Eat food. Not too much. Mostly plants. (2007) • Hacker News Board

Re: Eat food. Not too much. Mostly plants. (2007)

#11

Post by novia » Thu, Jan 26, 2017, 4:22 AM UTC

wyager wrote: post #8

pierrec wrote: post #6

Earlier quoted context omitted.

Sure, nutrition is littered with crackpot theories, voluntarily misleading research funded by corporations, smoke and mirrors, and so on. And if you only look at stuff that hits the news, then yes, things will change drastically every few years, probably forever. That's because people like sensationalism. However, there are reasonable diets that don't make the news, for example the "Mediterranean Diet". While it does…

> for example the "Mediterranean Diet". The irony here is delicious. You've gotten sucked into yet another evidence-free fad diet, but you're lecturing me on only listening to fad diets. And no, it's not a matter of "only looking at stuff that hits the news"; the FDA and FNS, which should be following strong scientific principles, change their dietary recommendations and rules all the time, without particularly stron…

The Mediterranean diet has been linked to longer lifespans.

http://www.bmj.com/content/349/bmj.g6674

https://www.ncbi.nlm.nih.gov/pubmed/15452459

Re: Eat food. Not too much. Mostly plants. (2007)

#12

Post by wyager » Thu, Jan 26, 2017, 4:29 AM UTC

novia wrote: post #11

wyager wrote: post #8

Earlier quoted context omitted.

> for example the "Mediterranean Diet". The irony here is delicious. You've gotten sucked into yet another evidence-free fad diet, but you're lecturing me on only listening to fad diets. And no, it's not a matter of "only looking at stuff that hits the news"; the FDA and FNS, which should be following strong scientific principles, change their dietary recommendations and rules all the time, without particularly stron…

The Mediterranean diet has been linked to longer lifespans. http://www.bmj.com/content/349/bmj.g6674 https://www.ncbi.nlm.nih.gov/pubmed/15452459

> The Mediterranean diet has been linked to longer lifespans

Did you read anything I wrote? Every single fad diet in the world has multiple scientific papers showing how great it is. (p = 0.05. Nevermind that there were 1000 research projects about this fad diet that didn't get published because they didn't reject the null hypothesis, so who cares. Nevermind that nutrition scientists don't practice good behavior to mitigate the effect of the base rate fallacy on popular subjects.)

Almost no scientific fields practice enough discipline to have very high predictive power. Particle physics mitigates base-rate failures by having an extremely high confidence requirement (5 sigma). Even with absolute best practices, it's sometimes not enough. http://slatestarcodex.com/2014/04/28/the-control-group-is-ou...

Nutrition science is nowhere near best practices.

Re: Eat food. Not too much. Mostly plants. (2007)

#13

Post by ericdykstra » Thu, Jan 26, 2017, 4:29 AM UTC

This is a great article, and should be widely read beyond that piece of succinct eating advice. For example, this is a great paragraph:

> In many cases, long familiarity between foods and their eaters leads to elaborate systems of communications up and down the food chain, so that a creature’s senses come to recognize foods as suitable by taste and smell and color, and our bodies learn what to do with these foods after they pass the test of the senses, producing in anticipation the chemicals necessary to break them down. Health depends on knowing how to read these biological signals: this smells spoiled; this looks ripe; that’s one good-looking cow. This is easier to do when a creature has long experience of a food, and much harder when a food has been designed expressly to deceive its senses — with artificial flavors, say, or synthetic sweeteners.

In fact, I recently read a book, The Dorito Effect which goes into this in much, much more detail, and I highly recommend it to anyone interested in the subject. I bought it thinking it would be a light read, based on the title and cover, but it was more engaging, better researched, and more informative than I expected.

Re: Eat food. Not too much. Mostly plants. (2007)

#14

Post by thanks4it » Thu, Jan 26, 2017, 4:35 AM UTC

wyager wrote: post #10

cup wrote: post #5

Earlier quoted context omitted.

Medicine, as a field, is mostly quackery and bullshit. There's a reason the advice changes drastically every few years, and at any time there are 10 competing treatments, each apparently supported by some number of reputable scientific authorities. We do not have the predictive power at this time to make authoritative recommendations about how each and every person should be treated, except for fairly straightforward…

> To shit all over nutritionists because you don't understand the science What makes you think A) that's my motivation or B) I don't understand the science? I'm not a nutritionist, but I'm well-versed in statistics and hypothesis testing, so I am at least equipped to make general observations about the predictive power of the field. > or don't think the science is moving fast enough When did I ever say anything to th…

Statistics is full of p hacking bullshit. You can mislead and lie with stats and stull sound intelligent. You being a case in point. It's funny you mention psychology, they don't accept p-value assholes in their journals any longer. You have a hammer and see nails. Foolish little fuck you are.

