Interpretation of stockfish evaluation

[wildfang2]

The new normalized evaluation is now linked to the probability of winning, with a 1.0 pawn advantage being a 0.5 (that is 50%) win probability.

What does that mean? Position that evaluated as 1.0 has 50% to have forced win (no matter what opponent does, the game result in his loss) ? or does that mean that if stockfish plays the same position with itself, for example, 10 times, so 5 games will result in win? but if so, how is it possible, like every time stockfish will choose another move in this pos?

[vdbergh]

It has to be understood in a conditional sense. It roughly means that in a suitable “representative” corpus of positions of Stockfish playing against itself, 50% of the positions with evaluation 1.0 belong to a won game.

[wildfang2]

so does it mean that 1.0 position has 50% chance to have forced win?

[wildfang2]

and not every position that results in 1.0 wins?