In Sowell's recent columns (for example, this one on negative advertising), he falls victim to the same phenomemon. His points about Obama's track record on various issues are only compelling under the assumption (not stated until the very end) that his record is a better predictor of his future actions than his currently stated beliefs.
When Sowell does get around to mentioning this assumption, he quickly dismisses any contrary opinion:
Does anyone in real life put more faith in what people say than in what they do? A few gullible people do -- and they often get deceived and defrauded big time.But this assumption is the basis of his entire argument. If you instead believe Obama's recent statements are better predictors of his future behavior, then nothing Sowell says in the article will be compelling. So perhaps it would be better to focus the article on addressing the evidence for this assumption rather than simply dismissing anyone who would disagree as "gullible".
Unfortunately for non-moderates, it is easy to find examples in either direction. In fact, republicans themselves do not always believe this assumption. For example, many seem to accept McCain at his word that he would not offer amnesty to illegal immigrants despite his clear track record of trying to do so.
The fact is that no simple rule -- either Sowell's assumption or the opposite -- is always true. The moderate voter is left to make the best prediction possible taking all the evidence into account on a case-by-case basis. Those who decide Obama means what he now says will ignore contrary evidence in his record, and conservative commentators like Sowell will talk right past them as they write article after article decrying that record.