5 ms·

>However, if 100% of these pressures were removed, would we have exactly 50% representation of men and women in tech? Many people appear to believe very strongl

>However, if 100% of these pressures were removed, would we have exactly 50% representation of men and women in tech? Many people appear to believe very strongly that this is true. What is the basis for this belief? We could have an even higher percentage of women, like we do in many top-notch classical outfits today. There is some history here I'm qualified to speak about as a woman who took many orchestral auditions in classical music. Most of decision makers in classical back in the day asked exactly what you are asking about tech right now. The results were stark. (see study below) My example is not intended to imply that tech is "the same" as classical music or anything--just pointing out a strong similarity in today's tech interviews to the gender/race-biased audition practices of orchestras in the pre-1970s, when it was believed that women/brown candidates were not as strong/accomplished/technically skilled/artistic/confident as their white European male colleagues. Here is a 1974 NYT article about "the change to screened auditions"; the reporter still felt comfortable calling Asian players "Oriental" at that time. http://www.nytimes.com/1974/10/15/archives/orchestra-auditions-the-old-systems-are-changing-system-is-changing.html http://www.nytimes.com/1974/10/15/archives/orchestra-auditio... The Met Orchestra is considered the overall "best orchestra" in the United States. It is a dream job in terms of pay and quality of music making. By best, I mean that winning a job at the met is like being a star player in the majors, more so than in any other orchestra. The Met's auditions are considered by classical players to be the most impartial, although they are still partial. They are the blindest, anyway, in terms of using a screen, if players can make it past the tape round. (that is another issue altogether for another time) Today, more than half of the NY Phil and 50% of the Met orchestra players are women, due to blinder audition practices. Still, most orchestras remove the screen after the first or second round today. (?) I believe the MET waits until the finals to remove the screen. (I didn't make it to the finals, but it is what I hear from players who have) I believe that tech could learn a great deal from classical music's hiring practice innovations of the 1970s and 80s. Today, most US Orchestras are comprised of at least 50% women, up from none or just a couple in the 1970s. Today, St. Louis is majority female. If anyone in tech is interested, they could dig into this study published in The American Economic Review http://www.academicroom.com/article/orchestrating-impartiality-impact-blind-auditions-female-musicians http://www.academicroom.com/article/orchestrating-impartiali...

haberman 9y ago
I had come across that study/story before, and consider it one of the great successes in combating discrimination that I have ever heard! Blinding the auditions is brilliant, because it isn't a "bias for bias" approach. It doesn't attempt to combat one bias by intentionally introducing another bias. It just removes the jury's ability to be biased, by concealing the musician's sex. Because of this, the results of that study are nearly unimpeachable. It's basically a perfect approach. If tech could do anything comparable to that, I would be 100% in favor.
- brooklyn_ashey 9y ago
  I believe they could! With all that technology? Shoot. All we had was a stage and a screen to work with. I will say that in another study it was found that the sound of women's dress shoes unconsciously biased juries. So we all (even men) walk to the stage on a carpet strip to mute that sound, or we play in stocking feet or soft shoes. Taking the screen away at the end is its only flaw. I will say something that a lot of non musicians may not realize, though. If you know a player (we all have known one another from childhood, in many cases) you can pick him/her out of 100 players just like you recognize someone's voice on the phone. So, there is always that bias. So, not 100% perfect, but better. If the conductor/principal player knows you and likes you...it helps...a lot.(and they are fairly likely to know you, at least) It doesn't matter what instrument they are playing either. I could recognize any player's playing I know by their kazoo playing, just like you can instantly hear the difference between Ray Charles and Stevie Wonder. Players whose playing is nondescript don't make it to the level of taking an orchestral audition, even poorly. I'm curious, is code at the highest levels the same? I'm not an accomplished enough programmer to know this. That would be an interesting difference between coding and playing music, but I suspect you could recognize a great programmer's coding style. Why aren't there totally blind coding interviews? I just don't get it! The science (I've read but can't remember the source now- if you know it, please share) says that in-person interviews are not helpful in helping to identify successful candidates, and are often harmful.
  - haberman 9y ago
    I can't speak to anything related to interviewing because I don't do interviews. Part of the reason I don't do interviews is because I have very low confidence in my ability to quickly tell who is good. Most of my colleagues that I respect the most (I work at Google) didn't make a strong impression on me when I first met them, or even within a few days. I feel like it takes days or weeks for me to get a sense of how much I like someone's work. So I definitely feel unsatisfied with the interviewing approach in general, but I don't have any practical ideas about how to improve it. "Work with someone for days or weeks first" isn't practical. I get your point that there still could be biases if you can recognize a musician's sound. I should have been more clear about what I meant by "perfect". What I meant is just that this intervention (adding screens to auditions) almost certainly didn't create any bias, it only removed it. It may not have removed 100% of it, but it didn't combat the bias by introducing a counter-bias. Can you recognize a coder's work by their style? Definitely to some extent. One aspect of working on a shared code-base with a whole bunch of engineers is that uniformity of style is a significant plus (makes code easier to read when it's consistent), so many of the differences that might show a "personal flare" are removed. But still, there is plenty that could belie a person's coding style. I think there are way too many programmers out there, however, for this to be an issue in interviews. Estimates say there are something like 2-4 million software engineers in the US. Compare this with orchestral musicians: it looks like about 60 orchestras in the US work 40+ weeks per year; if we estimate 100 players per orchestra this is 6,000 full-time orchestra spots in the USA (you might have better numbers on this). If you specialize this by instrument, it might be more like 240 full-time orchestral oboists in the whole country. It seems much more likely that you could recognize a specific person with those odds. (By the way, I'm a musician too. :) Though not in the orchestral world, and not full-time. I'm a singer (small ensembles mainly), though I did a lot of collaborations with orchestras as a boy growing up in Boychoir).

