In colonial India, the British government offered a bounty for every dead cobra delivered. The intention: reduce the dangerous snake population. The result: Indians began breeding cobras to collect rewards. When the program was cancelled, breeders released their now-worthless snakes—leaving more cobras than before. This is the Cobra Effect: a well-intentioned intervention that directly causes the problem it was meant to solve.
Part of our Mental Models Guide guide.
Horst Siebert coined the term in his 2001 book Der Kobra-Effekt, documenting how perverse incentives systematically produce outcomes opposite to their intent. I’ve watched this play out in education for five years. Our school implemented a “reading points” system—students earned rewards for finishing books. Within a month, kids were sprinting through the thinnest picture books they could find, gaming the metric instead of reading. We solved a behavior we never had.
How Perverse Incentives Form
The mechanism is always the same: a measurable proxy gets rewarded, and people optimize the proxy rather than the underlying goal. Charles Goodhart captured this as Goodhart’s Law: “When a measure becomes a target, it ceases to be a good measure.” The British cobra bounty measured dead snakes as a proxy for safety—so people produced dead snakes, not safety.
Siebert’s 2001 analysis showed the pattern repeats across domains: Soviet factories measured by tonnage produced absurdly heavy goods; sales teams measured by call volume make useless calls; hospitals measured by wait times discharge patients prematurely. Every system that rewards a proxy instead of an outcome eventually creates cobra farms.
Three Modern Cobra Effects
Content marketing metrics. Companies optimize for pageviews, so they produce content optimized for clicks—not value. Traffic rises, conversions fall. The proxy wins, the goal loses.
School accountability testing. High-stakes standardized tests were designed to measure learning. Schools began teaching to the test—drilling tested skills while cutting arts, recess, and deep inquiry. Test scores improved; actual learning arguably did not.
Wellness programs. Corporate step-count competitions reward people who already exercise, while sedentary employees game trackers or simply feel excluded. Health costs don’t budge.
The Fix: Design for Outcomes, Not Outputs
Siebert’s prescription is uncomfortable: you must measure what you actually care about, even when it’s hard. If you care about reading comprehension, test comprehension—not books finished. If you care about employee wellbeing, survey wellbeing—not steps walked.
After scrapping our reading points system, we moved to monthly book discussions where students had to explain ideas to classmates. Harder to game, harder to measure—and the kids actually read. The cobra farms closed.
Before designing any incentive system, ask one question: what behavior will a rational person take to maximize this reward? If the honest answer isn’t the behavior you want, you’re building a cobra farm.
References
Siebert, H. (2001). Der Kobra-Effekt. Deutsche Verlags-Anstalt.