Chris Elmendorf and Abby Wood have posted this draft on SSRN. Here is the abstract:

It is common to think of political elites—candidates, legislators, party officials, and campaign advisers—as specialists in learning the preferences of voters. But recent studies find that political elites believe public opinion within legislative districts to be more conservative than it actually is, and that extreme candidates are more electable than moderates (despite compelling evidence to the contrary). Campaign staffers overestimate their candidate’s electoral prospects. Moreover, natural and researcher-designed experiments show that informing legislators about constituency preferences changes roll-call votes, as legislators recalibrate to better represent public opinion in their districts.

This Article introduces the problem of elite political ignorance to the legal-academic literature. We make three principal contributions. First, we review political science findings on elite (mis)perceptions of voter preferences. Second, in light of ongoing technological developments that may provide elites with better information, we consider the likely benefits and costs of reducing elite political ignorance. The immediate impacts would probably include better alignment between the roll-call votes of representatives and the policy preferences of their constituents; reduced political polarization; less racial discrimination by campaigns and representatives; and lower-cost enforcement of the Voting Rights Act. However, over time, a reduction in elite ignorance could also engender more severe and enduring partisan gerrymanders; greater political and demographic skew in the population of regular voters; more inequity in the provision of constituent services; and micro-targeted campaigns that slowly erode democracy-sustaining norms and belief structures in the public.

We argue that most of the benefits of reduced political ignorance could be realized without incurring these costs if elites acquired better information about the distribution of voter preferences within districts, without learning the preferences of identifiable individuals. Our third contribution is to suggest public and private tools for realizing the twin objectives of constituency-level transparency and voter-level obscurity. Among other things, we argue that campaign-finance voucher programs have great and heretofore unappreciated potential for reducing elite ignorance, and that a suitably anonymized voucher regime could provide detailed constituency-level information while concealing the preferences of individual voters. Efforts to keep elites in the dark about the preferences of identifiable individuals will also depend on cooperation from social media companies.

This piece is worth your time.