Every two cards in a Dobble deck share exactly one symbol — never zero, never two. That is not careful hand-checking, it is a piece of nineteenth-century geometry called a finite projective plane. Here is how it works, and why the commercial deck is two cards short of the mathematical one.
Three ideas are enough to understand the whole construction.
Treat each symbol as a point and each card as a line drawn through some of them. A projective plane guarantees that any two lines meet at exactly one point — which is precisely the promise the game makes about any two cards.
Pick a prime number p. You get p+1 symbols on every card, p²+p+1 symbols in total, and p²+p+1 cards. Choose p and the whole deck size is decided; there is nothing left to tune.
The single-match property is a theorem, not a target. Once the plane is built correctly you cannot find two cards that share nothing, and you cannot find two that share more than one symbol.
Every legitimate Dobble size is p²+p+1 for some prime p.
Because the deck is a finite projective plane, in which any two lines intersect in exactly one point. Cards play the role of lines and symbols the role of points, so the geometric theorem becomes the game rule. It holds for every pair in the deck, without exception.
Both numbers are correct about different things. The retail box holds 55 cards. The complete mathematical set for p = 7 is 57. The published deck is the full set minus two cards, which is why you may see either figure quoted.
No. Seven symbols per card would need a projective plane of order 6, and the Bruck–Ryser theorem rules that out. The symbol counts jump from 6 to 8, which is why deck sizes jump from 31 to 57 with nothing in between.
Yes. Choose 13, 31, or 57 and the generator builds the plane for the matching prime and assigns your uploaded images to the symbol positions. You never have to check a pair by hand — the structure is correct before any picture is placed.