Tuesday's defeat of a sitting judge in California by a seemingly far less qualified candidate is a rarity. But the underlying mechanism is not:
Political consultants have long complained that judicial races -- particularly in a county as large as Los Angeles -- are somewhat arbitrary. Few voters have heard of any of the candidates, and yet most vote anyway. The voters' choice often comes down to the scant information in front of them in the voting booth: the candidates' names and job descriptions.
A candidate's ethnicity might thus influence things. A candidate's job title might, and often does, in favor of incumbents. Sometimes the candidate's having bought space in a slate mailer might influence things, though in sufficiently low-profile races even that's likely to have little effect. (A slate mailer is a political mass mailing that promotes a slate of candidates from the Governor on down, a slate that's often united chiefly by the candidates' willingness to pay money to be included in the slate mailing.)
Direct democracy is likely to yield good results only to the extent that the voters are actually likely to have meaningful information about the candidates. And voters, being rational people, aren't going to spend much time -- or often any time -- researching every candidate for every judgeship, community college district seat, and the like.
At least with ballot measures there are meaningful arguments in the ballot pamphlet, and the voters can use imprecise but tolerably helpful proxies to help them make the decision. With statewide ballot measures, there's also a decent amount of advertising and relatively memorable editorializing. Not so with most of the down-ticket races. Perhaps in a few cases there might be scandals that cause some trial judge to be duly removed from office by the voters. But that happens very rarely; most of the time the voters' choices are based on virtually no information.
One possible solution is to have races be partisan. At least then one can use party affiliation as a decent (though admittedly imperfect) proxy for a candidate's likely beliefs, and count on the party to do some direct or indirect screening of its representatives (at least at the general election). I'm pretty sure that this is much better than nonpartisan voting for local executive and legislative races, though in some heavily one-party cities it would make more sense to try to develop local parties beyond just the Democrats and the Republicans.
But if we think that partisan races for judicial office are bad -- I'm not sure about that, and I know some states have partisan judicial races, but for now let's assume they are -- then it's much better to have a purely appointed judiciary than an elected one. And if you do want elections for judges, at least limit them to the very few positions where the voters are likely to pay attention, such as state supreme court positions.
Incidentally, I'm told that the L.A. bar association puts out ratings of judicial candidates. Unfortunately, I'd only heard about this recently, and forgot to look into it in time. My error (which led me not to vote on any of the judicial candidates) -- but if even I made this error, and didn't have the association's endorsements in my voting booth, how many nonlawyers voters do you think had those endorsements in their voting booths?
Incidentally, when we all start voting more from our home computers, the situation might change, mostly (I think) for the better but possibly for the worse -- with home computer voting, we might at least be able to easily integrate various groups' recommendations into our own votes. I discuss this possibility in this short article. But if such a system isn't implemented (and in any event until that system is implemented), I think that voting for many down-ticket races, especially nonpartisan down-ticket races, is too much democracy.