It's insane to me that the automatic conclusion from the above is that organisation B is racist.

You just went from "Org B likely has some issues" to "automatic conclusion."

No, I actually went off the first sentence:

"So which report makes an organization less racist"

So that implies that org B is more racist(because it hired zero brown people) - is that not the conclusion the author of the comment wants us to reach?

* * *

If I say that my uncle is more racist than my cousin, do you think my uncle is racist, or not?

You roll a die 120 times, get 20 1's, 20 2's, 0 3's, 20 4's, 40 5's, and 20 6's.

Do you or do you not think it is incredibly likely this is an unfair die.

That's Company B.

If your company takes a pile of CVs and randomly picks a bunch of people to hire, then yes, you are absolutely correct, there is something very very wrong with company B if a truly random process results in an unequal distribution of hires.

But hiring isn't random. Maybe all the "brown" candidates lacked necessary qualifications. Or needed visas which they couldn't get. Or a million other reasons other than "org B is racist".

My point is: you need further context. To look at that one stat as given above and conclude org B is racist is....unwise.

Is that how you hire? You litearlly roll a dice for an interview? No bias of "This guy looks / acts / talks like me?"

I am not sure I have ever been on a team without some kind of like-me bias. It's real hard.

This is a very inappropriate analogy, because humans are very diverse; the exact opposite of a die's faces.

We’re not discussing f a human. We’re talking about fairly coarse demographic buckets

...or that Org A is "less racist", as I've demonstrated. Poor analogy by the OP.

>Org B likely has some issues

Or in the exact time slice you got the data there weren't any brown applicants.

10% out of how many?

> So which report makes an organization less racist:

> Org A) 10% of applicants are brown, 9% of hires are brown

That means nearly all brown applicants for got hired. You're probably thinking this is a "good stat" and therefore a "good employer". But what if the rest of that company's stats were "90% of applicants are white, 80% of hires are white." Is that still a "good stat/good company"? According to you, it should be (and I would tend to agree.) In both cases, we're pretty much showing "If you apply for a job here, you have a ~90% chance of getting in, regardless of skin color." Not bad!!

But wait. What if the demographic of applicants is more like 50% white and 50% brown. (Assume a non-remote workplace). Now, one has to wonder why, in a demographic where half the population is brown, why only 10% are applying to this organization? Could be many reasons. Culturally, maybe the brown folds simply choose not to apply here. Or maybe the organization has a history of mistreating brown folks, abusing them, paying them less, etc. That could explain the low rate of applicants here. Your good example of a "less racist" org doesn't look so good anymore. Many factors could be a play; some less intentionally-evil than others. Hard to draw conclusions simply by reporting on one stat in isolation.

> Hard to fix what you don't measure.

Yes, but harder to think critically about the data you do gather.

> That means nearly all brown applicants for got hired.

Nope, the denominator is different. If you select the same percentage from each category the proportions stay the same.

If there are 100 fruits where 10 are apples and 90 are pears and each fruit has 10% chance to be selected then you get on average 1 apple and 9 pears => 10% of the selected fruits are apples.