Thanks to a ballot proposition (California has far too many of those), California has a top-two nominating system. In the "primary," all candidates of all parties run against each other in one giant clusterfuck of a primary. "Clusterfuck," by the way, is a technical, political science term. You can use it on your exams! I'm a professor, and I say so. Any other professors who complain when they see you use the term, "clusterfuck," on exams may direct their complaints to Matthew Jarvis, California State University, Fullerton.
Anyway, in contrast to Thunderdome, two candidates leave the clusterfuck. The top two vote-getters in the everybody-against-everybody primary face each other, even if they are of the same party. (The Louisiana twist is that if someone gets over 50%, that person wins outright, and they cancel the general election, because everything is crazier in Louisiana).
Who would devise such a blinkered system?
The answer, of course: goo-goos. That's right, my perpetual adversaries. Goo-goos. Short for, "good-government" advocates. What's that that ails ya'? Well, I've got just the solution! My snake oil is guaranteed to fix everything wrong with politics! And it's sooooo simple!
What's wrong? Polarization. The problem? Primaries. Candidates are just moving to the extremes to appeal to primary voters. But, if everybody runs against everybody else in the California Clusterfuck (TM), then the Democrats can also win by pulling in Republican voters if they move to the center, and Republicans can also win by pulling in Democratic voters if they move to the center. The theory is to defang the extremist pressure of primaries. How sound is the theory?
Not very, for a lot of reasons. First, primaries aren't really the reason for extremism. Second, cross-over voting when you permit it by primary rules only really happens if you have a contest in which the race is just completely uninteresting on one side. In Congress, that means an incumbent, who wins the general election anyway, in which case the cross-over voting doesn't really do anything. That's the quick version, but really, the theory here is pretty weak.
And there's a big hole, scaring the crap out of a lot of Democrats right now. What happens if one party has more candidates dividing up the votes than the other? They get locked out of the general election. If, say, the GOP winnows its field down to two candidates, and the Democrats winnow their field down to three, with each party dividing their votes equally, the GOP has the top two vote getters unless the district tilts sufficiently Democratic to compensate.
This turns everything into what we call a coordination problem. Consider the price-fixing game in markets. Two firms in an oligopoly. They each consider raising their prices, and if they both raise prices by the same amount, a few consumers get pushed out of the market, but each firm does better by raising their profit margins. The problem is that if only one firm raises their prices, that firm gets screwed because nobody shops there. It's a prisoner's dilemma. The structure of the game is precisely the same. Two firms in an oligopoly can try to coordinate, though, and over time, you can get patterns of behavior in which they reward or punish in order to achieve coordination. The more firms you add to the market, though, the more difficult coordination becomes. That's part of why we like markets with lots of firms. They can't solve that coordination problem. A market with enough firms can't price-fix because they can't coordinate on a fixed price. That's part of why markets with lots of sellers are good. That's the coordination problem working for you, the consumer.
Wanna see it working against you? Let's say somewhere in the neighborhood of 50,000, plus or minus, will be voting in a California Clusterfuck primary. Your party will be splitting somewhere in the neighborhood of 30,000 between three candidates because three people couldn't coordinate to get two to drop out, and the other party's 20,000 or so (fewer than your party) will be coordinating around two candidates because they did a better job and got one more narcissist to drop out than your party did. Your side has more voters, but you might get locked out of the general. Why? Each candidate is going to get roughly 10,000 votes. Random chance will determine which two candidates get to the general. That might mean the disfavored party automatically wins the general!
Unless you can coordinate with the other 29,999. This is the coordination problem working against you. Your inability to solve that coordination problem-- the same principle that benefits consumers in a competitive market-- could lock your party out of a seat that it should win, thanks to the silliness of the California Clusterfuck primary system. Ah, goo-goos. Is there anything they can't fuck up?
Of course, the other way to solve the coordination problem is for parties-- we political scientists tend to like these things, partiers that we are in the most boring sense-- to get more narcissists to drop out before the primary. That's hard. Politicians are narcissists. Go away, politicians. Go away. Shoo.
You know what solved this coordination problem, though?
Regular, old fuckin' primaries! Of course, there are other approaches. In my opinion at this point, the problem with primaries is not that they select extremists, but that they select dipshits. See, for example, the recent research by Sarah Truel and Rachel Porter confirming what we really already knew, but by building the most awesome data set ever. Basically, Republican primary voters select against qualifications. That's... bad.
These days, I don't like primaries. The solution, though, is not the California Clusterfuck system, which is mathematically indefensible. The Westminster System? I dunno. Just throwin' that out there. Coordination problems are real, though, and creating them knowingly? No. Just... don't do that.