post by saying that I don't have a problem with most of my coworkers (male and female) being Asian. In fact, I prefer it. I'm not Asian, but the majority of my friends, classmates, and coworkers throughout my life have been, and that continues even now during my career in tech.

Whenever I see anything related to affirmative action being discussed nowadays, I think that it's only a matter of time until Asian women are treated the same way as Asian men. Once you've managed to close the gender gap in engineering, you have a new problem to deal with -- a lack of diversity among women engineers in regards to race and socio-economic class. A continued pursuit of diversity will require discrimination similar to that exhibited by prestigious American universities.

My coworkers (past and former) and friends who are women SWEs overwhelmingly fall into two buckets: American-born Chinese with parents who are middle-class or higher, and PRC-born Chinese with wealthy parents.

I get the strong impression that gender diversity is viewed as more important than race and class. I'm a male of color who has been in the industry for more than four years now, so I no longer have to worry about breaking in. If I were applying to CS programs or looking for my first job right now, I would feel some resentment. I've convinced and helped three of my friends to do a career switch because there's just so much assistance (financial, educational, and otherwise) available for women. From what I've seen, there is just so much more provided to help women get into the field.

I commend the big tech companies for lumping male URMs and all women together when it comes to prioritization, but this isn't the case for most companies, who are expending great effort on balancing the gender ratio while treating men of color as second-class URMs, or even ignoring their status completely. It's fortunate that the best jobs are the most fair, but even if a place sucks, a first SWE job is still a first SWE job.

What I predict will happen is the gender gap will begin to close, but the aforementioned diversity issues not related to gender will remain. This will be due to a combination of various factors, with the most significant being fatigue with affirmative action practices, and that discussing socio-economic and race is much more sensitive than discussing gender, Saying "stop hiring men" or "only hire women" is easy, even if you are asking people to discriminate against candidates similar to themselves. "Stop hiring Asian women" is not. And if you are willing to make that request, why would anyone listen?

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#223

Post by hkmurakami » Tue, Dec 03, 2019, 10:50 PM UTC

HeroOfAges wrote: post #196

spamizbad wrote: post #171

Earlier quoted context omitted.

It's a matter of perspective taking. If a "Never saw a line of code until I took this class" sits between two "Coding since 11" students, they're going to watch their neighbors effortlessly breeze through the curriculum. If they don't know those student's background, it's not unreasonable if they come to the (false) conclusion that they're not cut out to be a developer because it's just so much easier for "everyone e…

Even if you do have a "fast" track and a "slow" track, the only ways I can think of to get travelers on either of these paths to the same place at the same time is to either slow down those on the "fast" track, or speed up those on the "slow" track. What I seem to be witnessing is a willingness of people to slow down those on the "fast" track instead of expecting people on the "slow" track to work harder to catch up.

They definitely don't need to get to the same place at the same time, and don't even need to end up at the same final destination. Some will go onto PhD programs, some will not, etc. Having two (or more) separate tracks allows the late starters a better chance at gaining competence without getting crushed by the competition with more experience.

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#224

Post by inerte » Tue, Dec 03, 2019, 10:50 PM UTC

ForHackernews wrote: post #213

supernova87a wrote: post #187

With such politically charged issues, it's always a problem of people wanting to believe something is possible (and mandatory to pursue) banging their heads against what is true and actually possible. If the percentage of female computer science graduates (as a strong proxy for the available candidate population) is 18% (Google for the facts and stats), how is every company to hit an idealistic goal of 50% representa…

Does anyone have a goal of 50% female developers? Can we see some citations for that? Most companies I've worked at claimed to be trying for 20%, and they were failing abysmally at it.

I've heard from a Googler that yes, they want the workforce to match the outside world. With that in mind read https://diversity.google/annual-report/#!#_what-would-it-tak... and "That’s why in 2019 we challenge ourselves and others to think differently so that we widen pathways to tech. Only then will we reflect our consumer base, and truly elevate our ability to build products for everyone."

So Google wants to reflect the consumer base, so I guess we can take that to mean ~50% female devs?

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#225

Post by nl » Tue, Dec 03, 2019, 10:50 PM UTC

shadowgovt wrote: post #197

TeMPOraL wrote: post #120

Earlier quoted context omitted.

I'd agree with your observation about different realities, and I think it's woefully underappreciated. I think it only truly clicked with me once I entered a stable, long-term relationship with my now wife - which made me immersed in the daily minutiae of a woman's life, the stories, the perspectives, the concerns, the emotions. It's something that's hard to learn from just brief, casual interactions, no matter how o…

There's also history to consider. Computing was a woman-dominated field until relatively recently in history, which strongly suggests that the root causes of the shift in male/female ratios in technical fields isn't biological (unless there was a massive biological upheaval in the human species in the past 50 years).

the root causes of the shift in male/female ratios in technical fields isn't biological

I agree with this, but I'd just note that the shift isn't as dramatic as you are making out.

Computing was a woman-dominated field until relatively recently in history

The field of computing started in the 1950s (generously). Worker proportions are difficult to estimate, but the maximum proportion of female computer science degrees was 37% in around 1987[1]. In 1986 The Atlantic noted that "[in the federal government] only seven percent of the employees in the top five CS [pay grade] ratings were women, while more than three quarters in the bottom grades were."[2]

[1] https://www.aei.org/carpe-diem/chart-of-the-day-the-declinin...

