I can't speak for the Bay Area, but my region exhibits similar patterns.
Consider an example. A "good" school might have 100 slots for out-of-area kindergartners, distributed via a lottery system. However, the lottery system uses a biased distribution which preferentially selects from students who have been on the waitlist longer. This system is easily defensible by school administrators in the name of "fairness" to students who have waited longer.
The unspoken desire, however, is to preferentially select students with a higher socioeconomic index, because such students tend to score higher on standardized tests (and therefore maintain the school's current "good" ranking). Asking for this information explicitly is, naturally, forbidden.
Instead, the school waits for parents to come for site visits during working hours. These visits are attended overwhelmingly by stay-at-home parents, which correlate with socioeconomically-advantaged families, for the simple reason that they can afford to have a parent stay at home (the -economic side of socioeconomic). Underemployed parents cannot attend, because they are still working, but usually at jobs which do not offer flexible work hours. Unemployed parents can attend, but rarely do so, because the chronically unemployed statistically come from lower social backgrounds (the socio- side) that either do not value education as highly, or may not know that the concept of "school visits" even exist.
Then, when these parents visit, the schools inform them of the existence of the waiting list. However, information about the wait list (and especially details about its biased selectivity) will not be available by phone or website until enrollment opens. Thus, parents who attend the tours will be preferentially selected, since they are on the waiting list first, and the school can put their finger on the socioeconomic scale without ever asking parents for sensitive information.
For purely public schools which do not allow out-of-area students, this approach isn't relevant, because they are required to accept students from the local area and only the local area. These schools don't care, because the locations themselves correlate with socioeconomic factors, due to property values tracking school achievement.
For charter schools, which pass all students through similar systems, it allows them to accept statistically-advantageous students across all levels, leading to better outcomes overall (and shifting the remaining disadvantaged students, and the correspondingly lower level of achievement they statistically will achieve, to low-income public schools).