Glen wrote: > mgd circa Wed Jan 28 00:06:38 EST 2015: > > Consider counting boolean values. > > Trial one gives `1', `0', `0'. > > Trail two gives `1', `0', `0', and, `GodIsGreat'. > > The practical question is whether or not its good or adequate practice > to throw away the outlier. Obviously, I tend to avoid throwing away > the data even if it's nonsensical. Were I one of the tabulators, I'd > have to say it's either inconclusively between 1/4 and 1/3 or a broken > experiment, probably both.
The practical question to me is whether it is good to tolerate the design and proliferation of broken experiments. It's not even an outlier, because it is indeterminate. If the type system was inadequate it should have been redesigned before starting to state constraints and relate them in a consensus language. Marcus ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
