The poor showing of the Oakland A's has caused people to question whether the "Moneyball" idea is valid. The discussion has even penetrated the lawblog Volokh Conspiracy, where several commenters noted the limitations of the system, even at its best. There were several misunderstood statistical analyses which A's management could capitalize on by picking up undervalued players. They knew what others didn't. What happened since then is that some of the better teams have moved toward those strategies - notably OBP and WHIP - but been able to bid more for players as well. The A's were using 2nd-level money to get 8th-level results, getting them to the playoffs. But they were beaten by teams like the Yankees, spending 10th-level money for 9th level results, and the Red Sox, spending 9th-level money for 10th level results.
But the edge is vanishing. It turns out that OBP, while predicting very well which young players have command of the strike zone and will eventually hit for average and power - think Kevin Youkilis, who was known as the Greek God of Walks in the minor leagues - it is far less useful in predicting what players will do late in their careers - the names JD Drew and David Ortiz might come to mind. (Kudos to commenter Michael for alerting me to this.)
But Moneyball is not ultimately about On Base Percentage, but about finding players undervalued for any reason. In the 1940's and 50's, a person knowledgeable about the Negro Leagues could have assembled a great team of undervalued players. From 1960-2000, knowledge of OPS, outfield assists, and the break-even point for stolen bases would have been a great advantage. What is the next overlooked advantage? The NBA has learned the value of expiring contracts - what's the baseball equivalent?