On today's episode of Hang Up and Listen (Slate's sports podcast which I highly recommend), Josh Levin made reference to a 2004 Slate article titled "Building a Better World Series" in which mathematician Jordan Ellenberg offers an alternative Word Series formula that, on average, produces the "better" winner more quickly. It is a smart piece and an interesting read.
The basics are that a team wins the series when it is up 3-0, 4-1, 4-2, 5-3, or 5-4 (which ever comes first). Of course, the format lacks the simplicity and elegance of the "Best of..." format but I do think he is on to something. Naturally, the format has one of the same shortcomings as the current format in that as the series gets longer (it effectively turns into a best of 9 series) the less sure we are that the format has picked the best winner--such is Ellenberg's tradeoff to produce a winner quickly. Of course, at 3-3 or 4-4 the differences between the teams are likely narrower so it is going to be more difficult to make a "true" determination of which team is better. The problem is that these narrow difference situations are exactly the ones in which we want to know who the better team is. My (untested and probably more imperfect) solution: after both teams get 2 losses, the winning team must win 2 games in a row to be declared a winner (like tennis). However, 3-0 will end the series and so will 3-1 (this makes the 2-1 game very exciting). The result would be that every other game beginning with the 3-2 game would be really exciting. Yes, the series could go on forever but such is my trade-off for producing the best winner.
In any case, it is fun to think about.