Quoting Russell Coker (russell@coker.com.au):
One thing that's always annoying is when people make wild claims offering no evidence at all and then criticise people who reference Wikipedia.
There are much better and saner criticisms one can make of Wikipedia. One is that many pages are crazy-quilt junkyards of maniacal detail-freakery. See xkcd parodies thereof: https://xkcd.com/214/ https://xkcd.com/739/ https://xkcd.com/444/ https://xkcd.com/446/ Another is that some of the articles are pretty much incoherent and utterly fail to cover their subjects -- possibly as a result of camel-like committee editing. (Old saying: A camel is a horse that was designed by a committee.) Example: http://en.wikipedia.org/wiki/Prosecutor's_fallacy I happen to be a mathematics guy, and a Bayesian, so the really awful explanation at that page really pained me. Here's what I wrote about it on a mailing list: Quoting Terry W. Colvin (fortean1@mindspring.com):
Here's some examples of that sort of thing, generally called the "Prosecutor's Fallacy" http://en.wikipedia.org/wiki/Prosecutor's_fallacy#Conditional_probability
By the way, intending no criticism whatsoever of you or Ray, Terry, but the explanation of that concept at Wikipedia is truly dreadful. I expect most people reading it will think 'Well, it's complex math stuff; I can't expect to understand it.' That's a pity, because it's not that difficult to explain -- I think. I'll have a try at it (borrowing from http://www.dcs.qmul.ac.uk/researchgp/spotlight/legal.html): If you know it's the midnight hour, you can determine the chance that it's dark outside (i.e., pretty certain). If you know it's dark, you can also calculate the likelihood that it's the midnight hour (non-zero but small). The point is that the odds of A given that you know B is NOT THE SAME as the odds of B given that you know A. Take that scenario to court: The prosecutor speaks as if he/she is calculating the odds that the defendent is innocent given the DNA evidence presented (B given knowledge of A). _However_, what he or she actually _presents_ are statistical calculations about how likely the DNA evidence would be present, given that the defendent is innocent (A given knowledge of B). And, the point is, these are just not the same thing at all, and calculating one 'conditional probability' tells you absolutely nothing about the other. In the UK court case mentioned on http://www.dcs.qmul.ac.uk/researchgp/spotlight/legal.html, prosecutors sought conviction of defendent mother Sally Clark (whose two babies had died) by having an expert testify that the probability of two cot deaths occurring in a single family was 1 in 73 million. To restate, the prosecutors presented the odds of two cot deaths in a single family, given that the mother is innocent of murdering them (A given knowledge of B). Unfortunately, that is not a relevant calculation: They needed to calculate the calculate the ENTIRELY DIFFERENT AND UNRELATED probability of the mother being innocent of murdering her children, given occurrence of two cot deaths in the family -- B given knowledge of A. (That actually ended up being an extremely unlikely 1 in 2 billion chance, completely the opposite of the picture that prosecutors had painted. Ms. Clark was unjustly convicted based on this entirely bogus probablistic argument. Her convictions were overturned after two appeals, but then quite understandably she then drank herself to death.) Yeah, and now I can anticipate the question: So, Rick, if you think Wikipedia's explanation of the Prosecutor's Fallacy is so wretched and opaque, and you are sure you can do better, why aren't you editing the page? I might. If anyone else wishes to do so, I give my blessing to anyone who wants to borrow any or all of the above.