Institutions designed for worst-case knaves may actively corrupt virtuous people.

Evidence

It is granted on all sides that if we design institutions on the best-case assumption that people in power are all uncorrupt, and perhaps incorruptible, then the institutions may not work well in face of vice. But it is equally true that if we design institutions on the worst-case assumption that people are all corrupt, then the institutions may not work well in face of virtue: as we shall see later, for example, the institutions may actively serve to corrupt the uncorrupt! If we design institutions, however, on the middling-case assumption that people in power are corruptible, then we shall have to design institutions that work well both for agents who are not actually corrupt and, as a second line of defence, for agents who are. It is good to be given reason to cover both eventualities, and so I propose to conduct the discussion on the basis of this middlingcase assumption. — Philip Pettit, Republicanism, 950

Categorization

• DEEPTHINKING: A genuinely counterintuitive design principle: the paranoid assumption can be self-fulfilling and self-defeating. Worth returning to as it indexes dense institutional theory.
• XREF: connects to: Assuming officials are inherently corrupt may itself breed the corruption it seeks to prevent. — Frank Lovett, Republicanism, 63; Distrust-based institutional design can actively corrupt honest officials. — Philip Pettit, Republicanism: A Theory of Freedom and…, 950; Institutions exist to block the destructive tail risks of interacting with the untrustworthy — Russell Hardin, Trust and Trustworthiness, 511 (+1 more)

Broader Connections

• Distrust-based institutional design can actively corrupt honest officials. #corroborates (0.95) — Philip Pettit, Republicanism, 950
• Expectation of corruption can produce more corruption on its own. #corroborates (0.75) — Diego Gambetta, The Sicilian Mafia The Business of…, 1010
• Institutions exist to block the destructive tail risks of interacting with the untrustworthy #corroborates (0.72) — Russell Hardin, Trust and Trustworthiness, 511