Our final concern with a federation's safeguards is their imperfection. When safeguards make mistakes, they levy their sanctions too frequently, or not frequently enough. If they systematically fail to react, then the federation becomes vulnerable to opportunism. On the other hand, a safeguard might punish too frequently. While punishment will occur in every federation, if the safeguard punishes too frequently, frustration with the system mounts, as membership in the union becomes less attractive than an outside option of independence. Under either circumstance—sanctions that come too frequently or not frequently enough—the federation is jeopardized.

Referring to the baseline model of safeguards in Chapter 4, a safeguard's imperfection has two sources: an observation (the signal, ω), which is the safeguard's impression of reality, and a threshold (T), its “trigger point,” when the safeguard will levy its sanction. Imperfections arise because safeguards are not necessarily designed efficiently. Safeguards are often political creatures, or established for other purposes, or have interests of their own and reasons for acting that are peripheral to solving a federation's compliance problem. Tangential goals motivate the safeguard to trigger when it does; formally, its threshold is not necessarily set efficiently. It is also likely that what is optimal may change over time, converting an early efficiency into a flaw.

If the only concern were that a safeguard would fail to react, then the solution could be captured by two words: add safeguards.