I think the point is that it is getting stronger with an increasing number of players. In a 2-player game, it is mostly a 1-1 gain (Villager vs Coffers) and one opponent can better adapt to the situation by building a different deck. With more players, the ratios will increase on average, 1-2, 1-3 (per opponent-Magi player), and what I think might be crucial, it is getting increasingly difficult to go for different strategies (means key cards).
I don't get your point. I just showed that the average number of tokens that you get from Magi increases but the curve is concave, i.e. the differences themselves decrease.
No need to argue here. I agree.
To me it looks like a perfect, self-balancing design. It definitely has far less player number scaling issues than official cards like Pirate Ship or Jester.
Well, difficult to argue against my own card design. I actually like it very much the way I've presented it.
Different strategies? Either you need a splitter or you don't, either your Magi suffice or the Villagers you get via other Magi suffice. I don't see how this scales badly with player count. In multiplayer you are actually more likely to freeride on the Villagers, i.e. Magi might ironically be actually be weaker than in 2P in this respect.
This part is more difficult to answer properly. It is not only about Splitters. It could be an Attack card that the Magi player sets aside with the knowledge (or a high certainty) that the single opponent doesn't have it in hand or deck. In a similar way, it could be a Smithy allowing a big draw at the start of the next turn. Not a good idea if the opponent can make their Smithy non-terminal.
If it is a Village, and the opponents are more or less forced to play at least one Village, as their turn would otherwise suck, then with one opponent that one would get 1 Villager (assuming one Village was played) and the Magi player gets 1 Coffers (1-1 ratio). With 3 opponents, and all opponents play exactly 1 Village, each of them gets one Villager, but the Magi player gets 3 Coffers (1-3 ratio). I think that is what Gubump had in mind.