This kind of debate isn't that unusual, is it? To figure out what calculation to use, one has to made tradeoffs about desirable properties such as resilience to

This kind of debate isn't that unusual, is it? To figure out what calculation to use, one has to made tradeoffs about desirable properties such as resilience to outliers, being bias-free, and/or having the best predictive performance in least-squared error. Bringing some grizzled statisticians into the conversation will probably clear a lot up. They are familiar with the pros and cons of various measures of central tendency. In the era of multivariate models, machine learning, and AI, some of the classic wisdom from good-old linear modeling gets overlooked.