Allright, i give it to you that 3 as a minimum is nonsense, as getting four is the same in points with a strategic advantage.
Hm, i didn't take into account other players might go for Duke, too. Let's see, if i buy 4 Duchies, then alternate, and my opponent buys Duchies until they run out, he gets 7 Duchies at a point where i have 5 and two Dukes (assuming we both get equally many chances to buy them). If we both get no or one more Duke before the game ends, i win. if we both get two more, we are tied. So i guess the question is, how likely is it that the game ends before the "get all" strategy becomes better. For 3 players, i think it's reasonable to assume that it will indeed pay off, at least in a mirror situation. For 4 players, that's either 3 guys sharing Dukes (making them just better Duchies) or 2 guys buying Provinces (6 each). The first case makes "buy 4, then alternate" mean pretty much the same as "buy until they are out", while the latter makes it possibly better. And of course, if i'm the only player going for Dukes, the game might end before i get even one, at which point switching earlier would clearly have been the better decision.
As i said, 3 probably really was a stupid statement. For 4, i think it's reasonable to say that it depends on the player count and strategies whether "gotta catch em all" is actually better. I guess i just want to say that it's not a trivial decision.
For the counting VP cards, you usually want to stock up on the thing they count before you start getting them (except SR and Gardens in rushes). With a Feodum enabler, you really want to concentrate on winning the Silver split more than the Feodum split.
Sure, but Vineyard, Feodum, Gardens, even Fairgrounds count cards that are useful for other purposes then pushing your VP count. Acting as if three Silvers did the same to your deck as a Duchy isn't really fair.