The puzzles are often likely to be "solved" by guess rather than by analysis. Imagine a puzzle with only two legal moves in the starting position. Any player will guess it with 50% probability, but to see the winning variant may be beyond the abilities of a GM. Consequently at such puzzles the GM's rating will be same as the rating of a beginner.
The cause of that is the rule that in every position there must be only one winning move. Can this rule be dismissed? Let the player proceed further. The question arises, at which point to stop. That could be when the number of winning moves in the achieved position is enough large (e.g. 4 or more), which likely indicates that the position is "generally won" rather than a continued puzzle. Possibly there exist other criteria, I just suggest to think in that direction. The single winning move rule was imposed in the pre-computer era, now we have more opportunities.
The cause of that is the rule that in every position there must be only one winning move. Can this rule be dismissed? Let the player proceed further. The question arises, at which point to stop. That could be when the number of winning moves in the achieved position is enough large (e.g. 4 or more), which likely indicates that the position is "generally won" rather than a continued puzzle. Possibly there exist other criteria, I just suggest to think in that direction. The single winning move rule was imposed in the pre-computer era, now we have more opportunities.