All men contain several men inside them, and most of us bounce from one self to another without ever knowing who we are. - Paul Auster
In reinforcement learning literature there is something known as the credit assignment problem. In simplified terms it comes down to the fact that it is hard to assign credit to activities of artificial agents in order to provide them with a good reward signal for their actions. Obvious examples are games like chess. Creating a software agent that plays chess is somewhat tricky because it is hard to assign a score to all the intermediate moves in a game that end in a checkmate. When playing chess the signal is whether the game is won or lost but it is very hard to tell which actions were the ones that were responsible for the winning move. Human players are able to analyze games and pinpoint the positions that determine the fate of the winner and loser but it's not clear how to do this for artificial software agents.
Zooming out from games like chess it is easy to generalize and apply the same kind of analysis to human activities like politics. It is often not clear which political decisions will lead to good collective outcomes and which ones will lead to bad ones. I don't know if anyone has done research on how to properly assign credit to politicians and their decisions but seeing how most decisions are currently made I think it's fair to say we are still a long ways away from a proper theory of political action and credit assignment. One could even argue this is a very good reason to make sure political systems and the politicians that operate within them remain accountable to the people they are meant to serve because unaccountable politicians will more than likely end up with a distorted perception of their own performance.
A good traveler has no fixed plans and is not intent on arriving. - Lao Tzu