Here’s the Miami Herald’s take on Charlie Crist:

Gov. Charlie Crist — more popular than President Barack Obama — would trounce former Florida House speaker Marco Rubio 54-23 percent if the 2010 Republican primary for the U.S. Senate were held today, according to a new Quinnipiac University poll.

Obama’s job approval rating stands at 58 percent, while 62 percent approve of the job Crist is doing.

“Rubio is going to have to convince an awful lot of Republicans who have a favorable impression of Crist that they need to re-evaluate their views,” said Quinnipiac University pollster Peter Brown. “The only practical way to accomplish that would be a well-funded, negative campaign.”

Maybe. But we should keep in mind two words: Creigh Deeds. The Washington Post, which endorsed Deeds, declares his victory in yesterday’s primary to be “stunning.” That’s because all the “experts” got it wrong. They often do because they repeat three basic errors.

First, they take horserace poll numbers months or years before an election too seriously. (That’s how Hillary Clinton and Rudy Giuliani were crowned “front runners” a year before the Iowa primary.) These polls — the Crist/Rubio one included — have limited predictive value. They are largely a factor of name recognition. Especially in non-presidential elections, voters tune in late and make their minds up in the home stretch. If you want a better indicator, look at favorable/unfavorable ratings. (For example New Jersey Governor’s Jon Corzine’s 36-56% approval number [which is worse than the previous month] is likely more meaningful than a 10% gap in the polls because it suggests he really has a way to go in turning around voters’ perceptions.)

Second, the political “experts” disregard the centrality of personality and raw political talent in campaigns. It’s strange that political gurus miss this since they spend their professional lives following races; yet they usually do. Virginia voters found Terry McAuliffe obnoxious, plain and simple. His double-digit lead in the polls faded as soon as voters realized who he was. Likewise, Deeds is simply a nice guy. And voters got that. In short, money and issue positions matter, but not as much as professional political observers think.

And finally, if you don’t know what the campaign narrative is about you aren’t well positioned to figure out who’s going to win. ( Recall Hillary Clinton running on “experience” in a  classic “change” election.) If the race in Florida and around the country in 2010 is going to be about what a bust the Obama stimulus plan has been, then all those photos of Crist cheerleading for the president will be the kiss of death. On the other hand, if by 2010 things are looking up and Crist claims part of the credit for Florida’s improved situation he looks like a “dealmaker” who’s going to get the best deal for the state in the era of Obama.

So if you want to know if Crist or Rubio, Arlen Specter or Joe Sestak, or McDonnnell or Deeds will come out on top, you won’t be able to find that out in June 2009. And if professional pundits don’t want to be “stunned” they won’t tell you that you can.