3 ms·

I didn't touch gender because it's an equally long post with an entirely different set of base rate figures. With gender however, the base rates are more direc

I didn't touch gender because it's an equally long post with an entirely different set of base rate figures. With gender however, the base rates are more directly influence by society's treatment of gender in the many years between being born and arriving at the point where you start looking for a job. Girls are treated differently by adults and expected and encouraged to take interest in different things. I think that's unfortunate and I hope that changes. That said, it doesn't do anyone any good when people complain about a problem and only focus on a single contributing factor like racism and sexism instead of scientifically and methodically brainstorming all possible causes and contributing factors, measuring the magnitude of impact of each and then determining which of all the contributory factors provide the most bang for the buck in changing. There is an entire discipline for improving things you can measure. It's called statistical process control and its a much smarter way of going about improving things than debates we have today where people arguing from either side have no measurements or statistics or a poor understanding about how to measure things and use statistics or some combination of both. === As someone with a strong interest in the social history of how the hardware and software industries evolved since WWII, I've provided some other possible explanations, in addition to sexism, that likely contributed to the gender imbalance we see today: https://news.ycombinator.com/item?id=10928920 https://news.ycombinator.com/item?id=10928920 https://news.ycombinator.com/item?id=10929042 https://news.ycombinator.com/item?id=10929042 The big questions that remains in my mind after the observation I made in that second post are: - "Are there conditions that when afforded to boys and girls equally, generate equal interest in computer science and engineering in people regardless of where they lie on the gender spectrum?" - "If there are equal conditions that foster equal interest, does that equal condition generate maximal interest from anyone on the gender spectrum, or should we consider a offering a variety so that greater interest if fostered by pairing a person with the environment that is optimal for him her or anything in between?" - "If no condition exists where you can generate equal interest, how do you design two or more separate set of conditions that approximates equality by being equally equitable without necessarily being truly equal. === on race and ignore gender, which from what I've read is the much bigger problem in modern American tech I'm just curious how you arrived at that conclusion or where you got that conclusion from? I ask because it's strange to me to compare the "magnitude" of those two problems because to me those are apples and oranges and incomparable. You may measure both with the same final figure, but the moment you start investigating causes, you end up with more incomparable contributing factors than comparable ones. === Re: Gapjumpers Now that I've googled "gapjumpers" and seen what it is, I'm very inclined to give far less credence to the claims of nickpsecurity because I now know that gapjumpers is a startup selling a product based on their claims and not some non-profit funding independent academic resource with all the hallmarks of good science. I was really hoping for scholarly research and not an organization with a clear financial conflict of interest that is very likely to be hyperbolic or suffer from confirmation bias. I even went a bit farther and decided to look at some of their jobs and the blind audition task. Here are two engineering ones: https://www.gapjumpers.me/questions/sendgrid/qs-224/ https://www.gapjumpers.me/questions/sendgrid/qs-224/ https://www.gapjumpers.me/questions/bbc-digital/qs-256/ https://www.gapjumpers.me/questions/bbc-digital/qs-256/ Those have 32 and 36 applicants respectively. While I have no doubt that their process helps create a more equal process, what is see isn't suggestive of a large sample size to make firm claims from like nickpsecurity did. The figures on their home page are 54%/44% female to male. You can get at these figures with 9 females and 8 males. There is no indication that these results scale to dozens to hundreds of employees. If there were just 17 hired employees for those figures, the margin of error would be mega high (+/- 23.77%) margin. Furthermore, this hiring process is likely to suffer from confirmation bias, since those with the greatest interest in an unbiased process are those that hold the believe that they are most disadvantaged by the bias. This means that the pipeline of applicants into their process isn't representative of the base rates of available candidates in our industry currently. I'm really not trying to tear their startup apart because I think they are part of the solution, but bad science, bad statistics and unsubstantiated or misleading claims do deserve to be challenged and torn apart all day long. === Re: "not all men" Generalizing ... is the reason that "not all men" has become a nasty cliche. While I won't argue the merit of the "not all men" position (which is no more fallacious that the argument to which it is a response), I would like to point out that there is a certain irony that those that complain about others using that defense are engaged in generalizing themselves. Basically, someone makes a generalization (all/most X are), which is rebutted with another generalization (not all X are). Essentially this boils down to the immature playground argument of "Are too!", "Are not!", "Are too!", "Are not!", ad nauseam. "Hello Pot, this is Kettle calling..." Those who make generalizations about white people or men are as guilty as those who make generalizations about non-white people or women (or any other discretely identifiable group on the gender spectrum).

