5 ms·

Google has a separate compensation team/committee as well as separate hiring commitee, so the hiring manager doesn't have a say in this. What I don't know is h

Google has a separate compensation team/committee as well as separate hiring commitee, so the hiring manager doesn't have a say in this. What I don't know is how much the hiring is 'blind', if they're aware of the resume or if they just get the candidates' interview performance. This seems like a good case to make all of it blind. Once a candidate interviews they'd just be given an ID, and the aforementioned committees wouldn't know their gender/race/age/experience level at all. It would likely prevent some bias, as well as protect the company in lawsuits.

withinboredom 3y ago
The thing is, you can make a pretty good guess on gender just from word-choice in a resume[1]. It would need to be 'sanitized' or 'summarized' by a neutral 3rd party before given to the committee. Edit: this is probably a bad idea. A cursory search on 'gender effects on translations' shows that this would make it even worse (assuming that this is effectively translation).
- burnished 3y ago
  You forgot to include the link implied by [1] Was it that study done on words known/unknown by gender? Interesting stuff
calderwoodra 3y ago
Google used to do this - they worked very hard to remove all gender signals, and as a result, pay discrimination against women got worse. So now they no longer try to hide gender.
- bhelkey 3y ago
  Could you provide a source for this?
- tester756 3y ago
  What was reasoning behind it getting worse?
  - calderwoodra 3y ago
    I'm not sure - it was just a well known internal experiment that didn't have the intended result.
  - GreedClarifies 3y ago
    Seriously?
  - xienze 3y ago
    Not sure if the poster is talking about Google’s hiring practices in this case, but similar things have been tried during hiring elsewhere and quietly abandoned for approximately the same reason, because it didn’t result in more female hires. The reason is because the typical resume pipeline for an engineering position is roughly 90/10 male/female. If you take away anything indicating gender during the hiring process, in general you will end up with... a 90/10 mix. Which is decidedly worse than what these progressive companies are hoping for. The only real way to gain a more favorable ratio is to, well, introduce bias towards females in the hiring process (to overcome the pipeline factor). It’s why these blind hiring/review practices always fail — they don’t give the expected results.
    - VirusNewbie 3y ago
      > you take away anything indicating gender during the hiring process, in general you will end up with... a 90/10 mix What? That makes no sense. When I put up a job listing and I get 90% resumes of people who are under qualified, and 10% that are qualified, that doesn't skew my hire rate of under qualified people.
      - xienze 3y ago
        But you’re talking about something completely different. You can judge bad/good resumes in the absence of information about the applicant’s gender. If your goal is (for example) to achieve a 65/35 male/female hiring ratio and your pool of resumes is 90/10 male/female, you will find it very difficult to do so in a completely blind manner, because you _don’t_ have any distinguishing information with which to pare down the pool in a positive manner anymore. For example: 500 total resumes for one position. 90% are unqualified => 50 good ones. Assuming equal representation here, 45 of those resumes will be from men, 5 will be from women. You narrow it down to say 25 resumes for initial online screens (no voice chat so you can keep things gender-neutral!). Call it 22/3 male/female at this point. Second round, online again (there’s even been companies that have done voice changing during interviews so they can have real conversations!). Half get cut. Let’s call it 11/2 male/female now. How likely do you think it is that a man will get the position versus a woman? How likely do you think you’ll be to achieve a “good” gender balance in hiring over time using this strategy?
polishTar 3y ago
I'm not convinced it's a bad thing for hiring or compensation committees to consider relevant experience/education. Agreed on gender/race/age. That should be made as blind as feasibly possible.
- VirusNewbie 3y ago
  Why would someone's education be relevant, the point of the interview is to determine if they're actually educated on the relevant topics.
dmazzoni 3y ago
Overall, Google has a tendency to under-level on hire. If they need a leader for a small engineering team and they're not sure the candidate has L6 experience, they'll hire at L5 and let them get promoted if they exceed expectations. Far better to do that than to hire at L6 and then have them fail to meet expectations. But, I agree that there are fewer checks and balances when it comes to leveling. The hiring committee is going to recommend a conservative level based on interview feedback and resume - they don't know the context of other peers in that particular org. It's up to the manager to push back or fight for the candidate if they're hired and outperforming (they can be re-slotted quickly if needed). Point being: there are more checks and balances to ensure that women aren't being rejected more often than men, given similar background and interview performance - but it's not clear there are checks and balances to prevent women from being systematically under-leveled. Another issue: the levels in this lawsuit are pretty high - L8 is equivalent to director, L9 is equivalent to senior director. We're talking about the top 0.1% of the company in terms of seniority. The committees here would all be L9 and above, so basically all senior leadership in the company, which is less diverse than the company as a whole. Also, not nearly as many people are hired at this level, so the committees won't have nearly as many numbers to crunch. My personal feeling here, having worked at Google, is that Google does far more than the average company to try to make hiring and compensation fair. That doesn't mean there aren't mistakes sometimes - but it's because individuals are still biased, and company-level efforts to minimize bias overall don't eliminate it at the individual level. I have a very easy time believing that this particular individual was underpaid. That doesn't mean most women at Google are underpaid.

