Interesting means something else for every player, it's highly subjective
Yes, it's definitely an interesting problem to even define "interesting" in a way that doesn't exclude kingdoms that really are interesting.
My tentative attempt is: generate kingdoms where no card is trivially irrelevant to an experienced player. I think it's fair to say that kingdoms with few relevant cards are among the least interesting kingdoms. Kingdoms with many relevant cards also are less likely to admit boring 1-card or 2-card strategies.
[..]
Well, what I meant by "interesting is something different for everyone" also considers that some people like to tweak the heck out of simple strategies and find the optimal number of Courtyards for instance. So even when there is only a small number of interesting cards, it can be enjoyable for people to optimize such "simulator" strategies even if they buy only one or two kingdom cards.
Trying to make every card in the kingdom a possibly interesting card is probably not the best approach. There's nothing wrong with having some "traps" (like Scout) or for example some bad Swindler targets (like Scout).
The problem is that Scout isn't a trap for good players. If there's some legit reason to think Scout is good, then I'd actually say it's not "trivially irrelevant to an experienced player", although with closer analysis it may prove to be irrelevant. If Scout is serving as a Swindler target, then it's relevant in that sense.
The rule I gave is a bit of a cheat by relying on the judgment of experienced players, but I could imagine approximating it in a way a computer might be able to implement. For example, if we have access to conditioned-gain statistics (RIP councilroom.com), maybe we could deduce that Scout is almost never gained in games that lack Swindler, Vineyard, and hybrid VP. Then we could use that information to kick Scout from the kingdom and replace it with something else.