nickpsecurity 11y ago
"Now that I've googled "gapjumpers" and seen what it is, I'm very inclined to give far less credence to the claims of nickpsecurity because I now know that gapjumpers is a startup selling a product based on their claims and not some non-profit funding independent academic resource with all the hallmarks of good science. " And yet you've not cited a single, scholarly treatment of anything you're claiming. You immediately moved to an anecdote to save yourself time and facilitate explanation. Like I did. Upon request and when I found time, I dropped around 3 hours digging up old and some new references that supported my claims including scholarly work and experimental studies. Should I doubt all your future claims because you saved yourself time or didn't drop a Pastebin-sized set of references and commentary in your first comment? I think that would be a bit unfair, don't you? ;)
- malandrew 11y ago
  To be fair, I really should have phrased that as: "Now that I've googled "gapjumpers" and seen what it is, I'm very inclined to give far less credence to the claims of based on work by gapjumpers" The reason I made the jump to putting your name in their is because gapjumpers was your primary reference in a few comments, but no link was provided. I have few doubts that a blind interview process of capable candidates will approximate the base rate of available candidates that participate in the interview process (or mock hiring process for the purpose of demonstrating that a blind interview process eliminates or greatly reduces identity-based biases). I was never debating this claim. Providing references to this claim is tangential to the point I was making. I was debating your original claim. Your original claim: It all looks nice until we get numbers from groups like GapJumpers that institute blind auditions for hiring or performance evaluations where evaluator can't identify what group they are. Starting out, these "fair" companies are usually under 10% minority or women. Once they see only performance, it's suddenly closer to 50/50. Hmm. From what you wrote, I think it's fair to assume that companies exist that implemented blind auditions that are 50/50. Furthermore, since this debate is in the context of tech and because this was a reply to my comment about base rates, I also think it's fair to assume that the companies in question are tech companies. I'm also making a minor assumption that we're talking about engineering positions at these companies since that is where most of the debate regarding the gender ratio is centered at. For non-engineering positions, I've not noticed a huge discrepancy in hiring. At the tech companies where I've worked, the non-technical side of the business is very balanced gender-wise. FWIW the technical recruiting departments responsible for sourcing most of the engineering candidates we interview are mostly women. I don't know if that's important at all, but it's an interesting observation. With those two fair assumptions, both me and the other commenter considered your claim as written to be a very very bold claim. Your claim was tantamount to "blind auditions are a panacea to the gender (or other identity) diversity problem in hiring". Given that claim, that's was we want a citation for. Where are these mythical 50/50 tech companies and how to they achieve 50/50 without being discriminatory when the base rate they are drawing from is approximately 88/12 (the approximate distribution of men to women with CS degrees)? Given this, the "scholarly work and experimental studies" you found are interesting, but not really relevant because I never once challenged what blind auditions can achieve in an academic or experimental setting. I wanted a citation that references the "fair" companies you mentioned. because I was challenging that blind auditions can automagically achieve 50/50 without reverse discrimination in an industry where the population from which you draw candidates is far from 50/50. Had you limited your claim to academic and experimental studies, I wouldn't have challenged it at all. I probably wouldn't even had asked for a reference, since that's not even an extraordinary claim. With respect to your other comments. I'm not sure which claims of mine in particular you're addressing or what anecdote in particular you're talking about. I wrote a lot across several comments. The original claim I made, which was about the relevance of base rates with respect to race didn't require a scholarly citation since I cited the data directly, much of which was US Census data, which is as authoritative as you get for the claim I'm making. For the other references I used, I could try to find the original source material instead of secondary references. Let me know which secondary references you have doubts about and I'd be happy to scrounge up the primary sources so we can debate the likelihood that they are valid. FWIW, I'm 100% in favor of blind auditions. I would love to be able to reduce a candidate's identity down to a hash. I miss the internet back in 1995/6 when the only identity you had online was the one you created for yourself.

