A new paper from the School of Continuous Improvement catalogues five ways office tipping-competition players route around a confidence-weighted leaderboard once a reserved car space is attached to topping it, and finds the specific scoring formula barely matters.
A season-long study of nine office tipping competitions finds workaround tactics multiply once a reserved car space is tied to the leaderboard, whatever the scoring formula behind it
Attach a single reserved car space to an office tipping competition’s leaderboard, and the leaderboard stops being just a leaderboard. A new paper from the School of Continuous Improvement followed nine workplace footy-tipping competitions through a season to find out what changes once a real reward sits behind the ranking.
This sits alongside everything else the Adaptive Metrics Lab has traced about a measure and the behaviour that builds up around it. The car space is one of the smallest stakes we have found so far.
— Associate Professor Casimir Beng, Lead of the Adaptive Metrics Lab
Dr Beng, Dr Renke Sabel and Dr Marit Osayande compared five competitions that tie their confidence-weighted leaderboard to a reserved car space against four that publish the same leaderboard with nothing attached. Two coders read every submission’s timing, confidence choices and, where a competition ran one, its internal chat thread, sorting what they found into five recurring workarounds, from padding a sure thing’s confidence score to colleagues quietly splitting their picks to cover the perk between them. The pattern held all season and grew more pronounced as the fixture list narrowed toward the finals.
“The obvious next question is what happens once the car space is taken away again,” said Marit Osayande, Lecturer and Convenor of the Living Dashboard. “We don’t yet have a competition where we’ve watched that happen.”
A retrospective check, recomputing each round’s leader under a simple win-count instead of the platform’s confidence-weighted formula, changed almost nothing about how often the five workarounds turned up. The University reads that as the more useful half of the finding: the leaderboard’s arithmetic is doing far less work than the reward behind it.
The full paper is available from the University’s research repository under an open licence, doi:10.5555/slop.cm2xn5.