Another interesting thought experiment: Imagine two cards, one of which is Swindler and the other is Mountebank. One costs $3 and can potentially turn $5 Actions into Duchies, the other one costs $5 and doesn't really do anything if the opponent has a Curse in their hand, and while they both are very powerful cards, the Mountebank one is still a lot more powerful relative to its price.
I think it's kind of interesting to note the effect of your opponent's deck composition for Swindler and Mountebank. Swindler arguably gets better as your opponent's deck gets better - like you said, you can turn awesome $5 actions into Duchies, or even critical $4's into less-useful ones, but you can only turn junky $0 cards into other junky $0 cards.
Mountebank, on the other hand, gets worse as your opponent's deck gets worse. If she's drawing a Curse in every hand, a single Mountebank is effectively not much more than just a +$2. It has a negative feedback effect, since the more you play it successfully (while Curses are still around, at least), the less likely it is you'll play it successfully again.
I guess the upshot is that Mountebank will help you to get ahead, but then slow down in its usefulness. Swindler starts out OK, but can really help you catch up from behind if you Swindle some key cards. As we all know, in Dominion it's better to get ahead and stay there, which is exactly why Mountebank is so good.
At any rate, this discussion has really made me evaluate the benefit of Swindling a Copper into a Curse, which I used to think was the cat's pajamas. It's certainly better than hitting Estate -> Estate, but unless you really need the Copper early on, or it's a game decided by only a couple of points, the effect seems fairly mild. The real payoff is in trashing costly actions (or in some cases, treasure).