Analytics is useful and important. It is utterly baloney, though, to use still pretty simplistic analytics at current stage to gauge a player for a team sport that requires players to move and have direct body contacts constantly, like basketball and soccer.

For example, defense. There are better analytics than several years ago when rebounds, blocks, and steals were the primary statistics. But basketball is a team sport. The contribution of a player in team defense includes how much space a player can cover, how much a player's presence can direct the offense's ball movement, how uncomfortable a player can make his opponent feel, both physically and psychologically, how good a player is at directing his/her teammates in a defense scheme, how good a player is at anticipating a potential issue and cover it, how much a player can make his/her teammates trust his/her defense so that the teammates can aggressively pressure the opponents, and so on. One has to really understand the game to evaluate a player's defense correctly and fully.

It is the same with offense. A simple example is that Duncan requires more help defense than Barkley. That creates more space for his teammates, thus more opportunities on higher percentage shots, which will translate to more wins. It also help gets teammates more involved, make teammates better, etc.

I really like Barkley, both as a player and as a person. But it is nonsense to compare him to Duncan in regard to the best PF in NBA history.