How arrogant does one have to be to discount a valid explanation because it doesn't fit their own narrative?
And how transparently evasive to sidestep the on-topic question and then latch onto an irrelevant bit of side discussion -- especially after admonishing Loss Leader for what Michel deemed a change of subject (but isn't). For the record, I was able to find an old book that referred to "loss leader" or "lost leader" in the context of reel-to-reel audio tape. While perhaps obscure, I'm satisfied it's not just made up or misused. By the way, the UNIVAC tape drive in the black-and-white picture in Loss Leader's post is rather iconic. There's a great example of one in the computer history museum in Sunnyvale, CA.
Getting back to the actual topic, Michel tried to handwave his way around criticism by vague references to statistical tools used in experimental sciences.
There are some well known statistical tools to deal with these issues (binomial distribution, calculation of p-value and so on).
Yes, such tools exist. And they are very well-known. So well known, in fact, that many people in many scientific and technical fields are required to be competent in them, such that Michel is likely to have no trouble finding people to review his methodology who are confident in them, if not outright experts. So well known that it would be hard to bluff your way through an application of them without someone catching you. And loss leader called his bluff:
Assume a sample size of two people .... Assume a random chance rate of 25% (which is the chance in a 1-in-4 test). How many trials would you have to run to see a hit rate of 50% with a 95% degree of confidence?
That's a pretty simple, straightforward question one would expect to find on a freshman statistics exam. And since it responds directly to what Michel was trying to invoke as a vague means of deflecting criticism, it can hardly be dismissed as off-topic. "There's a problem with the numbers in your experiment." "No, no, there are statistical methods to get around that." A reasonable next step in such a conversation would be, "Would you please apply those methods to your experiment for us?"
Further, Michel tries to backpedal away from a test of his competence with this:
Besides, a hit rate equal to 50% is not something I would expect in general, and I would not want to construct a confidence interval around this hit rate.
But these are not Loss Leader's numbers. They're Michel's.
If your hit rate is for example 50% (instead of the 25% expected from random chance alone), and your sample is large enough, then one generally says that an ESP effect has been found.
All Loss Leader has asked Michel to do is solve his own example problem. And if he didn't like the 50% hit rate, he could have chosen literally any other number, worked the problem, and demonstrated his case.
So no, it's not off-topic to ask Michel to demonstrate the statistics that he says underlies his method and ensures its reliability. He probably has a different reason for not doing it. It could be that he's entirely incompetent and doesn't know how to do it. Or he might realize that solving the problem for his own example will expose to a wider number of people just how flawed his method is. He keeps referring to ganzfield-style experiments where numerous trials are conducted in order to accumulate enough data for the tools he mentions to have any toehold. He ignores critics who point out that his experiment design has nothing to with the ganzfield methods.
But then Michel has to distract from his unwillingness to demonstrate his own competence and/or test the validity of his method. So
he launches a challenge to an off-topic, inconsequential point: the origin of Loss Leader's nickname. It has the effect both of changing the subject and of pointing out an error on the part of one of his critics. Evasion and well-poisoning seem to be the orders of the day.
