The bigger the bench the more that it helps owners who can predict how players will do and draft accordingly. The smaller the bench, the more it helps people who need to wait a few weeks and see how the player is doing before they realize they should have already been on their roster. An argument is sometimes made that roster management is more difficult with a smaller bench, which could favor the more skilled owner. But I tend to think it is absolutely dwarfed by not letting the more skilled owner draft and roster in the first place the guy that he knew was going to be worth it.
There's also a good argument to be made that larger benches facilitate interaction within the league. If a player is on someone's roster you need to trade to acquire him. If he's on waivers you don't need to interact at all.
I think furthering the role of skill, and increasing interaction between the owners, are two of the things any league should be striving for. So obviously I'm in favor of bigger benches.
As for my leagues... one league starts 26 players at 13 different positions and we have 29 bench spots for a total of 55. I set it up that way as I thought everyone should be able to backup each of the 13 positions with 1 player, and then have an additional number of spots that you can carry about another 1.5 backups per offensive starter. Though of course they can be used however teams wish. It's a dynasty/contract league so I wanted enough backup spots people could hold onto players who are still developing.
The other league we start 19 players at 9 different positions and we have 33 roster spots. Enough to back up each position with 1 player and have 5 other players to backup the 11 starters on offense. I'd like more, I suggested 35 when we changed to this system, but some owners wanted fewer so we compromised. This one is a keep 4 league, so doesn't need as many spots as we don't have rookies from a rookie draft we need to hold for 2-3 years before they become worthwhile.
