I consider it irrational for someone to believe that a limited number of hand-selected studies can give you absolute confidence.
Even if it were a limited number of studies, gwern's not going for "absolute confidence". (Indeed, requiring "absolute confidence" before attempting a thing is irrational by the following sentence.) Irrational here means doing something with expected value less than zero. If you're a person living in an industrialized society with ready access to electrical lighting -- the possible gains are too large to be balanced out by the time wasted trying it. That's what gwern's saying.
A positive but cautious conclusion like http://www.vanderbilt.edu/AnS/psychology/health_psychology/m... would have been much more appropriate.
The black swan is a pretty useless concept at this point, because people like you will drag it out against anything they dislike. You cannot give any rigorous formulation or evidence for it, and the best you can do is drag out some old example like thalidomide.
Yes it's hard and often impossible to quantify black swans. But keeping them in mind is a way to avoid overconfident articles like the one you wrote.
Alright, let me clarify it then: by the remotely meaningful version presented by Taleb, your argument doesn't work and Taleb's writings support taking melatonin, rather than not taking melatonin. Why? Because even in the event of a negative 'black swan' your losses are strictly limited: the worse that can happen is you die. In fact, if I may borrow Taleb's 2x2 table from _The Black Swan_ where he brings up positive 'black swans' as well, it's easier to make a case that taking melatonin is a positive 'black swan': the negative possibilities are strictly in Mediocristan, with limited losses, easy to measure over large populations, predictable, not varying over time; while the positive possibilities look like some of Taleb's examples of attempting to exploit positive 'black swans' in Absurdistan - improved productivity, healthy, and seizing opportunities, things with potentially unlimited payoffs - such as moving to cities and networking with people.
> For example, a greater risk of dying after 20 or 30 years of use - which has not yet been studied.
FUD.
You do not mention any studies of long term effects and you consider FUD raising the possibility that there could be adverse effects. When more and more studies show negative long term effect of substances that were considered innocuous (like vitamins), it's not FUD, it's being reasonably cautious.
Not at all. Taleb like to use the example of selling insurance and shorts, where your downside is unlimited. Millions, billions, trillions... Think AIG or LTCM. Not everything has consequences solely for health.
> You do not mention any studies of long term effects and you consider FUD raising the possibility that there could be adverse effects.
In the absence of any evidence whatsoever for that, and plenty of positive correlations, short-term studies without any problem... Yes. It is FUD. Fear, uncertainty, and doubt.
> When more and more studies show negative long term effect of substances that were considered innocuous (like vitamins), it's not FUD, it's being reasonably cautious.
Only some vitamins show any mortality effects, the mortality effects while real are pretty small, the vitamins did not deliver the long-term health benefits promised (but never really demonstrated in experiments) which might offset the harm, and the vitamins were being consumed in huge doses. The first reduces the possibility of any such outcome, the second points out that you are exaggerating the harm, the third means that vitamins were a purely speculative play unlike melatonin, and the fourth reduces the concern that any such thing would happen with melatonin when one sticks to the suggested doses like 0.5mg rather than 5mg.
I don't argue that this doesn't help a fraction of people, but the case is pretty weak and the long-term results of tampering with a critical hormone are unknown.
No offense to the author but I've read at least 3 of his articles posted here and they all follow a similar pattern.
That kind of numbers is a red flag to me. It indicates that these studies may have issues, which I suspect they do.
[1] "Dietary supplement industry says 'no' to more information for consumers (again)" 3 October 2013
http://www.sciencebasedmedicine.org/big-supp-resists-giving-...