withinboredom 3y ago

The thing is, you can make a pretty good guess on gender just from word-choice in a resume[1]. It would need to be 'sanitized' or 'summarized' by a neutral 3rd party before given to the committee. Edit: this is probably a bad idea. A cursory search on 'gender effects on translations' shows that this would make it even worse (assuming that this is effectively translation).

burnished 3y ago
You forgot to include the link implied by [1] Was it that study done on words known/unknown by gender? Interesting stuff

calderwoodra 3y ago

Google used to do this - they worked very hard to remove all gender signals, and as a result, pay discrimination against women got worse. So now they no longer try to hide gender.

bhelkey 3y ago
Could you provide a source for this?
tester756 3y ago
What was reasoning behind it getting worse?
- calderwoodra 3y ago
  I'm not sure - it was just a well known internal experiment that didn't have the intended result.
- GreedClarifies 3y ago
  Seriously?
- xienze 3y ago
  Not sure if the poster is talking about Google’s hiring practices in this case, but similar things have been tried during hiring elsewhere and quietly abandoned for approximately the same reason, because it didn’t result in more female hires. The reason is because the typical resume pipeline for an engineering position is roughly 90/10 male/female. If you take away anything indicating gender during the hiring process, in general you will end up with... a 90/10 mix. Which is decidedly worse than what these progressive companies are hoping for. The only real way to gain a more favorable ratio is to, well, introduce bias towards females in the hiring process (to overcome the pipeline factor). It’s why these blind hiring/review practices always fail — they don’t give the expected results.
  - VirusNewbie 3y ago
    > you take away anything indicating gender during the hiring process, in general you will end up with... a 90/10 mix What? That makes no sense. When I put up a job listing and I get 90% resumes of people who are under qualified, and 10% that are qualified, that doesn't skew my hire rate of under qualified people.
    - xienze 3y ago
      But you’re talking about something completely different. You can judge bad/good resumes in the absence of information about the applicant’s gender. If your goal is (for example) to achieve a 65/35 male/female hiring ratio and your pool of resumes is 90/10 male/female, you will find it very difficult to do so in a completely blind manner, because you _don’t_ have any distinguishing information with which to pare down the pool in a positive manner anymore. For example: 500 total resumes for one position. 90% are unqualified => 50 good ones. Assuming equal representation here, 45 of those resumes will be from men, 5 will be from women. You narrow it down to say 25 resumes for initial online screens (no voice chat so you can keep things gender-neutral!). Call it 22/3 male/female at this point. Second round, online again (there’s even been companies that have done voice changing during interviews so they can have real conversations!). Half get cut. Let’s call it 11/2 male/female now. How likely do you think it is that a man will get the position versus a woman? How likely do you think you’ll be to achieve a “good” gender balance in hiring over time using this strategy?

polishTar 3y ago

I'm not convinced it's a bad thing for hiring or compensation committees to consider relevant experience/education. Agreed on gender/race/age. That should be made as blind as feasibly possible.

VirusNewbie 3y ago
Why would someone's education be relevant, the point of the interview is to determine if they're actually educated on the relevant topics.

dmazzoni 3y ago

Overall, Google has a tendency to under-level on hire. If they need a leader for a small engineering team and they're not sure the candidate has L6 experience, they'll hire at L5 and let them get promoted if they exceed expectations. Far better to do that than to hire at L6 and then have them fail to meet expectations. But, I agree that there are fewer checks and balances when it comes to leveling. The hiring committee is going to recommend a conservative level based on interview feedback and resume - they don't know the context of other peers in that particular org. It's up to the manager to push back or fight for the candidate if they're hired and outperforming (they can be re-slotted quickly if needed). Point being: there are more checks and balances to ensure that women aren't being rejected more often than men, given similar background and interview performance - but it's not clear there are checks and balances to prevent women from being systematically under-leveled. Another issue: the levels in this lawsuit are pretty high - L8 is equivalent to director, L9 is equivalent to senior director. We're talking about the top 0.1% of the company in terms of seniority. The committees here would all be L9 and above, so basically all senior leadership in the company, which is less diverse than the company as a whole. Also, not nearly as many people are hired at this level, so the committees won't have nearly as many numbers to crunch. My personal feeling here, having worked at Google, is that Google does far more than the average company to try to make hiring and compensation fair. That doesn't mean there aren't mistakes sometimes - but it's because individuals are still biased, and company-level efforts to minimize bias overall don't eliminate it at the individual level. I have a very easy time believing that this particular individual was underpaid. That doesn't mean most women at Google are underpaid.