Wikipedia defines a perverse incentive as one “that has an unintended and undesirable result that is contrary to the intentions of its designers”. The classic anecdotal example of this corollary to Goodhart’s Law is The Cobra Effect.
Legend has it that Way Back When the British wanted to do something about the population of highly-venomous cobras in Delhi, so they started offering a bounty on cobra heads. On its face this may not seem like an unreasonable approach, but once cobras became a monetizable commodity the natural thing to do was to become a snake breeder (thus increasing the overall cobra population). When the British government caught wind that this was happening they discontinued the bounty…at which point all of these now-worthless and still highly-venomous snakes were released into the wild (thus increasing the overall cobra population in the wild).
Clay Shirky references a study demonstrating an example of the inverse of this - a disincentive that actually encourages the behavior it’s supposed to discourage - in his talk on cognitive surplus. The tl;dr of the study is as follows: In order to test the theory that (financial) penalties can be used to curb undesirable human behavior, the researchers did an experiment whereby they introduced a late-pickup fine in a number of daycare centers. “If you pick up your kids more than 10 minutes past the appointed pickup time then we’re going to charge you ten bucks.” The result of this experiment was that the average number of late pickups actually doubled. As indicated by the title of the paper (“A Fine is a Price”), attaching a financial penalty meant that being late was no longer a breach of social contract. Instead, being late was simply an additional service for which you were billed. The super-interesting bit about this is that when the experiment was through and the fine was removed the number of late pickups did not decrease. The conclusion drawn: not only can introducing an intended (dis-)incentive break a culture of (social) accountability, but also once that culture is broken it stays broken.
This is fascinating…and this is scary.
Extending this to industry, Bradilation is perhaps one example of the Cobra Effect - if the goal is to close out tickets in a timely fashion, you got it. Another might be an effort to decrease service latency/errors - as it turns out it’s super-efficient to return a completely-empty 200 OK response! Another might be trying to drive down GCN counts - the easiest way to reduce the number of GCNs is to simply not open GCNs, yeah?
Hyperbolic?
Perhaps.
…but also, I think there’s a lesson here around being vigilant about any metric that might undermine a culture of accountability, that might provide even subtle back-pressure on individuals who are generally trying to Do the Right Thing (TM).
Be careful what you wish for.