Tedd Koren's behaviour towards Stephen Barrett falls a long way short of what I would expect from a professionally qualified adult, including repeating allegations without bothering to check if they are true or not and the inflammatory press release quoted above (I did a bit of checking to see what was behind it. The disposition in particular was pretty long-winded but somewhat illuminating), and I thank you for bringing him to my attention so I know of someone else to avoid. In some quarters there does seem to be a habit of trying to smear the person rather than challenge their ideas. I'm reminded of the time Gillian McKeith responded to a professor of nutrition who wanted to debate her work by threatening to sue him for libel.

However Koren's activities are not relevant to the current discussion which is Batmanghelidj's "water cure". My observations above have nothing to do with Quackwatch and are based on reading Batmanghelidj's own publications. I came to a different conclusion to him, but I'm not on death row and facing the gallows if I come up with something that doesn't work. The "cure" is a mixture of placebo and regression to the mean (in other words it would have got better even without the water). The tricky bit is proving that something has an effect over and above placebo.

Another interesting phenomenon is the opposite to placebo and which is called the nocebo effect, where something has negative effects such as lack of efficacy because you've convinced yourself that it won't work. You might do this knowingly (such as expecting an injection to hurt more than it does) or subconsciously (such as someone who might take the writings of a particular person seriously without considering other points of view that don't agree with theirs, and who then applies that thinking to every situation without even being aware that they're doing it). This is why questioning everything is a good idea, especially if it's something you agree with or if they have strong opinions one way or another. Scientists are always arguing with each other. There's even a formal way to do it: the peer review.

Nocebo and placebo explain a lot of anecdotal evidence. A proper trial takes steps to exclude them. One way is trying things out on more than one person to see if the results are reproducible. Another is double blinding, which is when neither the researcher nor the subject know who is in the experimental group and who is the control until after the study has concluded. Individual anecdotes are usually not regarded very highly because they can be biased by the person's beliefs and it can be difficult to show that something happening is directly because of one particular thing that they have done. Most conditions improve by themselves. Proving that a condition improved because you've done something is very difficult. One data point, or person, doesn't mean a lot. Having lots of data points does.