3 ms·

> so the company can say they have a process that is objective. I think it makes sense to spend some money on this kind of charlatan "rigor" from a legal CYA p

> so the company can say they have a process that is objective. I think it makes sense to spend some money on this kind of charlatan "rigor" from a legal CYA perspective. But I think the amount of energy and time and resources spent on it far exceeds what is reasonable for CYA, and that money would only be spent the way it is if either/and/or 1) the people making the decisions to do all this extra analysis benefit from it personally in some way, or 2) the people making the decisions truly believe that the data/analysis will yield productive results, or 3) that the people making the decision believe that this much charlatan-rigor is actually necessary for legal "CYA" purposes. I just feel that #2 and #3 reflect such a broad lack of understanding that I'm surprised there are no major corporate leaders bucking the trend. For hiring, specifically, I believe it's proven in mathematical theory that the best thing you can do is to have a two-stage hiring process: 1) only eliminate/select candidates based on provably strong signals. This will result in a massively large pool of candidates who "pass" the screening for each position. 2) use a random number generator to pick someone from that pool. A lot of people would propose using the algorithm from the secretary problem[0], but that algorithm assumes that you can "grade" on a candidate's quality on a continuous scale. I don't believe we have the ability to judge a candidate that is "60% likely to be amazing" vs. "80% likely to be amazing", so that algorithm won't work. We can, however, measure in a binary way whether we were happy or unhappy with a hiring decision, and determine which pre-hiring signals had any actual strong correlation to being happy with the hiring decision. Filter candidates based on that, then pick a candidate at random. Any additional filtering will result in systemically picking worse candidates at scale. 0: https://en.wikipedia.org/wiki/Secretary_problem https://en.wikipedia.org/wiki/Secretary_problem

throwway120385 3y ago
Consider for a moment that HR both identifies the rules by which a person's value to an organization is decided, and that they also must demonstrate their own value to an organization and you can see that a lot of this arises from a conflict of interest. HR really wants 2 to be true because many companies have drunk the "metrics" kool-aid and believe that the only things that matter are measurable. Metrics are a technical solution, and as organizations are social constructs HR is making the classic software engineer mistake and trying to solve a social problem(how do we identify high performers) with a technical solution(generate numbers and use those to identify high performers). The reality is that team performance doesn't yield to analytic techniques because it's so markedly a result of all of the people together and not specific individuals that it's not even funny.

throwway120385 3y ago

Consider for a moment that HR both identifies the rules by which a person's value to an organization is decided, and that they also must demonstrate their own value to an organization and you can see that a lot of this arises from a conflict of interest. HR really wants 2 to be true because many companies have drunk the "metrics" kool-aid and believe that the only things that matter are measurable. Metrics are a technical solution, and as organizations are social constructs HR is making the classic software engineer mistake and trying to solve a social problem(how do we identify high performers) with a technical solution(generate numbers and use those to identify high performers). The reality is that team performance doesn't yield to analytic techniques because it's so markedly a result of all of the people together and not specific individuals that it's not even funny.