The New York Times has a couple interesting pieces on this: a news analysis and a discussion panel. One person from the discussion panel, Steven Greenhut, does a good job explaining this, even though he's against this "gimmick":
It’s true that Democratic primary voters tilt left and Republican primary voters tilt right and that leads to general-election choices that strike some people as too extreme. But there’s something disturbing about instituting an election reform to elect a specific type of candidate, in this case more “moderates.”Though some supporters did talk up that this would bring in more moderates — as opposed to extreme right or extreme left candidates — I don't agree that that was necessarily the goal. Rather, it is designed to elect officials who are closer to the average voter in that district than we currently get.
To read more, click here.
Let's say the political spectrum in California runs from 0 for extreme liberal and 100 for extreme conservative, however you wish to define it. And let's say one-third of California voters are Dems and one-third are Republicans, while one-third are moderates or unaffiliateds who don't see eye-to-eye enough with either major party to join their ranks, and so they are then effectively locked out of the primary process and have no say in who the candidates are in November.
If there were only Democrats and only Republicans, we might have an "average value" of 25 for Democrats and 75 for Republicans, with the distance between the two typical candidates in an election being 50 spectrum points. The population as a whole has an average spectrum value of 50, so whichever candidate is elected will typically be an average of 25 points away from the average voter.
But since the middle* is occupied by moderates who are often non-affiliated, the actual spectrum value of the average Dem (and therefore the average Dem candidate) is lower than 25, maybe closer to 17 (the halfway mark between 0 and 33). Meanwhile the same is true of the average Republican and the average Republican candidate on the other side, about 83 (the halfway mark between 67 and 100). That means the distance between the average California voter and his/her eventual representative or executive would be 33 points. Furthermore, the distance between the Democratic establishment in Sacramento would be 66 spectrum points instead of 50.
The result, proponents of Proposition 14 say, is Sacramento's unceasing gridlock.
With the party primary system, people averaging 40, 50, or 60 are forced to choose between a 17 and an 83. The 35 or the 65 are more likely to choose the 17 or the 83, but the large group in the middle hovering between 40 to 60 may be at a loss of whom to vote for, since there is a large enough difference between them and a 17 and a 50 or even a 40 that they may be dissatisfied.
And that brings us to the issue of voter disgruntlement. Nobody likes to see their candidate lose, but with Democratic and Republican candidates hovering around 17 and 83, respectively, instead of 25 and 75, there is even more distance between the "lost-out voter" and the winning candidate.
And this is not the only form of disgruntlement. The candidates who are closer to the average spectrum value of the electorate, those from 40 to 60, are so far removed from that of the electorate in either major party's average spectrum value (17 and 83) that they have difficulty getting elected except when negative factors force a candidate closer to 17 or 83 lose.
In a top-two primary, however, the 17 and 83 will not likely garner enough votes to be top two. Instead, it will be candidates in the middle (say, a 40, 50, or 60) who can garner votes not just from the Democratic or Republican core (17's voting for a 40 and 83's voting for a 60), but also from the unaffiliated middle, where the distance between 40 or 60 and the average moderate value of 50 is very small.
This means that ideological difference will matter less because the average voter will already be choosing candidates close to them, so innovations in policy ideas will become more prominent reasons for candidate selection.
So this is where I disagree with Mr Greenhut. While the system appears to favor moderates right now, that is only because moderates are the norm in California. Were there to be a shift left or a shift right, the top-two primary system would necessarily lead to such results.
In short, the system is designed to elect candidates whose views are closer to those of the average voter.
An added benefit is that more voters will own the candidate in the sense that more will have chosen them in June to run in November. A similar problem exists in South Korea, with only a plurality of voters owning the presidential decision by having chosen him/her. The head-of-state is hobbled from Inauguration Day by facing an electorate that mostly didn't want him/her as their first choice.
South Korea needs to adopt a two-part election system for president, governorships, mayorships, and other important posts, a so-called runoff. If no one gets at least 50% plus one on the first ballot, the top two (or three) campaign against each other in another election about three or four weeks later (and if it's three candidates, then if no one gets a 50% majority in that one, then a third election is held).
More on this idea later, which I would like to push to get adopted for the 2012 election. None of the democratically elected ROK presidents under whom I've lived, from Roh Taewoo on up to Lee, has received a true majority of the vote.
Yeah, this might sound expensive, but if it's important enough to vote on, then it's important enough to vote on right.
*Actually, since voters would probably appear in the political spectrum not as an even distribution throughout but in some sort of normal curve, the one-third who are moderates would more likely be squished together in the middle, meaning the spectrum values for Dems or Republicans would be higher in both the Dem-Republican-only scenario and the Dem-Republican-moderate scenario, but in a similar way, so it's not necessary to add this complicated layer to what is intended as a simple discussion.