3 ms·

Melatonin is a clear-cut Good Thing. The gains I have laid out are large enough I consider it irrational for someone not to use it. I consider it irrational fo

by pesenti 13y ago

Melatonin is a clear-cut Good Thing. The gains I have laid out are large enough I consider it irrational for someone not to use it. I consider it irrational for someone to believe that a limited number of hand-selected studies can give you absolute confidence.

thebooktocome 13y ago
First, it's not a "limited number of hand-selected studies." The Brzezinski meta-study ( http://www.gwern.net/docs/melatonin/2005-brzezinski.pdf http://www.gwern.net/docs/melatonin/2005-brzezinski.pdf ) alone covered 17 studies and 284 participants. There's even a clear, well-documented causal mechanism, which can't be said for most of the supplement industry. 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.
- pesenti 13y ago
  And how do you claim to calculate expected value here? You are making the same error he is by ignoring the possible "black swan", which, with every drug, can have dramatic consequences. A positive but cautious conclusion like http://www.vanderbilt.edu/AnS/psychology/health_psychology/melatonin.htm#Conclusions%20on%20Melatonin http://www.vanderbilt.edu/AnS/psychology/health_psychology/m... would have been much more appropriate.
  - julioademar 13y ago
    I recommend you read the article first. Check out the ROI section: http://www.gwern.net/Melatonin#roi http://www.gwern.net/Melatonin#roi
  - gwern 13y ago
    > You are making the same error he is by ignoring the possible "black swan", which, with every drug, can have dramatic consequences. 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.
    - pesenti 13y ago
      No rigorous formulation? The black swan has a very simple mathematical model: an unlikely event with an effect so great that it will affect the expected value. For example, a greater risk of dying after 20 or 30 years of use - which has not yet been studied. In your calculation of ROI you completely ignore unknown long term effects. 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.
      - gwern 13y ago
        > The black swan has a very simple mathematical model: an unlikely event with an effect so great that it will affect the expected value. For example, a greater risk of dying after 20 or 30 years of use - which has not yet been studied. In your calculation of ROI you completely ignore unknown long term effects. 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.
- domdip 13y ago
  Using a meta-study that showed tiny changes (statistical significance is irrelevant here, virtually any intervention could have had that kind of impact) to bolster a handful of anecdotes is pretty silly. 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.
- harvestmoon 13y ago
  17 studies, 284 participants?? And that's supposed to be reassuring? That kind of numbers is a red flag to me. It indicates that these studies may have issues, which I suspect they do.
tokenadult 13y ago
In general, the pattern of regulation of "dietary supplements"[1] in United States law allows a lot of claims that go beyond the available evidence to be made without legal consequences. In an environment like this, I would do more than just listen to anecdotes from satisfied users. [1] "Dietary supplement industry says 'no' to more information for consumers (again)" 3 October 2013 http://www.sciencebasedmedicine.org/big-supp-resists-giving-consumers-safety-and-effectiveness-information/ http://www.sciencebasedmedicine.org/big-supp-resists-giving-...
wissler 13y ago
To the extent that it may permanently inhibit natural melatonin production it is not a good thing. There are tradeoffs.
- __--__ 13y ago
  Do you have any references for that? I'm not pro or con either way (yet), I'm just gathering research. I'd love to know if there are any recorded cases of this and the doses involved.

by pesenti 13y ago

thebooktocome 13y ago

First, it's not a "limited number of hand-selected studies." The Brzezinski meta-study ( http://www.gwern.net/docs/melatonin/2005-brzezinski.pdf http://www.gwern.net/docs/melatonin/2005-brzezinski.pdf ) alone covered 17 studies and 284 participants. There's even a clear, well-documented causal mechanism, which can't be said for most of the supplement industry. 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.

pesenti 13y ago
And how do you claim to calculate expected value here? You are making the same error he is by ignoring the possible "black swan", which, with every drug, can have dramatic consequences. A positive but cautious conclusion like http://www.vanderbilt.edu/AnS/psychology/health_psychology/melatonin.htm#Conclusions%20on%20Melatonin http://www.vanderbilt.edu/AnS/psychology/health_psychology/m... would have been much more appropriate.
- julioademar 13y ago
  I recommend you read the article first. Check out the ROI section: http://www.gwern.net/Melatonin#roi http://www.gwern.net/Melatonin#roi
- gwern 13y ago
  > You are making the same error he is by ignoring the possible "black swan", which, with every drug, can have dramatic consequences. 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.
  - pesenti 13y ago
    No rigorous formulation? The black swan has a very simple mathematical model: an unlikely event with an effect so great that it will affect the expected value. For example, a greater risk of dying after 20 or 30 years of use - which has not yet been studied. In your calculation of ROI you completely ignore unknown long term effects. 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.
    - gwern 13y ago
      > The black swan has a very simple mathematical model: an unlikely event with an effect so great that it will affect the expected value. For example, a greater risk of dying after 20 or 30 years of use - which has not yet been studied. In your calculation of ROI you completely ignore unknown long term effects. 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.
domdip 13y ago
Using a meta-study that showed tiny changes (statistical significance is irrelevant here, virtually any intervention could have had that kind of impact) to bolster a handful of anecdotes is pretty silly. 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.
harvestmoon 13y ago
17 studies, 284 participants?? And that's supposed to be reassuring? That kind of numbers is a red flag to me. It indicates that these studies may have issues, which I suspect they do.

tokenadult 13y ago

In general, the pattern of regulation of "dietary supplements"[1] in United States law allows a lot of claims that go beyond the available evidence to be made without legal consequences. In an environment like this, I would do more than just listen to anecdotes from satisfied users. [1] "Dietary supplement industry says 'no' to more information for consumers (again)" 3 October 2013 http://www.sciencebasedmedicine.org/big-supp-resists-giving-consumers-safety-and-effectiveness-information/ http://www.sciencebasedmedicine.org/big-supp-resists-giving-...

wissler 13y ago

To the extent that it may permanently inhibit natural melatonin production it is not a good thing. There are tradeoffs.

__--__ 13y ago
Do you have any references for that? I'm not pro or con either way (yet), I'm just gathering research. I'd love to know if there are any recorded cases of this and the doses involved.