I agree that this method will produce a better estimate of expected mean real world performance under certain load conditions, but still contend it just muddies

I agree that this method will produce a better estimate of expected mean real world performance under certain load conditions, but still contend it just muddies the waters about which solution is in fact the fastest. And here is Andrei Alexandrscu arguing the same in 2012: https://forum.dlang.org/thread/mailman.73.1347916419.5162.digitalmars-d@puremagic.com?page=3 https://forum.dlang.org/thread/mailman.73.1347916419.5162.di...