There is a sender and two receivers, L and R. The sender has information, the state of the world, and can also make a decision. There is a socially optimal decision, the ethical decision, which depends on the true state of the world. The sender’s preferences depend on the gap between her decision and the ethical decision and on her “reputation” – more on reputation below.

If the sender only cared about making the ethical decision, her problem is simple – she implements the socially optimal policy for the privately observed state. But she also cares how much support her decision gets from the two receivers.The sender might care about her support because it affects the long run viability of the institution she belongs to. Or she might just want to be seen as being bipartisan. So the sender’s reputational concerns may arise for ethical reasons.

The receivers do not know the sender’s preferences – is the sender ethical or biased towards one receiver or the other? A receiver is more likely to support the sender the closer is the sender’s decision is to his own policy preference. The receivers’ preferences have a common value component. They might potentially update their preferences based on the information content of the sender’s decision.

Suppose the sender faces a decision where where the ethical decision favors L. If she makes that decision, she loses support from R and that is bad for her reputation. So, she biases her decision rightward to maintain support. Perhaps she makes the same rightish decision for a set of information states from left to right in a pooling equilibrium. She loses support from the L receiver but gets more support from the R receiver. For ethical reasons, she makes  a decision which is not ethical from the one period perspective.

Hence, the press presents “both sides” even when it knows one side is right. The press has a super high standard of proof for calling someone out as a liar. Obama gave McConnell veto power over announcing Russian interference in the election. He could have gone ahead and announced the interference himself at the cost of being non-bipartisan but he did not. Comey broke FBI and DoJ policy by commenting when himself dismissing the case against Hillary Clinton. But he went with policy by not discussing the investigation of Russian interference and communication with members of the Trump campaign.

So, this framework helps to unify the strategic logic behind the way different institutions and decision-makers operate(d). But it also reveals a flaw in their decisions. This comes from the common value component of the receivers’ preferences. Receivers update on what the right decision is based on what they learn in equilibrium from the sender’s action. Some may switch their preference from left to right or vice-versa based on what they learn.  In fact, majority opinion may change. If there is some outcome that depends on majority opinion like an election, this switch can make dramatic consequences. If the cost of making an error is large,  the sender should condition on this event when they determine their message. Hence, if the reputational loss from changing majority opinion is large enough, the one period ethical decision and the long run ethical decision will be the same.

So, in this interpretation, the main Comey-NYT-Obama mistake was not to condition on the event they were pivotal but instead to focus on what was probable given the polls.

(HT: This is speculation based on Stephen Morris’s paper on Political Correctness and Ely-Valimaki on Bad Reputation. Here I am positing multiple principals rather than just one principal and there are is some common value component to principals’ preferences.)