CIA? See, I Am Policy Relevant

As most things I encounter, This New York Times story got me to thinking about, well, me.  Specifically, the article—discussing the Senate’s attempts to oversee the CIA’s interrogation programs—touches upon two strands of my research that, at first glance, might appear related only in that they both use mathematical models to analyze and characterize political phenomena.  One of these strands revolves around the use and acquisition of information in political institutions (specifically, but not exclusively, bureaucratic agencies).  The second focuses on how one might discriminate between legitimate political procedures and illegitimate ones.

Information and Oversight: Ex Ante vs. Ex Post Incentives. One story at the heart of the article revolves around the existence of a classified internal CIA report that members of the Senate Intelligence Committee would like to see.  The central question here is whether the internal report confirms the Senate’s (still classified) own report and/or contradicts the subsequent disclosures/admissions/justifications offered by the CIA to the Senate Intelligence Committee.

I have written, with Sean Gailmard, several articles and working papers (e.g., here, here, here, and here) and a book, Learning While Governing, that examine in various settings the incentives for individuals within hierarchical organizations to acquire, use, and honestly report information to their superiors.[1]

While I am sure you want to read each of these in their entirety, with the proper “SPOILER ALERT” I can summarize a principal thread linking the theories as follows:

The incentive to collect, use, and share information in a faithful way depends on expectations about who will subsequently get the information and how they will use it.

In terms of the Senate Intelligence Committee’s oversight efforts, the implication of this is as follows:

To the degree that the information revealed and shared within the CIA in writing its internal report is potentially used by the Senate to punish the CIA (in whatever form), successfully extracting the report may hinder the CIA’s efforts to internally collect and share information in the future.

Of course, we are not the first to make this point,[2] but it is often forgotten.  Put another way, attempts to keep the CIA’s internal report hidden need not indicate nefarious motives.  Rather, there is a coherent logic that justifies a lack of transparency, or stonewalling, in somewhat ironic pursuit of information in the future.[3]

Colloquially understood, “oversight” is an ex post phenomenon (it occurs after the actions of interest).  But game theoretic institutional analysis helps illustrate and remind us that this ex post procedure can have ex ante effects (individuals may change their behavior—sometimes in unexpected, or “perverse,” ways—in anticipation of it).

Legitimacy and Oversight. In line with the Senate Intelligence Committee’s pursuit of the CIA internal report, some Senators are also pursuing the Department of Justice’s internal classified memos that supposedly provide a legal rationale that justifies some of the interrogation techniques in question.  Quoting from the end of the article,

Much of Tuesday’s hearing was consumed by a debate about whether the White House should be forced to share Justice Department legal memos.

Under polite but persistent questioning by members of both parties, Ms. Krass repeatedly said that while the two congressional intelligence committees need to “fully understand” the legal basis for C.I.A. activities, they were not entitled to see the Justice Department memos that provide the legal blueprint for secret programs.

The opinions “represent pre-decisional, confidential legal advice that has been provided,” she said, adding that the confidentiality of the legal advice was necessary to allow a “full and frank discussion amongst clients and policy makers and their lawyers within the executive branch.”

Senator Feinstein appeared unmoved. “Unless we know the administration’s basis for sanctioning a program, it is very hard to oversee it,” she said.

In an article and a forthcoming book, Social Choice and Legitimacy: The Possibilities of Impossibility, Maggie Penn and I have tackled the question of when a government policy is legitimate, which colloquially means that it is consistent with an underlying set of principles or criteria (e.g., fairness, efficiency, equality, etc.).  The argument is social choice theoretic, and boils down to the following.

A policy is legitimate if one can construct a sequence of decisions that justify the policy in the sense that no earlier, intervening decision is strictly better than the final policy, and every unchosen policy that is arguably better than the final policy is itself inferior to one or more of the intervening decisions in the sequence that justifies the final policy.

Informally, the theory provides a precise characterization of justifying a choice through providing (or accompanying the decision with) other decisions that can serve as counter-objections to any alternative choice that one might propose to replace the final choice.

A direct implication of Arrow’s impossibility theorem is that there may in general be multiple legitimate decisions.  However, it is generally the case that there are many illegitimate—and non-legitimizable decisions, too.

A fundamental starting point/implication of our notion of legitimacy is that, in order to verify the legitimacy of a decision, one must have access to the rationale or justification for the decision.  As portrayed in our work, this can be thought of as an argument about both the principles that ought to (and/or did) guide the choice and an demonstration that the final decision is in fact justifiable—sometimes necessarily through a complicated process of reasoning.

In a nutshell, then, our work provides a systematic (though, of course, contestable) argument in favor of the executive branch sharing the DOJ memos with Congress.

Putting It All Together. In addition to being applicable to the same story, the two strands of research—which leverage different (though I just argued not that different) veins of theory—also complement each other, particularly in the light cast by situations such as this.  In particular, placed side-by-side, they demonstrate one of the most fundamental realities of politics: every choice we debate for more than a couple of seconds necessarily involves an important trade-off.

In other words, seeking the truth can sometimes ironically further its obfuscation, just like banning certain types of Superbowl Ads can ironically create an incentive to create ads that will be banned.  Finally, recognizing the ubiquity of such trade-offs leads to the recognition of the fundamental importance of social choice theory. To quote Maggie Penn and myself,

Rather than taking [the impossibility theorems of social choice theory] as negatives, to be either ignored or worked around, … these results motivate the entire study of politics. The potential irreconcilability of multiple societal and/or individual goals is exactly the raison d’être of government.

This has obviously been an even more blatantly self-promoting blog post than usual. Instead of feigning an apology for that, I leave you this, the greatest pop song ever written.


[1] In solely authored work, I also tackle this topic in this article and this working paper.

[2] To be fair, though, I think we make it in a variety of new ways and draw new institutional implications of it.

[3] For the cognescenti of these kinds of models, there is always a delicate question of commitment at this point.  I am setting that to the side.