Let's take the immediately preceding example, where the author had a black woman as a supervisor. The author describes this as 'diversity lightning'. If you interpret this narrowly, this makes absolutely no sense: there's no way in which a single person can be considered diverse. On the other hand, if you think about it at an environmental or company level, things look different.
Suppose we take a random US tech company's employees and dump them in a vase. (Let's ignore the ethical issues with this for a moment.) At this point, we'd expect the odds of a random draw to reflect the tech industry's demographics, which is to say, predominantly white and asian males. If you draw this 'predominantly black' development team from the vase, though, you need to update your priors. You probably shouldn't calculate your posterior as 'everyone at this company is black' -- but you'd still probably guess that you're looking at a tech company where there are more black people and women than average. Which is to say, you'd expect that the company as a whole was more diverse.
I think this second reading is a lot more charitable to the author. I also think it's the more interesting reading; the demographic representation in a small group is subject to a huge amount of randomness, but the demographics of a entire organization are more telling, and can reveal more interesting things about policy and bias. As a community, software people are normally biased toward systems-level thinking; it's weird to me that in this particular issue people have this laser-focus on small groups and tokenism.
That is a more interesting interpretation. I think it's more of a steel man [1] than a charitable interpretation, but that's absolutely worth bringing up.
Let's take the immediately preceding example, where the author had a black woman as a supervisor. The author describes this as 'diversity lightning'. If you interpret this narrowly, this makes absolutely no sense: there's no way in which a single person can be considered diverse. On the other hand, if you think about it at an environmental or company level, things look different.
Suppose we take a random US tech company's employees and dump them in a vase. (Let's ignore the ethical issues with this for a moment.) At this point, we'd expect the odds of a random draw to reflect the tech industry's demographics, which is to say, predominantly white and asian males. If you draw this 'predominantly black' development team from the vase, though, you need to update your priors. You probably shouldn't calculate your posterior as 'everyone at this company is black' -- but you'd still probably guess that you're looking at a tech company where there are more black people and women than average. Which is to say, you'd expect that the company as a whole was more diverse.
I think this second reading is a lot more charitable to the author. I also think it's the more interesting reading; the demographic representation in a small group is subject to a huge amount of randomness, but the demographics of a entire organization are more telling, and can reveal more interesting things about policy and bias. As a community, software people are normally biased toward systems-level thinking; it's weird to me that in this particular issue people have this laser-focus on small groups and tokenism.