We often make lists of cards that are good and cards that are bad. And we look at openings, which openings are the best. But I think we don't often pay enough attention to WHEN cards are good and bad. Almost every card is better at some point in the game than at some other, form estates, which everybody knows get much much better over time (i.e. you want to trash them at the beginning and buy them at the end) to trading post, which is great on turn 1 and much worse later. And everything in between. There are hints of this analysis out there, but it's not well-developed.
Every card has this strength to some extent. Some aren't very interesting (silver, province, etc.). Some are incredibly important.
An example for you - the paradox of trashers
The most obvious application of this is to trashers. Chapel is by far the best it's going to be before the first reshuffle (I'll call this the zeroth reshuffle). To the extent that it's super-powered then, maybe average on the first reshuffle, and bad afterwards. The same goes for cards like trading post, steward, ambassador, remake, and even forge. The reason for this is that the purpose of these cards is to remove unwanted cards from your deck (well, not entirely, but most often), and in order to do that, you have to be able to get them in your hand together with those unwanted cards, which is by far easier early in the game, when a large percentage of the cards you draw these with will be unwanted.
Now, there's a whole bunch of trashers that I've left out, you might be thinking, and these are your trash-for-benefit cards. Why have I left them out? Because they function quite differently than the others. You use these cards to convert what are generally longer-term investments into immediate gains. What makes them different? The biggest thing is that they only trash one card at a time, which isn't really effective at trashing your deck into a compact engine. The other big thing is that they give you benefits in proportion to the cost of the trashed card. It's for this reason, along with the fact that when you trash those expensive cards, you want to do it at a time when you won't miss their presence as much, i.e. close to the last reshuffle, that trash-for-benefit cards achieve their highest power in the late game.
Time for a big list
Engine components are obviously most powerful later in the game, when you have sufficient density of them to be able to reliably chain them together in an engine.
Terminal draw tends to be stronger late in the game as well; not only are you drawing better things, but (assuming you've planned your deck well), your risk of collisions tend to be decreased then as well. And, these cards get 'caught in the reshuffle' much less later on, too.
Copper is best at the very beginning of the game and at the very end, when your money density is dreadful.
Gardens and Philosopher's stone obviously get stronger with time, as your deck gets bigger.
Venture gets stronger later too, as the average value of your treasure grows.
Silk Road, Fairgrounds, and Vineyards get better over time, as you get more and more likely to bump up to the next of their conditions (silk road especially, since it compounds on itself).
Sifters (e.g. warehouse) are best early, worst in the mid-late game, and decent again when you're greening late. The particular reason for their goodness early on is that not only do they skip your relatively high ratio of estates, they also help you get to your key one or two strong cards.
Duration cards should get a special note here, as they get significantly longer the later you are in the game, because they are much less likely to get 'caught in the reshuffle'
Then there are all kinds of 'interaction' cards.
Attacks
Cursing attacks have essentially two components. Curses are bad because they give you negative VP. This is unchanged throughout the game. Curses are also bad because they clog up your deck. The earlier you get the curse, the thinner your deck probably is, and the more times you're going to see it throughout the rest of the game. Combined with the compounding effect of earlier buys being more important than late ones, and cursing attacks are definitely early game cards.
Handsize-reduction attacks get to their strongest point just before greening happens. This is when you really need all of your cards, and you probably don't have all that much you'd be ok with discarding at that point.
Bureaucrat's attack is strongest when there's significant green in the deck, which is at the end but more especially the beginning of the game. And getting flooded with silver is generally better early than late, so it's pretty much an early game card.
Cutpurse is a huge early-game card; not only are you more likely to hit a copper earlier in the game, it's also much more likely to make a meaningful difference then, too.
Jester consistently gets stronger the later you go in the game, as there's better stuff to steal.
Inspection (i.e. spy-like) attacks remain fairly consistent throughout the game. Early on, you're more likely to be able to pin them to a dead card on top of the deck. But later, it's more painful.
Non-attack interaction
Possession is obviously strongest late in the game, when you can steal better stuff. On the other hand, you want to possess people as often as possible, and if you possess someone in the middle of the game, you do get the compounding effect of 'early buys are better than later'. So possession is actually fairly neutral time-wise - it's much more dependent on other factors, like what's in your opponent's deck, than time.
Masquerade is strong early in the game, because of the trashing effect. In certain circumstances, particularly if you can chain it with handsize-reduction attacks, or if your opponent has trashed even more than you, the passing effect can be like an attack, which gets more powerful with time.