> Look at the examples of CVs/books sent around with just the name/gender changed and the huge difference in response rate to see the bias.

That's not true. The original studies used low class names for blacks and hispanics, but not for whites. (For example, they didn't use names like Cletus or Billy Ray.) Once this was corrected for, they found almost no bias in resume screening.[1]

1. https://economics.missouri.edu/working-papers/2014/wp1419_ko...

So looking at their methodology, they chose the "high class" black names ("Anderson" and "Thompson") (EDIT: I misread, these aren't the black names) based on census proportions, but that doesn't necessarily mean those names are actually read as "black" by most people.

Moreover, using census results ignores exposure to foreigners with these names. My own strongest association for "Thompson" is the white police officers in Tintin, and more generally for white English people. (Although I'm British so that may not be representative.)

As long as "low class" black names are more readily associated with blackness than "high class" ones, this methodology isn't really isolating anything, just doing a slightly more intersectional analysis.

You misread the paper. The surnames Anderson and Thompson have white connotations, while Jefferson and Washington have black connotations. The researchers backed up their choices with data. From page 3:

> For example, based on data from the United States Census, 90 and 75 percent of individuals with Washington and Jefferson surnames are African American, respectively. Similarly, 90 percent of individuals with a surname of either Hernandez or Garcia are Hispanic, and 70 percent of Andersons and Thompsons are white.

Ah yes, you're correct, I got confused by the first-name/surname thing.

I would still question the validity of this approach, and more generally, I don't think you can meaningfully isolate class and race, since they're so inter-related in the US.

They acknowledge that the names may not be strong signals, but the very reason that's the case is that people are less likely to associate black people with high socioeconomic status in the first place.

Even if you object to their choices of black surnames, they also ran experiments with male and female names, and with hispanic names. In those cases, resume screeners were all but certain as to the race and sex of the applicant. Still, there was almost no discrimination detected.

Honestly, I feel like I'm playing whack-a-mole here. You started by claiming that there was a massive gender & race bias:

> CVs/books sent around with just the name/gender changed and the huge difference in response rate to see the bias.

When I corrected that and linked to a study, you replied with an argument based on a misreading. When I corrected your misreading, you fell back to, "I don't think you can meaningfully isolate class and race, since they're so inter-related in the US." If that's the case, then it also refutes the conclusions of the original (flawed) studies claiming racism. After all, if your argument is true, those can be just as easily explained as classism.

I'm not sure there is anything that could convince you that you are mistaken.

I don't think that shows a lack of racism, rather the existance of classism.

People are being unfairly judged based on racial and class identifiers, rather than on their CV.

The study I linked to shows that there's almost no race or sex discrimination for names on resumes, even in the southeast US. Regarding class discrimination: I don't think you'll get much disagreement from reasonable people. It's ubiquitous. That said, I do think classism is far less common in tech than in most fields.

The emily and jamal paper controlled for mother's education level distributions.

This doesn't disprove the original contention.

From an organisation’s point of view, does it matter how one person came to be a better performer than another? Surely they just care about current state: this person performs better than this other person now.

Perhaps had their backgrounds been different, their performance would be different. Perhaps if the under-performer had more training, their performance would be better.

But right now, things aren’t different. From the organisation’s self-interested point of view, they should just choose the better performing candidate.

This only makes sense if you assume the organisation's interest is only to maximise productivity in the short term.

An open source organisation may explicitly be interested in attracting marginalised people into the tech sector, to improve equality in society as a whole.

Or even from a cynical perspective, a profit-led corporation may desire to increase their ability to hire competent but undervalued (and thus cheaper) employees in future.