Phil Singer mostly nails this:
Political attacks work best when the charge they make is both echoed by the subject of those attacks and resonate with voter perceptions of that candidate. Case in point: The flip-flop attack on John Kerry wouldn’t have been nearly as effective as it was if he hadn’t told voters in West Virginia that he voted for the $87 billion before he voted against it. Kerry gave the Republicans a real time example of the negative storyline they were driving against him.
Fast forward to 2008: It’s tough to make the McSame attack stick because John McCain rose to national prominence by being a thorn in George W. Bush’s side. McCain might have voted for 90 percent of the Bush agenda but the public got to know him as a pain in Bush’s behind - a perception aided by the fact that Democrats rushed to exploit the McCain-Bush schism that came out of the 2000 primaries.
So does that mean the Obama campaign should ignore the fact that McCain voted 90% of the time with Bush. Absolutely not.
It means that the Obama campaign needs to focus its energies on generating some real time examples of McCain hugging Bush. (I think there are some other areas to hit as well but that’s a post for another time.)
Democrats need to test McCain’s maverick claims by creating news stories that force the Republican to choose between opposing the Bush Administration OR adopting the Bush position on an issue playing out in the headlines.
Therein lies the rub of the McSame strategy and the challenges Obama has attacking. The McSame attack directly contradicts McCain's image as Bush's #1 rival in the Republican Party, something Democrats were all too happy to stoke back in the day.
This is the Obama campaign going with an Attack 1.0 strategy -- pick your opponent's theoretically most damaging vulnerability and hammer away at it, regardless of how initially believable it is. The premise: repetition will make an initially farfetched but damaging attack believable.
The McCain campaign and the Steve Schmidt machine is pursuing an Attack 2.0 strategy. Pick the most believable attack (or the one most likely to get picked up by earned media, which magnifies paid media by orders of magnitude) even if it isn't the most damaging, and hammer away until it is the most relevant and therefore damaging.
Attack 2.0 beats Attack 1.0 because there is some kernel of public belief in the attack that allows it to go viral.