KERALA RISING · FROM EXPORTING PEOPLE TO EXPORTING IDEAS – Durability: no significant decline in participation in the final two months versus the peak, indicating a forming habit rather than a novelty spike. – Cost: actual verification cost per action at or below the conservative one-rupee assumption. The failure threshold, stated before the pilot runs Equally, the conditions under which the pilot is declared a failure and stopped are fixed in advance, because a pilot that cannot fail proves nothing. The pilot is judged a failure, and the unwind protocol triggered, if any of the following holds at month twelve: – Participation never exceeds 15 percent of households in any month, indicating the loop did not take hold. – Logged participation rises but independentlyaudited street cleanliness does not improve by at least 15 percent, indicating people are gaming the points rather than changing behaviour, the most important failure to catch. – Participation in the poorest quartile runs below one-third of the richest quartile’s, indicating the design entrenches advantage and should not be scaled as built. 347