I know there are at least hints of this in the secret histories ("this card never changed" and "this took a long time to playtest and changed a lot"), but what cards were among the easiest/quickest to both come up with and playtest, and which ones were among the hardest/longest to both come up with and playtest? Are there whole expansions that were generally easier or harder to finish (expansion size notwithstanding)? I realize this question may not be answerable, but just in case. Thank you.
I will pick two cards from each set.
Main: Mine is a day one card where the only change was "up to +$3" rather than "copper to silver, silver to gold." It matters now but did not much with just the main set. Witch went from "$3, they gain Curse" through "$5, pay $1 to give them Curse" to what it is.
Intrigue: Masquerade is a good example of a nontrivial card that didn't change. It was carefully built to make the most of passing a card left. Secret Chamber started as "victory cards are also Copper this turn" for $4.
Seaside: Several of these were good to go - Merchant Ship, Caravan, Bazaar, Warehouse. None of these had so many versions; maybe Outpost has the record here. I'm going to say this set was the easiest.
Alchemy: Potion never changed. Vineyard just went from $4 to $P. Maybe Philosopher's Stone had the most versions; the oldest one was an action for $3, +1 buy, +$1 per 4 cards left in your deck (did not count discard).
Prosperity: Platinum never changed. The $7's all just changed in cost, plus the "may" on King's Court; Expand I think spent the most time as is. A bunch of discard attacks tried out for the Goons slot; that version may not have had multiple versions, but it feels like the most work went into that slot.
Cornucopia: Remake and Hamlet never changed. Remake got more playtesting focus. Tournament took the most work, but Horn of Plenty also has a long history, starting with "+$1 per card you played this turn" in Intrigue.
Hinterlands: Cache survived unchanged from the first month or so of Dominion, and even got to keep its name. The Margrave slot ate up the most time, depending on what you count. For a long time there was a discard attack that hit you the turn they bought it. There were multiple versions and well hooray they're gone. Margrave itself descended from another attack tried in a couple versions that didn't work out. Then Margrave itself didn't change once I had that particular card, but it was a focus of testing because of the old-Crossroads / Margrave deck.
Dark Ages: Probably this set took the most work over all. This set was last, so ideas that sounded good but hadn't worked out trickled down into it, to be worked on one last time. And it changed themes and is large. What month are the fewest children born in? February. Anyway Armory and Altar are cards that never changed. The Knights probably took the most work, although there was certainly some time spend on a bunch of these.
Attacks take the most work, both playtesting/changing and also just thinking of good ones that feel new. Very basic effects are easy to think of, but some of my initial cost guesses were way off, and sometimes the simple cards didn't start simple. Some of the easiest things I listed were Seaside's "do something basic but next turn also" and Prosperity's "do the big version of something basic." "Choose one of these basic things" wasn't hard either. Whereas "care about variety," you can quickly list the basic approaches (variety in your hand, in play, in your deck), but the cards didn't just happen.