More on the Varieties of Policymaker Ignorance
A taxonomy of ignorant policymakers
As promised on last week’s episode of The Week in Policymaker Ignorance, here is a bit more formal analysis of the different epistemic states that policymakers might occupy at any given time, with respect to any given policy goal. A similar analysis appears in my first book, F. A. Hayek and the Epistemology of Politics, and is discussed in Book Ten of Dialogues concerning Natural Politics.
Policymakers might be ignorant of some policy end desired or demanded by their constituents. They can fail to know of some specific goal that constituents want them to pursue. It is well to keep in mind how faint are the signals that policymakers receive from constituents in this regard. For starters, there really isn’t such thing(s) as “what constituents (or voters) want” (a problem recently adroitly analyzed by Cyril Hédoin at The Archimedean Point). Constituent demands are always vague and in some degree of tension, both in the minds of constituents and as conveyed to policymakers. However, even if there were some univocal ranking of constituents’ preferences, the mechanisms we have conjured for communicating these preferences to policymakers are not well-suited to the task. Whatever the virtues of democratic mechanisms, the clear expression of constituents’ specific policy demands is not among them. This problem is even worse in non-democratic contexts, of course.
But, even if policymakers could acquire from constituents some coherent preference ranking of their policy wishes, demands, needs, desires, etc., policymakers might still be ignorant how to realize associated ends. Even if constituents could convey to policymakers their unambiguous preference to have, say, the homelessness problem addressed before all others, policymakers might fail to know how to mitigate homelessness and its attendant evils to an extent adequate to satisfy constituents. No mechanism, democratic or otherwise, exists to address ignorance of required know-how. If policymakers do not already know how to realize constituents’ preferred policy goals, the results of elections will not convey the required causal knowledge.[1]
So, policymakers can be ignorant either of or how to satisfactorily realize various policy goals preferred by their constituents.
But, these are not the only kinds of policymaker ignorance. We need to differentiate policymakers’ first-order ignorance of some policy end, or of means to deliberately realize it, from their possible second-order ignorance of their first-order epistemic condition regarding a policy goal. Policymakers might be ignorant either about or how to realize some policy objective, and they might be ignorant about this ignorance. Either first-order or second-order ignorance can distort policymakers’ incentives to pursue their constituents’ preferred policy objectives, and, thus, can affect whether policymakers pursue constituent-minded or other goals.
In much of my work on the problem of policymaker ignorance, I start from the simplifying assumption that policymakers know they are ignorant of some of the knowledge required to deliberately realize a given policy goal. That is, I assume that policymakers know their constituents’ policy preferences (perhaps counterfactually), but not how to realize goals associated with these preferences, and that they recognize this latter ignorance.
Policymakers who are second-order knowledgeable about their first-order ignorance know that they do not know how to deliberately realize the relevant objective and that pursuing the objective despite their ignorance will end in failure unless spontaneous considerations outside their ken and control intervene in the causal nexus to counter the goal-undermining consequences of their ignorance. Other things equal, knowledge of their ignorance to deliberately realize some potential policy end makes its pursuit a less attractive option to policymakers, who will either ignore it altogether or systematically discount it in their preference rankings relative to other options that they take themselves to be better epistemically positioned to realize. If, relative to other, less constituent-minded, pursuits, policymakers know that they are ignorant either of or how to realize objectives their constituents demand, the latter objectives are relatively unlikely to be pursued compared to the former, less constituent-minded, goals.
To offer a more concrete example, if policymakers know that they do not know either that (say) homelessness is their constituents’ top priority or how to adequately address homelessness through policy measures, but know that they know how to engage in a public relations campaign to convince constituents that they earnestly care about, want to realize, and are hard at work pursuing, goals associated with their constituents’ interests, then sincerely pursuing the satisfactory mitigation of homelessness will be systematically discounted in their preference rankings relative to the option of merely pretending to care about and pursue goals in their constituents’ interests.
Imagine that policymakers know that constituents most want homelessness addressed, but also know that they have no idea how to address it in a way and to an extent that will satisfy constituents. Are they likely to earnestly pursue the mitigation of homelessness, with all of the obvious risks to their political futures that this choice entails, or are they more likely to merely go through the motions, declare “war on homelessness,” and create eye-catching and ostentatious, but mostly impotent, “watchdog committees” and “expert blue-ribbon panels” ostensibly committed to addressing homelessness? If policymakers know that their constituents cannot distinguish earnest from merely pretended pursuit of a homelessness-mitigation strategy and that homelessness is unlikely to be adequately mitigated either way, but that constituents are inclined to reward policymakers who appear to “do something” about homelessness, policymakers would seem more likely to flatter to deceive constituents than sincerely pursue their policy preferences.
When policymakers know that they are relevantly ignorant, they tend to choose the epistemically easiest option that they think they can justify to constituents, while pretending that they are doing something else, i.e., earnestly trying to satisfy constituents’ demands.[2]
There are still other kinds of policymaker ignorance.
