After the Potomac primaries on Feb. 12, I wrote:

And if Obama wins his 10th straight primary after Wisconsin, it is extremely difficult to see why Ohio and Texas will essentially decide to have a brokered convention.

Well, they did.

Not only was my prediction yesterday wrong, that Obama would win TX outright (though he still may win more delegates there.)

But my entire read of the Democratic electorate has been wrong.

Fears of a brokered convention did not dominate. In all likelihood, most primary voters were not even aware of the scenario, as it was not an element in most media coverage.

I also thought the flurry of attacks at the end amounted to a blur that would not favor either candidate. The Wisconsin run-up struck me as far more clear anti-Obama storyline -- and it ended up backfiring on Clinton for going so lamely negative.

But the "kitchen sink" approach finally worked. Clinton did better with voters who made up their mind in the last three days.

What exactly did it?

Red phone ad?

Fresh round of Rezko stories, thanks to his trial starting up?

Canada-NAFTA flap, with a possible assist from the right-wing Canadian government? (Clinton ran a late radio ad on it disguised as a news report from the Associated Press.)

The Muslim smears, which can't necessarily be traced to the Clinton campaign, but seemed to make a impact for the first time, both anecdotally and in the exit poll data.

Who knows, it's impossible to parse it all out.

This time, for the first time since Nevada two months, was Clinton able to pull off the "raise doubts about the new guy" tactic, without it backfiring on herself.

That is particularly striking, because the Clinton has done so many things wrong in the past weeks.

Inconsistent messages. Jarring shifts in tone from the candidate. Turf battles among staff played out in the media. Generally poor media coverage, both undeserved and deserved.

But Clinton can make more mistakes without bleeding support, because the Clinton brand is a more proven brand. Obama has to be near-perfect to overtake the more established figure.

Which for the most part in this campaign, he has been. His red phone rapid response ad was genius.

But he didn't respond to the Canada flap with all the facts in hand, and that helped drag the story out. And he had difficulty putting to rest the Rezko non-story (though once the trial is over, it would presumably be done with).

For the upcoming 12 contests, if Clinton keeps doing well in the Midwest, and Obama does well in the West and states with significant African-American populations, they'll likely split what's left.

Leaving Obama with the delegate edge he has today, but not enough to avoid a brokered convention.

There's no way Obama can be chased out of the race when he has more delegates.

But after tonight, it would take more than a split to chase Clinton out, if that.