Introduced as a bounded exception or emergency measure, a change runs long enough to show its real effects, and that operating record becomes the case for making it standard.
An example
During a winter of ice storms, the vice president of field service at a regional telecom provider asked for a temporary exception: technicians could close repair tickets on site without a supervisor's sign-off, for ninety days, to clear a backlog of 9,000 outages. She had wanted the change for years and said so in the request, along with the sunset date and a named owner in quality assurance who would review results before it expired. The exception ran through the storms. Repeat visits fell, average restoration time dropped by a day, and an audit of a sample of closed tickets found error rates no higher than before. At the ninety-day review she presented the data and asked for the exception to become standard practice. The operations committee approved it with a quarterly audit attached.
Why it works
Bounded framing makes approving the exception look easier, since approvers are not setting policy, and the urgency makes it look better, a sign of responsive leadership. Once the exception has run without harm, reverting looks worse, because the old rule's value rested on harms that never appeared, and costlier, because reversal now means disrupting working processes and retraining staff. Four things carry it: a credible special case at the start, enough time for the exception to become routine, an exception that runs without conspicuous failure, and decision-makers whose attention does not return to the original carve-out.
The thinking behind it
Avinash K. Dixit Dixit and Pindyck's Investment under Uncertainty (1994) treats the ability to delay an irreversible commitment as an option with value, since waiting brings better information. A ninety-day exception is a cheap way to learn before committing to a permanent rule, which is why approvers grant it readily. The sunset keeps the option open, and the operating record is the information it buys.
William Samuelson Samuelson and Zeckhauser's Status Quo Bias in Decision Making (1988) found that people disproportionately keep whatever is framed as the current state, with transition costs among the reasons; a later replication supported most of their scenarios but not all. After ninety days of technicians closing their own tickets, the exception is the arrangement people know, and reverting carries transition costs. Presenting the record at the sunset uses that advantage openly, through a normal decision.
When it fits
Changes whose benefits are hard to prove in argument and easy to prove in operation are the candidates, at a moment when a real special case justifies trying one. You need standing to request the exception, the ability to measure it honestly, and a willingness to let the sunset do its job if the results disappoint.
How to use it well
Request the exception for the real reason it is needed now, and state your longer-term view in the same request. The exception gives you evidence. Asking for permanence on that evidence, in the open, is a fair fight you are well placed to win, and it keeps the carve-out above suspicion.
- Write the sunset date and a named reviewer into the approval yourself.
- Measure the effects honestly, including the ones that cut against you, and audit a sample.
- At the sunset, present the operating record and ask for a decision on permanence through the normal forum.
What it costs and where it backfires
Visible problems during the exception hand your opponents proof. Approvers who learn you always meant to make it permanent may refuse your next request. Agreeing to an expiry you plan to dodge, inventing the emergency, or leaving out the date and owner so the carve-out drifts into the background turns an honest trial into a trap.
When it is used on you
Before granting any exception, write the sunset into the approval and name someone to enforce it. Ask the requester directly whether they intend to seek permanence. An auditor or board member who tracks carve-outs can catch the second stage before it lands, and can insist that permanence be decided as a new policy.