Let’s try the experiment. Two startups short on cash. One hires your way and one does not. Which one do you think is more likely to survive?
That's not the point though. The point is to remove bias.
It's easy to solve for one variable; beyond that, one has to balance things so that primary goals (hiring capable and personable talent) are still achieved.
"We want the best talent."
"No we don't. We want the to remove bias."
"Isn't talent more important?"
"Removing bias is finding the best talent."
That was like reading a conversation with an AI chatbot. We all agree we want the best talent then, yes?
So asking whether removing bias is "more important" than improving decision making is a non-sequitar: removing bias is a tool towards that goal. In a sense it's "less important" because other tools might do the job better, and in a sense it's more important because simply desiring to improve decision-making does nothing without tools to implement that desire.
That said, I wouldn’t blank out schools and employers. Discrimination by school and employer may or may not be wrong for the business. But those aren’t protected categories.
The problem is that those institutions aren’t perfect in their hiring/accepting either, and some initial mistake by the college is propagated and amplified by subsequent decisions deferring to it.
No idea how to solve this. Maybe the data could be coded into some numeric or ordinal value. I bet the actual terms like “Harvard” and “MIT” are more impressive than a completely equivalent abstraction such as “Ivy League” or “top tier”.
https://www.abc.net.au/news/2017-06-30/bilnd-recruitment-tri...
"The trial found assigning a male name to a candidate made them 3.2 per cent less likely to get a job interview.
Adding a woman's name to a CV made the candidate 2.9 per cent more likely to get a foot in the door.
"We should hit pause and be very cautious about introducing this as a way of improving diversity, as it can have the opposite effect," Professor Hiscox said."
It was fashionable in France for some time, and it was even attempted. It backfired however. Researchers realized that hires were less diverse when using anonymized resume.
> replace experience on resumes with "random corp 1", random corp 2, random school 1 etc.
That would remove one of the best predictors of future performance. Especially in CS, not all internships or schools are created equal.
Now, with all the "affirmative action" going on, these should be interpreted with a grain of salt. Asians, for example, need much better scores to get into Ivies than some other minorities. And YouTube was caught trashing all white and Asian applicant’s resumes a while ago [0].
[0] https://www.theverge.com/2018/3/2/17070624/google-youtube-wi...
These articles always pull the same tricks. They invent a new racial category, "white or Asian", to avoid stating that only 61% were white - i.e. slightly underrepresented compared to the 63.7% white 2010 US demographics [1]. Although by 2014 the white % had probably dropped enough that they are neither over- nor under-represented at Google (the 2020 census reports non-Latino whites as only 57.8% of the US - a 5.9% drop in only 10 years [2]).
But you wouldn't know this from "91% white or Asian". Maybe they can make the problem go away entirely by saying Google is "32% Black or Asian"?
[1] https://en.wikipedia.org/w/index.php?title=Demographics_of_t...
[2] https://en.wikipedia.org/w/index.php?title=Demographics_of_t...
Especially with admission scandals at top schools, there seems to be an unspoken systemic bias against asians perpetuated by some activists.
I wouldn't go as far as what you're suggesting though, because verbal communication skills, etc, are a relevant skill for the job.
It’s famous for the ridiculous differences he policy change made. Women's success rates doubled overnight.
https://statmodeling.stat.columbia.edu/2019/05/11/did-blind-...
Blind auditions seem like a good idea regardless, don't get me wrong. But the headlines talking about 50% increases are probably wrong.
If there are particular attributes that you want to hire for, those should be in the job/person spec and candidates should be assessed against them.
I'm sure that kind of redacting names &c would help, but it wouldn't be perfect.
https://www.reuters.com/article/us-amazon-com-jobs-automatio...
I'm not sure that's a valid take.
It's my understanding that Amazon calibrated their machine learning model to answer the question of "help me find more people like the one we already have", and they just so happened to have within their ranks way more people with a specific academic and professional background, which was happened to refer to be men who did activities dominated by men.