Distance Weights Flip the Vote

A query point's five nearest neighbors, in order:

neighbor	distance	class
1	1.0	A
2	3.0	B
3	4.0	B
4	5.0	B
5	10.0	B

KNeighborsClassifier(n_neighbors=5) defaults to weights='uniform'; weights='distance' weights each neighbor by $1/d$ . What does each setting predict?

uniform predicts B, and distance also predicts B by 22.0 to 1.0, since the weights are the distances themselves.

uniform predicts B by 4 votes to 1; distance predicts A, 1.000 against 0.883.

Both predict B: the closest point is still one neighbor, and no positive weighting can overturn a four-to-one majority.

Both predict A, since weights='distance' is the default and the nearest neighbor already dominates.