[2] https://www.theatlantic.com/magazine/archive/1986/09/women-i...

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#226

Post by gjmacd » Tue, Dec 03, 2019, 10:51 PM UTC

Not one mention of age as being a factor as passing over good talent because of ageism. I'm 54, I walk through the door to interview and people run the other way before I even sit down with them because I probably look like their Dad -- meanwhile, i'm not only as current on skills than anybody I'm talking to, I'm probably the best option at a startup since I've seen it all (three and four times over). It's ridiculous that you age out of this industry at 45.

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#227

Post by TheFiend7 » Tue, Dec 03, 2019, 10:51 PM UTC

ForHackernews wrote: post #213

supernova87a wrote: post #187

With such politically charged issues, it's always a problem of people wanting to believe something is possible (and mandatory to pursue) banging their heads against what is true and actually possible. If the percentage of female computer science graduates (as a strong proxy for the available candidate population) is 18% (Google for the facts and stats), how is every company to hit an idealistic goal of 50% representa…

Does anyone have a goal of 50% female developers? Can we see some citations for that? Most companies I've worked at claimed to be trying for 20%, and they were failing abysmally at it.

Even so why do diversity weights exist at all? Should it not be based on performance or results? I understand the notion that people can be biased resulting in biased hirings, but why does that insist we must twist the bias in the opposite direction? Both sides of this fight are equally guilty of not hiring on the right principles which is how well can a person do a job.

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#228

Post by baddox » Tue, Dec 03, 2019, 10:53 PM UTC

drdeca wrote: post #198

baddox wrote: post #132

Earlier quoted context omitted.

> wherever you try to improve the pipeline, you cannot expect to make things better than they are at the input of your pipe segment. But that's basically just complaining that there's no quick and easy fix to your hiring problem. Well, too bad, maybe it actually is a really hard problem that you're going to have to work really hard at for a long time. If there was some relatively trivial solution like "have all hirin…

I don't understand what you mean. Are you saying that the people who are receiving a biased mix should introduce an opposite bias to counteract it before sending on to the next step, or not? Possibly I am misunderstanding what a "pipeline problem" refers to. Wow, I'm experiencing some deja vu. I feel like I already (like, a year or more ago) sent this comment and got a reply, but I don't remember what the reply was.…

My point is that, if there is in fact a pipeline problem, that probably does not constitute a shift in responsibility away from the hiring company. Saying "you cannot expect to make things better than they are at the input of your pipe segment" is explicitly attempting to shift responsibility away from the hiring company. I simply disagree. I think a company is still largely responsible for the outcomes of its hiring process, even if fixing those outcomes turns out to be an extremely hard problem that cannot be solved overnight with a diversity webinar.

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#229

Post by nl » Tue, Dec 03, 2019, 10:54 PM UTC

TeMPOraL wrote: post #214

nl wrote: post #210

Earlier quoted context omitted.

In terms of pipeline argument, the memo can be viewed as arguing that the pipeline is biased at every stage That's your interpretation, and isn't mentioned in the Wikipedia page nor the original memo. Both the memo and the Wikipedia page discuss the changes Damore suggested making at Google , not anything about changes in the pipeline.

> That's your interpretation, and isn't mentioned in the Wikipedia page nor the original memo. Yes, it's my interpretation. I quoted Wikipedia to show directly (the quote is self-evident) that GP is misrepresenting the very article they're linking to, and I then followed with reinterpreting that quote through the lens of the "pipeline argument". As far as I recall the memo (I read it two years ago), Damore didn't exp…

In that case I think it's fair to point out that the OP's interpretation of the memo is valid ("Damore's memo (which I assume is what you're talking about) absolutely did not attribute Google's lack of women candidates to a lack of graduates in the pipeline.") and has to call into question your accusation of misquoting ("You're misreporting the very link you quoted.")

The OP's comment seems closer aligned to what Damore's actual argument was (as opposed to your interpretation).

Re: We ran the numbers, and there really is a pipeline problem in engineering hiring

#230

Post by hkmurakami » Tue, Dec 03, 2019, 10:54 PM UTC

HeroOfAges wrote: post #161

kaikai wrote: post #149

Earlier quoted context omitted.

Inexperience doesn't equal unintelligence. Making sure that your intro class structure supports people with varying experience levels means you're more likely to engage people with the right interest and capabilities, not less so.

I've never experienced or witnessed a scenario in which trying to support varying levels of experience didn't result in lowering of standards. People with the right interest and capabilities will strive to clear the bar and will not expect the bar to be lowered for them.

There's the "math majors' math track" and the non math majors' math track at most institutions. Even within the majors track, there's separation between experience levels and intended tracks within the discipline. Ex:

MAT215-217: Recommended for math majors, some prior experience in constructing formal proofs is useful but not required.

MAT216-218: Accelerated math major sequence, for students with substantial background in university-level proof-based analysis courses.

MAT214-204-203: Alternative path to majoring in math, with a more algebraic introduction to proofs

MAT203-204-215: Alternative path to majoring in math; good path for the applied math certificate.

https://www.math.princeton.edu/undergraduate/placement/seque...

Lumping all these students into one general course would be a disservice to everyone, like you say, and should be spilt up.