Bah, OK, loophole with gainers. Let's try a slightly more complicated version that is closer to what I wanted:
Switch - Action - $2*
-----------------------------------
+2 Cards
Each other player gains a Curse card.
-----------------------------------
During your Buy phase, this costs $5.
When you would gain this other than during your Buy phase, gain a Curse instead.
Now what? Still a bit crazy with HoP, but I could finagle it even more to get rid of that if necessary. The point is, this will only ever be bought for $5. It technically costs $2 because that's what it says on the card. It essentially costs $5 because that's what you'll pay.
I agree that it functionally costs $5. Just liked Peddler is functionally unbuyable, decent, or lucrative depending on how many actions you have played. Grand Market "functionally" costs virtual coin, Silver, and Gold.
If there were a ton of cards like Switch, my guess is cost would end up being a faulty way of basing these rankings.
As for your thought exercise, LGG is almost identical to IGG, so it's just slightly worse on the $5s list. On the $2s list... you manage to underscore the reason we organize things by cost
Because it's really hard to rank it there. I mean, it's just IGG. Let's design a new list, "The Best Cards That Either Cost $2 or Are IGG." Is IGG number 1? Well, it's not better than Chapel, right we all say that. Maybe not Courtyard either? I don't know, it's like top 3 or 4 probably definitely. But it ends up being kind of hard to compare the relative goodness of IGG vs. the relative goodness of, say Squire, because you also have to say, well IGG is so much harder to buy than Squire, because of cost. Is Goons better than Masquerade? Is Ambassador better than Cultist? Is Margrave better than Fishing Village? Are Nobles better than Silk Roads? There are objective answers to these questions. One of Goons/Masquerade IS better than the other. However, I think it's hard to tell, and more importantly, I think in a community-based ranking the community would be more likely to commit errors (possibly just always ranking the stronger cards better despite their steeper costs).
So we remove cost as a consideration, by only comparing like-costed things. In doing so, we can make intelligent considerations on the margins, for cards like Peddler and Duchess and such. And I think it's perfectly reasonable to do the same for Masterpiece. It's also technically correct. So I stand by it.