I think Vanilla Village is a good example of a card that would actually be a relatively high, but not bonkers, power level. How many boards is it skippable? It is often a key piece of an engine board. How often would you buy another village instead if a different one were available at $4? It costs $3, and I'd much rather have it than Shanty Town most of the time. Sure, it's boring, but it's appropriately costed.
It's especially interesting to compare it to cards like City, which, on the wrong board, is a $5 Village for most of the game. I've probably skipped City more than I have Village. When City's good, it's great, but it can be a trap, especially if your opponent ignores it and you can't empty another pile quickly enough.
Factoring in the cost when evaluating the cards is definitely the trickiest part, but it's a critical part. Expand would be utterly broken at $4 and would be an unskippable rock-crusher of a card on almost any board. As-is, it's often (if not usually) skipped. This is why the cost has to be taken into account. The rating gels it all down to one number that gives a general sense of how good the card is.
Never mind Village for a second; allow me to suggest a basic card rating comparison: Silver is better than Gold. It gives you $2 for only $3 investment, rather than $3 for $6. It's usually not skippable because of the need to hit $5 early. Sliver flooding is a thing. Gold flooding isn't. Sure, some cards give you free Gold, but does that make it a better card overall? Not really. The existence of Gold adds a lot to the intrinsic value of Tunnel; The existence of Tunnel doesn't add much to the intrinsic value of Gold. I've skipped Gold on plenty of engine boards, but I've only skipped Silver on BM boards that ramp to $6 almost immediately, like Gear-BM.
That said, you use Silver as a means to an end, not because you want to, but because you have to. Sometimes it even gets trashed along the way. Some people would even call it a glorified Copper. Then again, even Copper has utility (I wouldn't rate it a zero. That's for Curses and Ruins.) But all Dominion cards are a means to an end, so that's not meaningful.
I think the numbers will be interesting with a large enough sample size. Sure, some people will overrate and others will underrate, but that's the whole point of creating averages. Personally, once there's enough data, I'd be interested in just seeing a list of all of the card-shaped things sorted by rating the average rating shown to one or two decimal places.
