A research poster from the School of Continuous Improvement compares a standard stopping rule with the best possible stop across a thousand runs of one heat calculation.
A thousand trial runs on a cooling rod set a standard stopping rule beside the stop that hindsight would have chosen
Every calculation that improves by repetition has to be told when to stop, and the School of Continuous Improvement has now measured what the telling costs. Stopping Before the Valley Floor: Discrepancy principle versus oracle stopping in a backward heat problem is a research poster by Dr Osei Vandermeer, Research Fellow and Convenor, Review Cadence Observatory, with Dr Bram Ntuli, Lecturer and Convenor, Improvement Grand Rounds, and Associate Professor Casimir Beng, Associate Professor and Lead, Adaptive Metrics Lab.
Set side by side on the poster are a long-established stopping rule and the stop that hindsight would have chosen, across a thousand trial runs of a single problem: working out how warm a rod was from a later reading of it. The rule stopped ahead of hindsight in every run, often by a wide margin in steps, and gave up comparatively little in accuracy for doing so.
I spent some years in a trade where finishing early was a line on a report. It is a relief to work on a problem where it is a number, and a smaller one than I had been led to fear.
— Dr Osei Vandermeer, Research Fellow and Convenor, Review Cadence Observatory
The University recognises in the poster a school prepared to turn its own name into a question and to accept a measured answer. Dr Ntuli said the three authors had “agreed at the outset on when we would stop, which in the circumstances seemed only polite”.
The full poster is available from the University’s research repository under an open licence, doi:10.5555/slop.x27jfz.