The Art Vote
Two cards. Pick the better illustration. Enough of those and the whole catalogue falls into order.
Where it stands
All of it →Nothing has been compared yet. 310 illustration cards are waiting.
The first few hundred comparisons decide the top of this list. Yours are the ones that build it.
Why two at a time
Every “best card art” list online is a poll, and a poll measures the order of its own list. The first three options win because they are first, a five-star scale collapses into fives and ones, and nobody has an opinion about card #147 until they are looking at it.
Nobody can rank eight thousand cards. Anybody can say which of these two is better. Enough of those forced choices and the ranking falls out of them — the same method used to rank chess players, applied to the thing it suits best. A card that beats cards which beat other cards climbs; a card that loses to everything sinks; and neither needs anyone to have seen the whole catalogue.
Each rating is printed with an error bar, because a rating built on four comparisons is not the same object as one built on four hundred, and a ranked list that hides the difference is lying by layout.
- provisional
- Fewer than ten comparisons. The number will move a lot.
- ranked
- Enough comparisons to place it, not enough to argue over one position.
- settled
- Forty comparisons or more. This is where the ranking stops moving much.
One vote per person per pairing, enforced in the database rather than by asking nicely, and a vote can only be cast on a pair we handed out in the last ten minutes. That does not make it unstuffable — nothing without accounts is — but it makes it expensive, and every comparison is kept so the ordering can be re-derived if it ever needs to be. No IP addresses are stored.