Re: Eat food. Not too much. Mostly plants. (2007)

#15

Post by pierrec » Thu, Jan 26, 2017, 4:36 AM UTC

wyager wrote: post #8

pierrec wrote: post #6

Earlier quoted context omitted.

Sure, nutrition is littered with crackpot theories, voluntarily misleading research funded by corporations, smoke and mirrors, and so on. And if you only look at stuff that hits the news, then yes, things will change drastically every few years, probably forever. That's because people like sensationalism. However, there are reasonable diets that don't make the news, for example the "Mediterranean Diet". While it does…

> for example the "Mediterranean Diet". The irony here is delicious. You've gotten sucked into yet another evidence-free fad diet, but you're lecturing me on only listening to fad diets. And no, it's not a matter of "only looking at stuff that hits the news"; the FDA and FNS, which should be following strong scientific principles, change their dietary recommendations and rules all the time, without particularly stron…

An evidence-free fad diet? We're looking at dozens (maybe hundreds) of research papers published over the past 40 years. Most of them are not based on conjectures but on clinical and dietary surveys looking at long-term health effects. Perhaps you think this is irrelevant because of selection bias, however, if a study found the diet had any relatively negative health effects, I'm pretty sure it would also get published. Of course, this can always turn into an unsolvable "faith in research" argument.

"Boring" diets, yes, because these long-term studies are necessary in order to convincingly support the idea that a diet is healthy, which necessarily implies that the diet is pretty old and probably won't hit the news all that much.

Finally, and as an edit, don't worry: I haven't gotten sucked into the diet that I'm using as an example. I don't follow or endorse it, in part because it relies too much on unsustainable fisheries.

Re: Eat food. Not too much. Mostly plants. (2007)

#16

Post by gibbitz » Thu, Jan 26, 2017, 4:39 AM UTC

wyager wrote: post #2

Nutrition, as a field, is mostly quackery and bullshit. There's a reason the advice changes drastically every few years, and at any time there are 10 competing fad diets, each apparently supported by some number of reputable scientific authorities. We do not have the predictive power at this time to make authoritative recommendations about what each and every person should eat, except for fairly straightforward thing…

I'm 42 and this is what the USDA has been saying all along with their food pyramids and four food groups and what have you. I think as a whole, nutritionists have always been on the same page, but those selling diets keep changing their opinions. Sure, there has been flip-flopping recently on fats and sugars, but the edict "Eat food. Not too much. Mostly plants" doesn't concern itself with those. I can't recall anyone ever promoting a diet with less vegetables (even the Atkins fad diet known for it's meat and dairy included vegetables) and make sure to overeat at every meal and I highly doubt either of those will ever be considered a good idea (provided the vegetables aren't tainted by radiation or chemicals or the diet's authors aren't overly politicizing GMO or some other arbitrary thing about production). Fad diets are all about selling supplements and books and are indeed snake oil. The problem is separating the wheat from the chaff and that's what this article is really about.

Re: Eat food. Not too much. Mostly plants. (2007)

#17

Post by wyager » Thu, Jan 26, 2017, 4:45 AM UTC

gibbitz wrote: post #16

wyager wrote: post #2

Nutrition, as a field, is mostly quackery and bullshit. There's a reason the advice changes drastically every few years, and at any time there are 10 competing fad diets, each apparently supported by some number of reputable scientific authorities. We do not have the predictive power at this time to make authoritative recommendations about what each and every person should eat, except for fairly straightforward thing…

I'm 42 and this is what the USDA has been saying all along with their food pyramids and four food groups and what have you. I think as a whole, nutritionists have always been on the same page, but those selling diets keep changing their opinions. Sure, there has been flip-flopping recently on fats and sugars, but the edict "Eat food. Not too much. Mostly plants" doesn't concern itself with those. I can't recall anyon…

> I'm 42 and this is what the USDA has been saying all along with their food pyramids and four food groups and what have you.

You are not remembering correctly.

https://en.wikipedia.org/wiki/History_of_USDA_nutrition_guid...

> I think as a whole, nutritionists have always been on the same page,

No, they certainly haven't. "Best practices" change all the time. This happened quite recently; http://articles.mercola.com/sites/articles/archive/2015/06/1...

