Is there actually a reliable way to measure gender pay gap
If everyone is making widgets doing a gender pay gap analysis is easy. It works well for bus drivers too.
With artists or creative knowledge workers where everyone is doing different things for different money it's a fraught statistical process. Hard, or maybe impossible. Whatever the result, someone will claim it is wrong, maybe even with some justification. More likely, with varying combinations of ideological moral certainty and cluelessness.
Google would crazy to institute this kind of lose-lose project. It would require arcane statistical methods that just about no one would trust. If an analysis showed they were gender neutral, they'd be accused of cooking the books before anyone had read the report. If it turned out they weren't gender fair, that result would be distrusted by others. Then, what are they supposed to do: inflate wages artificially or promote people with lower perceived competence, or what? There's a reasonable to good market for this type of employment. We aren't talking about a sweatshop in rural Pakistan.
Personally, I'd want to see the reliable methodology appropriate to Google's workplace before we bang our dainty fists on tables and demand that they produce a bunch of meaningless numbers. "Everybody knows" doesn't cut it with me.