I think it's worth considering the following points:

* The critics are talking about Software Engineering (~1/3rd of the company?), whereas the source you cited is about all of Google. I wouldn't be surprised if the White/Asian/Male splits are far worse when you only look at software engineers.

* Google has many international offices (e.g. there's > 1000 employees in Google Japan).

> The critics are talking about Software Engineering (~1/3rd of the company?)

It's closer to 2/3.

Google hires many people who are not from the US to come work in the US, so I don't think the 61.5% is reflective of their hiring pool.

Oh, okay. So we can keep claiming Google, a US company, favors white workers until when? Until less than 10%* of their US workers are white? Is that what being favored means? That companies in your own country look elsewhere to import workers?

*Ballpark white global population.

All I'm saying is that your statistic does not disprove the allegation that Google's hiring processes might favor white workers. You're making an apples-to-oranges comparison, and drawing a conclusion that isn't demonstrated by the data.

I'm also not saying that these numbers prove that Google does favor white workers. Only that they are not sufficient to disprove the idea.

Should the null hypothesis be that Google discriminates in favor of white, Asian, and male applicants? Or should it be that Google is not discriminating improperly?

It seems strange to put people in a position where showing a datapoint that would otherwise tend to refute a hypothesis (which IMO should not be assumed to be true in the first place) is not enough to be considered as evidence against the hypothesis being true.

I'm not making any claim about what the null hypothesis should be here. I'm simply disagreeing that the datapoint says what the poster claims it says.

I'll say that you'd probably want to set your null hypothesis at "presumption of innocence" here.

The general trend I see in the community is that there's a strong belief that google, and by extension, the tech industry, is biased in favor of white workers.

The choice of hiring pool is part of the hiring process. Expanding that pool to other countries is itself an act of disfavor to US workers.

Google serves the world as their customers. It seems to me that having a preponderance of US-based employees that's wildly out of proportion to their customer (or revenue) base is still heavily in favor of US workers.

Many large companies serve the world (Xiaomi, Hikvision, Hitachi, Samsung, Mitsubishi..) while their employees in their home countries are overwhelmingly native. So if international customer bases are the norm, but only Google (or US companies) have significant international hiring, that effectively disfavors US workers.

What's "white" even mean? Descendants of north Europeans? Surface albedo of skin greater than 0.85 in the visible spectrum?

Race is self-reported. There isn't someone in Google HR researching your ancestry, they ask you what race you identify as when you join. If you self-identify as white then you are white according to Google HR.

Interesting, that introduces some very skewed incentives: from a game theory perspective it would always be better to lie about your race in HR surveys. Then according to the data, your actual race would be underrepresented and the industry would try to hire more people who look like you, making your next job search easier.

Descendants of Europeans. It's not ambiguous.

Just so you know, Middle Eastern folks from Lebanon and Egypt fall under the race of "White" in the US census. Source: https://www.latimes.com/projects/la-me-census-middle-east-no...

The US Census categories are incorrect.

What race categories are correct? Race is arbitrarily defined so any categorization is going to have flaws.

At least Census is consistent.

10 bucks says you could meet some of these folks on the street and you'd think they were European.

There is definitely a spectrum between Mediterranean and Middle Eastern.

You know nothing about demographics and race. Do some research before for you continue this conversation.

What degree of purity is required to count as a European? Europeans can and do mix with non-europeans.

Under US law, 'one drop of blood' is sufficient to count as nonwhite. Might be an outgrowth of puritanism, I'm not sure.

That was only true in some states during Jim Crow. Others had different definitions of "white". In fact it was possible for some "black" people to become "white" just by crossing a state boundary.

Racism is stupid.

"One drop" is not and was never federal law in the US[1].

[1]: https://en.wikipedia.org/wiki/One-drop_rule

The 99% of white Americans for whom all ancestors were from Europe, are European.

Just checked that there are 7 millions Jews in the US. That's a lot more than 1%. About 3 millions are Slavic people. Many millions are european-type Arabic people. They all look like white Americans and if they grew up in the US, they won't even have an accent. The number of "true north-european descendats" must be a lot lower, maybe 60% of white Americans. And this number is going downhill rapidly because Americans don't hesitate to mix with other races and nationalities. My point is that albedo of skin is a really poor predictor of ancestry, about as good as a coin toss.

Every human being descends from the original humans in Africa so I don't think the term "true north-European descendants" makes much sense.

> The 99% of white Americans for whom all ancestors were from Europe, are European.

“White Americans” average a little under 99% Eurooean ancestry, but far fewer than 99% have exclusively European ancestry.