> but the edict "Eat food. Not too much. Mostly plants" doesn't concern itself with those

It also contradicts the recently popularized "keto" diet, which has roughly as much evidence behind it as any other popular dietary recommendation (i.e. not much).

Really the only things we can say authoritatively is "if you miss out on certain nutrients, you get sick". Beyond that, we're not really very sure.

> The problem is separating the wheat from the chaff

Your mistake is in thinking that there's a definitive difference. Almost ever piece of nutritional research available today has an extremely low predictive power. Where you draw the wheat/chaff line is more or less arbitrary. Even what should, ideally, be the most staunchly scientific food-related organizations, the FDA and FNS, change the recommendations all the time because they're not actually using hard science.

Re: Eat food. Not too much. Mostly plants. (2007)

#18

Post by novia » Thu, Jan 26, 2017, 4:48 AM UTC

wyager wrote: post #12

novia wrote: post #11

Earlier quoted context omitted.

The Mediterranean diet has been linked to longer lifespans. http://www.bmj.com/content/349/bmj.g6674 https://www.ncbi.nlm.nih.gov/pubmed/15452459

> The Mediterranean diet has been linked to longer lifespans Did you read anything I wrote? Every single fad diet in the world has multiple scientific papers showing how great it is. (p = 0.05. Nevermind that there were 1000 research projects about this fad diet that didn't get published because they didn't reject the null hypothesis, so who cares. Nevermind that nutrition scientists don't practice good behavior to m…

It's been a while since I've read Michael Pollan's book, An Omnivore's Dilemma, but if I am remembering correctly, there are several locations in the world where people live an abnormally long time, with good quality of life. Researchers went to those places and found out what people usually eat there. Then, comparing the diets of the different qualifying populations, the diet advice Pollan came up with to live a long healthy life was, "eat food, not too much, mostly vegetables."

Re: Eat food. Not too much. Mostly plants. (2007)

#19

Post by closeparen » Thu, Jan 26, 2017, 4:49 AM UTC

gibbitz wrote: post #16

wyager wrote: post #2

Nutrition, as a field, is mostly quackery and bullshit. There's a reason the advice changes drastically every few years, and at any time there are 10 competing fad diets, each apparently supported by some number of reputable scientific authorities. We do not have the predictive power at this time to make authoritative recommendations about what each and every person should eat, except for fairly straightforward thing…

I'm 42 and this is what the USDA has been saying all along with their food pyramids and four food groups and what have you. I think as a whole, nutritionists have always been on the same page, but those selling diets keep changing their opinions. Sure, there has been flip-flopping recently on fats and sugars, but the edict "Eat food. Not too much. Mostly plants" doesn't concern itself with those. I can't recall anyon…

I'm half your age and the USDA recommended grains as the plurality of your diet until nutrition stopped showing up in school.

Re: Eat food. Not too much. Mostly plants. (2007)

#20

Post by wyager » Thu, Jan 26, 2017, 4:50 AM UTC

pierrec wrote: post #15

wyager wrote: post #8

Earlier quoted context omitted.

> for example the "Mediterranean Diet". The irony here is delicious. You've gotten sucked into yet another evidence-free fad diet, but you're lecturing me on only listening to fad diets. And no, it's not a matter of "only looking at stuff that hits the news"; the FDA and FNS, which should be following strong scientific principles, change their dietary recommendations and rules all the time, without particularly stron…

An evidence-free fad diet? We're looking at dozens (maybe hundreds) of research papers published over the past 40 years. Most of them are not based on conjectures but on clinical and dietary surveys looking at long-term health effects. Perhaps you think this is irrelevant because of selection bias, however, if a study found the diet had any relatively negative health effects, I'm pretty sure it would also get publish…

> We're looking at dozens (maybe hundreds) of research papers published over the past 40 years.

And how many nutrition researchers and students do you think there are? How many do you think choose to do a study about a popular diet? Do you think everyone who fails to reject the null hypothesis with whatever mediocre p-value criteria their supervising professor decided on publishes their paper anyway? Do you think most journals would accept such a paper?

> however, if a study found the diet had any relatively negative health effects

That's the thing. Almost any semi-reasonable diet isn't going to kill you or make you horribly sick. Every dietary recommendation you find online will probably keep you alive and healthy. Then, by the nature of having lots of scientific papers and the way publishing works, you can find as many papers as you like claiming that your favorite dietary recommendation is slightly better than the other ones.