Bug in the test cases or in the description?

According to the test case provided, it says [1,2,2] wins [1,1,3] but it is not true. Actually, out of the possible 9 pairs of numbers, the player wins in only 4 pairs.

I think the checking function doesn't work as is stated in the description of the problem. The description says "output some die which will win against opponent more than half the time," but the checking function counts the player's and the opponent's win, and it returns true if the number of the player's wins is bigger. This is not straight understanding of the description. I mean it is just checking if the player is more likely to win than the opponent. It doesn't count on the draws. Even if the probability that the player wins is higher than the probability the opponent wins, the probability the player wins might be lower than 1/2.

The judging function or the description should be rewritten. I think rewriting the description is easier.