I expect that you could have several biases:
- Preference for new technologies or languages will favor people who were recently trained or recently spent a period training, which will tend to be people younger in their careers. Can test for this bias versus ageism by comparing pairs of candidates with similar learned technologies, and seeing just their outcomes. (May have to control for confounding factors, though.)
- Preference for candidates who are willing to accept more intrusive or abusive working conditions. This will tend to skew towards people with less work experience and fewer other commitments in life, which again would bias towards younger candidates. Again, you could test for this versus ageism, but the pairings of matched candidates becomes harder because it depends on psychological profiles and life circumstances.
- Google has a bona fide business reason to prefer recent graduates (ie, even if this is agist, it might still be legal), because there's a higher variability in their career engineering RoI (lifetime value generated over lifetime cost to employ) versus established engineers, and Google's main mechanism of value creation is people farming -- getting high returns on engineers, which requires the underlying variability of recent graduates to generate high returns. It's similar to CDOs on mortgages: you can generate a higher return by increasing the variability of the underlying asset and decreasing the risk of systemic failure. Google just does that with engineering talent. This is testable versus ageism by pairing people based on recent graduation, or a score of outcome variability, and see if Google is preferring highly variable people.
I would expect that the reality is a combination of these, often enacted by different departments: HR prefers the abusable hires; technical interviewers prefer the new technologies and languages; both prefer the variability, as part of a broader corporate strategy, and because it provides benefits to each department individually.
However, with several legitimate business decisions each giving a slight bias towards younger candidates, you can get a situation with a heavy hiring bias against older candidates without any agist policy in place (or any agist bias on the part of hiring manager or interviewers), simply because lots of other developmental processes correlate highly with the process of aging.