haberman 9y ago

I had come across that study/story before, and consider it one of the great successes in combating discrimination that I have ever heard! Blinding the auditions is brilliant, because it isn't a "bias for bias" approach. It doesn't attempt to combat one bias by intentionally introducing another bias. It just removes the jury's ability to be biased, by concealing the musician's sex. Because of this, the results of that study are nearly unimpeachable. It's basically a perfect approach. If tech could do anything comparable to that, I would be 100% in favor.

brooklyn_ashey 9y ago
I believe they could! With all that technology? Shoot. All we had was a stage and a screen to work with. I will say that in another study it was found that the sound of women's dress shoes unconsciously biased juries. So we all (even men) walk to the stage on a carpet strip to mute that sound, or we play in stocking feet or soft shoes. Taking the screen away at the end is its only flaw. I will say something that a lot of non musicians may not realize, though. If you know a player (we all have known one another from childhood, in many cases) you can pick him/her out of 100 players just like you recognize someone's voice on the phone. So, there is always that bias. So, not 100% perfect, but better. If the conductor/principal player knows you and likes you...it helps...a lot.(and they are fairly likely to know you, at least) It doesn't matter what instrument they are playing either. I could recognize any player's playing I know by their kazoo playing, just like you can instantly hear the difference between Ray Charles and Stevie Wonder. Players whose playing is nondescript don't make it to the level of taking an orchestral audition, even poorly. I'm curious, is code at the highest levels the same? I'm not an accomplished enough programmer to know this. That would be an interesting difference between coding and playing music, but I suspect you could recognize a great programmer's coding style. Why aren't there totally blind coding interviews? I just don't get it! The science (I've read but can't remember the source now- if you know it, please share) says that in-person interviews are not helpful in helping to identify successful candidates, and are often harmful.
- haberman 9y ago
  I can't speak to anything related to interviewing because I don't do interviews. Part of the reason I don't do interviews is because I have very low confidence in my ability to quickly tell who is good. Most of my colleagues that I respect the most (I work at Google) didn't make a strong impression on me when I first met them, or even within a few days. I feel like it takes days or weeks for me to get a sense of how much I like someone's work. So I definitely feel unsatisfied with the interviewing approach in general, but I don't have any practical ideas about how to improve it. "Work with someone for days or weeks first" isn't practical. I get your point that there still could be biases if you can recognize a musician's sound. I should have been more clear about what I meant by "perfect". What I meant is just that this intervention (adding screens to auditions) almost certainly didn't create any bias, it only removed it. It may not have removed 100% of it, but it didn't combat the bias by introducing a counter-bias. Can you recognize a coder's work by their style? Definitely to some extent. One aspect of working on a shared code-base with a whole bunch of engineers is that uniformity of style is a significant plus (makes code easier to read when it's consistent), so many of the differences that might show a "personal flare" are removed. But still, there is plenty that could belie a person's coding style. I think there are way too many programmers out there, however, for this to be an issue in interviews. Estimates say there are something like 2-4 million software engineers in the US. Compare this with orchestral musicians: it looks like about 60 orchestras in the US work 40+ weeks per year; if we estimate 100 players per orchestra this is 6,000 full-time orchestra spots in the USA (you might have better numbers on this). If you specialize this by instrument, it might be more like 240 full-time orchestral oboists in the whole country. It seems much more likely that you could recognize a specific person with those odds. (By the way, I'm a musician too. :) Though not in the orchestral world, and not full-time. I'm a singer (small ensembles mainly), though I did a lot of collaborations with orchestras as a boy growing up in Boychoir).