The electability argument doesn't say what you think it says, Paul Krugman

I have a strong distrust of electability arguments, because they are usually made by people who are not making an observation, but beating the drums for a cause. In fact, it is a mathematical truth, in the modern American system, that one of the candidates from either the Republican or Democratic party will win the election. It is another truth that the GOP or Dem candidate will have won the majority of the primaries. Is it, however, true that the person who wins the primaries in a party isn't always the most electable in the general? What that means, what that should mean, is not that the candidate who lost the primaries could lose less the general, but that the candidate who lost the primaries could have won the general. Personally, I think this is totally unlikely. The argument of electability is usually manipulated by Democratic centrists, and they usually pick McGovern for their punching bag. The problem is evident, however. For if Mcgovern was a uniquely bad choice, what they are contending is that his opponents - basically, Humphrey and Muskie - were better. But when you go back to the Gallup polls, there's absolutely no evidence for this. Humphrey and Muskie both did worse against Nixon in the polls in May, 1972, than McGovern. Intuitively, it seems more probable that the winner of the primary is probably the most electable candidate that the party has running. This intuition is borne out, partly, by the fact that it is rare (in fact, I can't think of an instance) that a person who lost his party's nomination in one election cycle to a person who lost the general was elected in the next cycle by the party and defeated the incumbent in the general. To give an example that is less muddy: say, Kerry had lost to Gore in 2000, who lost to Bush, and then Kerry won the Dem nomination in 2004 and defeated Bush. In fact, most of the time, those who lose in the primaries never get a chance to be nominated - unless they are VPs. Humphrey lost in the 1960 primary, and did get a chance in the 1968 election. Even so, he lost.
Now, given this, I think it is important to note that the electability argument has been used to promote a buncha egregious losers to the Democratic coronation: Humphrey, Mondale, Dukakis, and Kerry. All we were told were electable - in fact, before any substance, this is what they were supposed to be. But they weren't. They were losers.
That said, I do think Clinton is electable, though I am for Sanders in the primary. But she is a bit like Humphrey, which is a bit frightening to me.