i don't buy the argument that requiring a member from each of the 32 teams degrades the level of the rosters. perhaps in practice, where some scrub reliever is selected to represent his schlubby team, but really, it doesn't have to. if, using fangraphs' WAR metric, you selected the top player at each position for each league, as well as the top starter and reliever, you would have 7 AL teams represented, and 8 NL teams. if you expanded your roster to include the top two players at each position (counting david ortiz as a 1b), you would have representatives from 12 AL teams (only seattle, and baltimore excluded) and 12 NL teams (colorado, san francisco, los angeles, and chicago excluded). add a 3rd starting pitcher, and you bring in seattle. and that only has 21 players per side. you bring in baltimore with the 3rd best catcher or centerfielder, or 6th best starting pitcher. you bring in colorado with the 3rd best centerfielder you bring in san francisco with the 6th best starting pitcher (grumble grumble matt cain grumble grumble) you bring in los angeles with either the 3rd best catcher, right fielder, centerfielder (tied with colorado), or the 8th best starting pitcher chicago, of course, gives you some trouble. their best player is only the 4th best shortstop. but even he is only 0.1 WAR behind hte #3 guy. and since, in reality, the all star rosters tend to average 35 men, if you set that as your number, you can have the top 3 players at each position, plus 9 pitchers (7 starters, 2 relievers), and you only have to bend the rules to fit in a cub who barely got edged.