AlanCrowe comments on Quantified Health Prize Deadline Extended - Less Wrong

3 Post author: alyssavance 05 January 2012 09:28AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (45)

You are viewing a single comment's thread. Show more comments above.

Comment author: RichardKennaway 05 January 2012 05:13:01PM 8 points [-]

How do you suppose that 'eat like a sane person' (as though that were precise advice) gives exactly optimal nutrition, with no gains to be had from any increase or decrease of anything? It seems vanishingly improbable that there is no substance that a human could benefit from getting slightly more or less of in the diet.

It does not seem vanishingly improbable to me. Any organism that must eat complex foodstuffs (which includes all animals big enough to see) must deal with the fact that they have no way to obtain precisely the right quantity and proportions of everything that they need. Their bodies therefore need to be robust to wide variations in their dietary content, resulting in a plateau, possibly a very wide one, between deficits and excesses that do measurable harm. If this is so, then there is no such thing as the exactly optimal amount of a nutrient. Instead, there is a broad range, and if you manage to hit that barn door it doesn't matter where.

Comment author: AlanCrowe 05 January 2012 06:37:06PM 3 points [-]

I think that what you say has to be true at the population level. (The panda provides an obvious counter example, but since the panda is going extinct, that is merely a nit pick. Successful species, such as rats or humans are robust to wide variations in their dietary content.

However, at the micro-level, the truth of this proposition is maintained by ruthless culling. The rat population is riddled with weaklings, who lose out in life due to diet-induced health problems.

As for the human population, we have a different perspective on these matters. I can start from a prior belief that my body is probably robust to wide variations in dietary content, but I need non-zero probabilities on a wide variety of dietary vulnerabilities so that updating will work if events supply evidence.