Re: Transmission, Inheritance, Emulation 18
Just fwiw. My background (for 17 years anyway) was statistics in a human performance research environment. I've read Gladwell's book and although I have a few issues with his work, overall I found it a fun read. He is using the term "outliers" the way we used it in statistics. There is no connotation of "good" or "bad". Those are labels that depend on some other value judgement or set of criteria. An outlier is simply some datapoint that falls well outside the predicted/expected results. That guy who screws up the grading curve in class. It isn't an issue of "good" or "bad", simply being far outside predicted/expected range. They are a source of constant pain and frustration for statisticians because they can wreak havoc with correlations and studies due to their extremity. But you run great risk in ignoring outliers as the existence of outlying data is often a signal of another unidentified factor.
Just fwiw. "Outlier" does not imply any sort of value judgement. It is just an expression of a piece of data that doesn't "fit".