I do see the speed advantage of group init when init is first rolled, and I see the speed advantage because one side has a third of its force on the ground in round one because the other side does focus fire, but we've never had issues with individual initiatives (since the players have a prop that handles this for all of the PCs and NPCs, and the DM has that task handled for him). Granted, individual init does have that up front prep time. But the concept that individual init is slow during combat can be countered. For one, don't let players kibitz during encounters. It's Frank's turn, Frank talks and decides his PC's action. Adding in this simple rule means that combat itself is just as fast as with group initiative.
I could even argue that group initiative is slower during combat because there is so much kibitzing going on that it slows down combat. That may be overcome by superior group tactics, but then again, it might not.
But the real advantage of individual initiative is that the DM typically cannot just wipe out a third to half of the party on round one (shy of Dragon breath and failed saves) in a tough encounter (and vice versa, the party doesn't wipe out a third to half of the enemies either). This to me, trumps the up front speed advantage of group initiative. And combining surprise with group initiative could be an extremely deadly situation for one side.
I could even argue that group initiative is slower during combat because there is so much kibitzing going on that it slows down combat. That may be overcome by superior group tactics, but then again, it might not.
But the real advantage of individual initiative is that the DM typically cannot just wipe out a third to half of the party on round one (shy of Dragon breath and failed saves) in a tough encounter (and vice versa, the party doesn't wipe out a third to half of the enemies either). This to me, trumps the up front speed advantage of group initiative. And combining surprise with group initiative could be an extremely deadly situation for one side.