Policymakers might be second-order ignorant of their relevant first-order knowledge. That is, policymakers might mistakenly believe that they are ignorant with respect to a given policy objective. In fact, such policymakers know enough to realize the goal, but erroneously believe that they are ignorant of some of the required knowledge. They mistakenly believe that acting on the objective will end in failure unless spontaneity intervenes.
Like policymakers who know they are ignorant, policymakers who are ignorant of their relevant knowledge will either ignore the given objective altogether or systematically discount it in their incentive structures relative to other objectives they believe themselves more knowledgeable and capable to realize. Policymakers who are second-order ignorant of their first-order knowledge are less inclined, other things the same, to pursue the given objective than they would be if they correctly assessed their first-order knowledge. If they recognized the adequacy of their knowledge to deliberately realize the goal, they would be, ceteris paribus, more likely to pursue its realization than they are when they mistakenly believe themselves ignorant of some of the required knowledge or capacities. Such policymakers are relatively more likely to believe, albeit perhaps mistakenly, that they know better how to pretend to pursue constituents’ interests than how to deliberately realize goals associated with these interests. Like policymakers knowledgeable of their ignorance, they are more likely to flatter to deceive their constituents.[3]
Policymakers can also be ignorant of their ignorance with respect to some policy objective. Such policymakers are potentially quite dangerous, because they can too easily convince themselves that they know enough to deliberately realize ends that they do not in fact know enough to realize. They face an epistemically distorted incentive to pursue particular policy objectives that they would not face if their second-order knowledge improved enough to permit them to recognize their first-order ignorance.
The late former Defense Secretary Donald Rumsfeld may have been onto something when he lamented the effects of what he claimed were “unknown unknowns” on the decision of the Bush administration to invade Iraq in the erroneous belief that Saddam Hussein possessed weapons of mass destruction. Members of the Bush administration falsely believed – put another way, they did not know that they did not know – that Hussein possessed WMDs and were incentivized by this false belief (and perhaps other relevant considerations) to invade Iraq. If Rumsfeld is to be believed, members of the Bush administration would have been less inclined to invade Iraq had they correctly assessed their relevant ignorance.
Policymakers who are ignorant of their ignorance mistakenly believe that pursuit of relevant policy objectives will end in success and, so, do not discount such pursuit as deeply as they would, if they correctly assessed their, in fact, deficient epistemic circumstances.[4]
Finally, consider the privileged epistemic status of policymakers who know that they know how to realize a policy objective, policymakers who are second-order knowledgeable of their relevant first-order knowledge. These policymakers know how to realize a given policy goal and know that they know how to realize the goal.
I have argued that such policymakers are like the wise captain of the ship of state that Plato discusses in Book VI of The Republic. The wise captain (or “true pilot”) of the ship of state knows that he possesses all of the knowledge required to steer the ship to safe harbor. The wise captain of the ship of state can proceed confidently in the pursuit of his constituents’ policy demands; ignorance cannot dissuade him from their pursuit.
Of course, whether the wise captain of the ship of state is motivated to pursue goals associated with his constituents’ interests is distinct – and logically ancillary – to the question whether he is knowledgeable and capable enough to deliberately realize constituent-minded goals. In a world populated exclusively with wise captains of the ship of state, i.e., omniscient and omnipotent policymakers, ignorance cannot distort policymakers’ motives, because they’re not ignorant. It is only in such a world that we need to be singularly focused on policymakers’ motives, on whether they are naturally inclined to self-interested or constituent-minded policies. In all other contexts, their motives, intentions, incentives, etc., to be more or less constituent-minded will be determined to a large extent by their relevant epistemic circumstances.
To sum up, there are four different epistemic states that policymakers can occupy at any given time with regard to some potential policy goal. They might be knowledgeable of their relevant ignorance or ignorant of their relevant knowledge. They might be pretenders to knowledge, who are ignorant of their relevant ignorance, or they might be true pilots of the ship of state, who are knowledgeable about their relevant knowledge. Each of these epistemic states implies specific effects on policymakers’ incentives to pursue constituent-minded or other policy goals.
[1] Of course, it is the ostensible function of scientific and other “experts” to mitigate this problem. Public officials may not know how to solve various problems, but their expert helpers are supposed to know, an assumption called into doubt in Book Five of Dialogues concerning Natural Politics.
[2] I have argued in several places (e.g., here, here, and here) that this explains much of what was observed during the pandemic, i.e., the choice of an epistemically easier policy option – lockdown – and the rejection of the more epistemically challenging option of a focused protection strategy, and the endless political theater that policymakers around the world engaged in to publicize their purported concern for constituents.
[3] Of course, it is relatively unusual that policymakers underestimate their knowledge. Simple observation of political history suggests that policymakers tend far more often to exaggerate than deprecate their knowledge.
[4] In his 1974 Nobel Prize lecture, cheekily titled “The Pretence of Knowledge,” F. A. Hayek diagnosed central bankers and other economic policymakers as ignorant of their relevant ignorance. They think they know how to manage the economy, so they make decisions on the basis of this belief, but they do not, in fact, know how to manage the economy, so their decisions succeed rarely, and only inasmuch as spontaneous forces compensate for their relevant ignorance.