This isn't surprising because humans will do what they're incentivized to do (speaking mainly in a professional setting here).
Might be effective to add a measure to the transparency that shows something to the effect of, "likelihood to perform life-saving and risky procedures." Therefore a high win-loss with a lower risk threshold would be weighted lower than a so-so win-loss with a higher risk threshold.
Might be effective to add a measure to the transparency that shows something to the effect of, "likelihood to perform life-saving and risky procedures." Therefore a high win-loss with a lower risk threshold would be weighted lower than a so-so win-loss with a higher risk threshold.