A quick thought on the problem of confirmation bias.
Skeptics know that many conspiracy theories, and all general works of pseudo scholarship, are based on some fundamental principles. As follows:
Most pseudo-scholarly conspiracy theories are based upon one or both of two premises (1) That in their own particular field of enquiry there is “no such thing as coincidence” and (2) That the fact of missing data is in fact important data itself because missing data/evidence is proof that what is missing is missing because it’s incriminating.
To that, we must then add the problem of pattern building confirmation bias. One of the neatest little, and interesting, papers you can read on that subject is by Will Dowd and it's available free on Caltech’s eSkeptic magazine. Here is the direct link to that paper: http://www.skeptic.com/eskeptic/13-06-19/#feature
I have only one question about confirmation bias and it is this: ‘How can we guard against the problem of confirmation bias when it comes to assessing the likelihood that something unlikely actually did occur?’ I mean, if I know about the problems of confirmation bias leading other people to join the dots while looking for confirming evidence for their pet hypothesis, whilst paying scant regard to disconfirming evidence, how do I know I’m not doing the same when I claim that their account is shaped by their confirmation bias?’
In sum: since we have no algorithm to help us assess confirmation bias, those who spot it may also be victims of it, because their own bias towards spotting confirmation bias prevents them from objectively assessing that the improbable might just be real.