"If I beat you in a 12 game match it probably does mean I am better than you ."
By this logic, anyone Carlsen has a minus score against, he is worse than. The problem is, if that player he is "worse than" has a minus score against some random 2600, even if he has a plus score against 95% of them, is that player now worse than that 2600? And maybe that 2600 can beat almost all 2400s but just has one 2400 that has a style that conflicts him and thus the 2600 has a minus score against that particular 2400 -- now the 2600 is worse than that 2400. The transitive property would seemingly tell us that Carlsen is worse than a 2400 player -- if we kept going he might end up weaker than a class player.
Exaggerations aside, the point is it's entirely plausible that even the best players have a counter (as does everyone! Even the "GM killers" (e.g., a strong amateur who beats lots of titled players in internet blitz) probably struggle against some weaker player ) -- surely there are some players Kasparov did worse against than others -- but overall his style/play proved to be the most consistently good one in his time -- tested by playing many different types of players.
I view match play results as another piece of data -- a valid one, but not a totally definitive one, unless we can somehow confirm that it is done in a purely scientific manner.
"Estragon, I don't know if you remember the match between Anand and Topalov. The games were absolutely fantastic, but apart from an opening blunder in game one and an incomprehenisble mistake in a drawn opposite colored Bishop endgame"
To be fair, a top player shouldn't need to rely on people using the phrase "apart from" a whole lot. Besides, you could use the apart from idea to justify Topalov too. Topalov did great, apart from the games he lost. Ultimately, neither guy played perfectly -- the deciding factor was of course the final score, which was narrowly in Anand's favor.