There are six officially recognised racial categories: White, African American, American Indian/Alaskan Native, Asian American, Native Hawaiian, and Other Pacific Islander.
There are two official ethnicities: Hispanic and Latino Americans, and Not Hispanic and Latino. (Note: Hispanic is actually a cultural group, not an ethnicity, and in the real world the terms hispanic and latino are orthogonal. Try not to think about it too hard)
Most hispanics and latinos are officially categorised as white. That means that whenever you see the term "white" you'll need to do some further digging to see if that means the official "white term (including most hispanics), or if it's a short hand for "white, non-hispanic and latino". The 77% you cite as "white" includes latinos; the number you probably are looking for is 61%. The 56% in google's numbers seems to exclude latino's, so the comparison is probably 56% to 61%. (Probably!)
Also note that if your goal is equity for all groups, it doesn't make much sense to average minorities together. The Hispanic and latino group (including both whites and non-whites) makes up 18% of the US population, but the apparent matching group at Google makes up 4% of their employee base. Conversely, Asians make up about 5% of the US population, but apparently around 35% of Google's employee base using the closest equivalent measure. Clearly Google shouldn't get a pass for discriminating against hispanics (if that is indeed what they're doing) purely because they discriminate in favour of Asians (if that is indeed what they're doing).
> Doesn't that mean that racial minorities are actually overrepresented at Google?
Some (well, one) are. Others are not. Non-hispanic white's are broadly in line with the major population share.
With regard to Asian overrepresentation and Latino/Hispanic underrepresentation, I think there's a really good chance that how those respective immigrant groups came to be in the United States probably has a large role to play in this disparity. East and South Asian immigrants have to cross an ocean to get here, whereas most Latino/Hispanic immigrants could cross a land border (on foot in many cases!). The disparity in wealth required to pursue these alternate immigration paths probably significantly skews the class mixture of these groups and, thus, their representation in white collar professions.
These groups are generally forgotten about or elided when talking about how Google or US tech has a diversity problem - there is just a diversity problem with blacks, latinos, and women, and my hunch is that it has nothing to do with tech being racist or sexist and has everything to do with either larger structural issues in the US, or people choosing through free will to not enter the field.
As a hunch, yes I used to be of that opinion as well. I thought it was nonsense. Sorry that I don't have the data handy to back it up, but seriously (not being snide) if you care enough to matter you can research it.
In particular, though, the only over-represented group is asians: 35% vs 5%.
Also keep in mind that in official US usage that term basically means "someone from a Spanish cultural background OR a Latin American country". Although some agencies specifically exclude countries from Latin America without a Spanish or Portuguese culture.
So a white guy whose grandparents immigrated from Spain counts (the grandparents spoke Spanish), and a Quechua from Bolivia who recently immigrated and only speaks Runasimi counts (Bolivia has a Spanish culture), and a black guy from Guyana might count. The OMB says no (Guyana was colonised by the British), but the Census says yes (it's in South America, so close enough).
The system makes very little sense.
You either have to go one step further and break that down by available tech talent to be recruited, with the premise being that you can't magically manifest a certain % of white/hispanic/black/asian tech employees if they're not available locally (eg if the population is 12% hispanic, but only 5% of the tech employees available locally are hispanic; that presents a very difficult recruiting scenario).
Or the argument is that a company like Google can afford to recruit nationally and relocate whatever demographics they see fit to the bay area, such that local restrictions do not apply as much to them. If the local bay area pool of tech talent is 3% black, and the national pool is 6% black, Google can afford to relocate black employees from around the US to bolster their diversity efforts locally if that's their aim.
Though Asians are only 5% of US, they are maybe 1/3 of Tech degrees, which is consistent with Google hiring.
Similarly for women.
Very roughly, Google seems to be selecting in a fairly unbiased fashion from the candidate pool.
Where the issue gets more sticky is at upper levels of management, which are fairly male dominated.