You may have noticed that the Romney campaign has gone through a couple of different core critiques of President Obama. First, they said he was a nice guy who was in over his head. Then they decided that they don't actually think he's a nice guy after all, but instead he's a crypto-communist who despises free enterprise and hates entrepreneurs. Now they may be reverting to the old message again. The Obama campaign looks much different. Very early on, they decided—presumably because their polling and focus groups told them this was the right approach—that they were not going to attack Mitt Romney as a flip-flopper, despite the fact that this attack has been effective against other politicians in the past, and Romney is without question the flippy-floppiest party nominee in American political history. Instead, they argue that Romney believes the things he says and only cares about helping the wealthy. While every once in a while you hear an insufficiently prepared Obama surrogate call Romney a flip-flopper, for the most part they stick to the plutocrat attack. That's message discipline, and all winning campaigns demonstrate it.
But David Karpf makes an interesting point about this. He argues that it isn't that message discipline wins campaigns, but that if you're winning, you can afford to have message discipline:
Any dimestore campaign hack can tell you, message discipline is one of the keys to victory. Winning campaigns offer the same consistent message, week after week, in making their case to the voters. Losing campaigns tend to waver. They try out a lot of different messages. None of those messages stick.
They're correct about the correlation, but wrong about the causal arrow.
Message discipline doesn't cause a winning campaign. Winning campaigns reinforce message discipline.
The Romney campaign now provides a perfect example: throughout the summer, the Romney strategy has been to paint President Obama as a good guy who was in over his head. You see this in the RNC’s latest ad "Break Up With Obama," where a former Obama supporter* tells a cutout of the President "It's not me, it's you." This strategy acknowledges Obama's high likability ratings and tries to work around them. It's a reasonable strategy, more promising than a Republican base turnout strategy would be. But in the aftermath of the conventions, polling suggests that it isn't working.
There's still a long time between now and election day. But if you're an Obama strategist, the current polling lets you comfortably stay "on message." If you're a Romney strategist, now is the time when you try something new.
I'd say I agree only partly. Dave is absolutely right that when you're winning you keep doing what you're doing, and when you're losing you often try something new. But it's also true that a campaign that has carefully designed the story it wants to tell won't get tempted into throwing a series of random darts at the board with every daily fluctuation in the tracking polls. You might remember that in 2008, the Obama campaign was under constant pressure from its supporters to change what it was doing, yet even when they were running behind John McCain, they adamantly stuck to their message because they were confident it would work in the end. And it did.
Nevertheless, Dave's warning is a useful one. In retrospect, we tend to think not only that everything the winning campaign did was brilliant and everything the losing campaign did was stupid, but also that we can then generalize those results to a set of instructions about how to win a campaign. And now we get to do it in real time. But it's often wrong, or at least seriously overstated.