nickpsecurity 11y ago

"Now that I've googled "gapjumpers" and seen what it is, I'm very inclined to give far less credence to the claims of nickpsecurity because I now know that gapjumpers is a startup selling a product based on their claims and not some non-profit funding independent academic resource with all the hallmarks of good science. " And yet you've not cited a single, scholarly treatment of anything you're claiming. You immediately moved to an anecdote to save yourself time and facilitate explanation. Like I did. Upon request and when I found time, I dropped around 3 hours digging up old and some new references that supported my claims including scholarly work and experimental studies. Should I doubt all your future claims because you saved yourself time or didn't drop a Pastebin-sized set of references and commentary in your first comment? I think that would be a bit unfair, don't you? ;)

malandrew 11y ago
To be fair, I really should have phrased that as: "Now that I've googled "gapjumpers" and seen what it is, I'm very inclined to give far less credence to the claims of based on work by gapjumpers" The reason I made the jump to putting your name in their is because gapjumpers was your primary reference in a few comments, but no link was provided. I have few doubts that a blind interview process of capable candidates will approximate the base rate of available candidates that participate in the interview process (or mock hiring process for the purpose of demonstrating that a blind interview process eliminates or greatly reduces identity-based biases). I was never debating this claim. Providing references to this claim is tangential to the point I was making. I was debating your original claim. Your original claim: It all looks nice until we get numbers from groups like GapJumpers that institute blind auditions for hiring or performance evaluations where evaluator can't identify what group they are. Starting out, these "fair" companies are usually under 10% minority or women. Once they see only performance, it's suddenly closer to 50/50. Hmm. From what you wrote, I think it's fair to assume that companies exist that implemented blind auditions that are 50/50. Furthermore, since this debate is in the context of tech and because this was a reply to my comment about base rates, I also think it's fair to assume that the companies in question are tech companies. I'm also making a minor assumption that we're talking about engineering positions at these companies since that is where most of the debate regarding the gender ratio is centered at. For non-engineering positions, I've not noticed a huge discrepancy in hiring. At the tech companies where I've worked, the non-technical side of the business is very balanced gender-wise. FWIW the technical recruiting departments responsible for sourcing most of the engineering candidates we interview are mostly women. I don't know if that's important at all, but it's an interesting observation. With those two fair assumptions, both me and the other commenter considered your claim as written to be a very very bold claim. Your claim was tantamount to "blind auditions are a panacea to the gender (or other identity) diversity problem in hiring". Given that claim, that's was we want a citation for. Where are these mythical 50/50 tech companies and how to they achieve 50/50 without being discriminatory when the base rate they are drawing from is approximately 88/12 (the approximate distribution of men to women with CS degrees)? Given this, the "scholarly work and experimental studies" you found are interesting, but not really relevant because I never once challenged what blind auditions can achieve in an academic or experimental setting. I wanted a citation that references the "fair" companies you mentioned. because I was challenging that blind auditions can automagically achieve 50/50 without reverse discrimination in an industry where the population from which you draw candidates is far from 50/50. Had you limited your claim to academic and experimental studies, I wouldn't have challenged it at all. I probably wouldn't even had asked for a reference, since that's not even an extraordinary claim. With respect to your other comments. I'm not sure which claims of mine in particular you're addressing or what anecdote in particular you're talking about. I wrote a lot across several comments. The original claim I made, which was about the relevance of base rates with respect to race didn't require a scholarly citation since I cited the data directly, much of which was US Census data, which is as authoritative as you get for the claim I'm making. For the other references I used, I could try to find the original source material instead of secondary references. Let me know which secondary references you have doubts about and I'd be happy to scrounge up the primary sources so we can debate the likelihood that they are valid. FWIW, I'm 100% in favor of blind auditions. I would love to be able to reduce a candidate's identity down to a hash. I miss the internet back in 1995/6 when the only identity you had online was the one you created for yourself.