Some rumors suggest that the behavior is due to the piston disappearing during retraction, and while this may be a related problem, I think the immediate culprit is that the top surface of a piston is currently solid only in the unextended state, whereas extended pistons were never changed. A block update causes any tiles placed upon an extended pistons to behave accordingly, which the retraction causes. This(but not the dismissal of the disappearing piston hypothesis) is proven by placing a tile on a piston, powering the piston, and then placing a block adjacent to the tile.
Kevin O'Brien
Some rumors suggest that the behavior is due to the piston disappearing during retraction, and while this may be a related problem, I think the immediate culprit is that the top surface of a piston is currently solid only in the unextended state, whereas extended pistons were never changed. A block update causes any tiles placed upon an extended pistons to behave accordingly, which the retraction causes. This(but not the dismissal of the disappearing piston hypothesis) is proven by placing a tile on a piston, powering the piston, and then placing a block adjacent to the tile.