Comment by lionhearted | Hacker News Reader

Comment by lionhearted | Hacker News Reader

Parent Full thread

This author doesn't read his own studies or check his numbers. Here's the most glaring example:

> One of Lustig’s opening assertions is that The Atkins diet and the Japanese diet share one thing in common: the absence of fructose. This is flat-out false because it implies that the Japanese don’t eat fruit. On the contrary, bananas, grapefruits, Mandarin oranges, apples, grapes, watermelons, pears, persimmons, peaches, and strawberries are significant staples of the Japanese diet [17].

Except the Japanese DON'T eat much fruit compared to Americans. It's in his own citation:

Japan fruit consumption: 41.5 kg per person per year

USA fruit consumption: 126.5 kg per person per year

http://www.ers.usda.gov/publications/wrs0406/wrs0406h.pdf

http://www.usda.gov/factbook/chapter2.pdf

Americans eat over 300% more fruit than Japanese. I don't know what a "significant staple" is, but from my time in Tokyo, Osaka, the rest of Kansai, and Hokkaido, Japanese have a heck of a lot less fructose than Americans do. It's mostly rice, meat, fish, and vegetables, and the main desserts are only semi-sweet like mochi and red bean.

The most powerful argument Lustig makes is one he barely makes any fuss about in the video, but can't just be waved aside: "I use this information to craft diets that consistently cause obese children to lose weight."

If you dig into actual dietary outcomes, "a person consistently loses weight on this diet" is a milestone hardly any diet reaches. Certainly not diets based on the idea that obesity is caused by eating too much and so the solution is to eat less; those have a terrible track record.

The way to cut through the scientific fog is not to argue about mechanisms and causation; the way is to look at who makes correct predictions, and consequently in the case of diets, what works and what doesn't. Fructose may not be evil, it may merely be strongly correlated with something else that is evil, but that's good enough for me, for today.

I also didn't think the problem was fruit, but rather things like HFCS-sweetened sodas. A banana or an apple has about 100 calories, but a 12 ounce can of soda has about 150. I know people (I was one) that will drink 32 ounces (about a liter) of soda just at lunch but won't have more than 2 pieces of fruit a day.

There's also the fact that the apple or banana have fiber and other nutrients the sodas are missing. The sodas (or often other HFCS-rich foods) don't have.

Yeah, at one point Lestig points out that generally fruit is ok because the amount of fructose in fruit is relatively low and accompanied by lots of fiber and micronutrients. Fruit also tends to be more expensive so you can't eat as much even if you wanted to.

A mango is a sweet and not particularly fibrous fruit, and it's only about 15% sugar. According to nutritiondata.com, A pound of mango pulp has about 65 grams of sugar. That would probably cost about $7.00 prepared (you can get them sliced fresh at Whole Foods). According to thedailyplate, a 20oz bottle of Pepsi, which feels expensive at $1.25 as you can get liter bottles for only slightly more, has 69 grams of sugar.

And like I said, mango is not a particularly low-sugar fruit. Pineapple is 8.5% sugar and the bromelain makes your mouth hurt if you eat too much. Apples 10.5%.

Agreed. If you actually watch the video it's the HFCS that's the major source of fructose, NOT fruit.

Actually, a banana has about 200 calories, 23g sugars, and a fairly high glycemic index:

http://www.nutritiondata.com/facts/fruits-and-fruit-juices/1...

That's why a lot of the paleo diet folks avoid them.

That's 1 cup, mashed. I looked up 1 medium banana:

http://www.weightlossforall.com/calories-banana.htm

Edit: To note that I still do see your point about why certain diets avoid them. I think of bananas as the 3 Musketeers Bar of fruit. :-)

Sweetener control becomes a lot easier if you elect to add it yourself: For example, to keep my own cravings at a minimum I've moved towards a combination of fruits and individual sugar packets, or more often noncaloric sweeteners, for sweetening. The sugar packets are 5g each(as are sugar cubes), which means you would have to dump eight of them in your coffee to make it as sweet as a 12oz Coke(40g).

Nobody dumps in eight sugar packets into plain coffee - they buy a sweetened coffee beverage instead, which is about as sugary as the Coke and costs more to boot.

Candy bars and cookies are similarly difficult to portion control, because, again, they're overconcentrated.

Also, the author makes one of the most annoying forms of argument: "truth through averaging." "So, what’s the upper safe limit of fructose per day (all sources considered)? [Author than compares high and low estimates]... Figuring that both sides are biased, the middle figure between the two camps is roughly 50 grams for active adults."

Lessee -- scientists argue that the Earth formed ~6 billion years ago, creationists say ~6000 years ago. Figuring both sides are biased, the middle figure between the two camps is roughly 3 billion years ago. So the Earth is 3 billion years old. QED.

It completely depends on what probability distribution you assume for the range. If it's a normal distribution with a high confidence interval from 25g to 90g, taking the average is reasonable. For the age of the Earth example, clearly the distribution is a bit more binary.

Maybe we're getting "hung up on the trivial minutia of an exact gram amount", though.

I think the word you want is "bimodal".

I guess you're technically correct, though the two peaks are almost narrow enough to be binary. :-)

Although I don't completely agree with the author, he accuses Lustig of a similar fallacy: using the popularity of his Youtube video to justify where the truth stands.

See his retrospective of the original article: http://www.alanaragonblog.com/2010/02/19/a-retrospective-of-...

It has a decent (although one-sided) summarization of the comments from the original.

He also says that:

"High-fructose corn syrup (HFCS) is nearly identical to sucrose in structure and function."

But we had this article recently on HN that shows the exact opposite:

http://www.princeton.edu/main/news/archive/S26/91/22K07/

To be fair, that study has a few procedural gaps that make me want to wait for confirmation and expanded testing.

Even though I'm predisposed to agree with it's conclusions --or perhaps because I'm so predisposed to agree with it's conclusions-- I'd prefer to wait for better science before calling the race.

The resulting chemical structure, after digestion, is the same, but with different amounts of each chemical. Sucrose is a 50/50 mix of fructose to glucose. HFCS has a customized ratio, 55/45, 42/58, or 90/10.

A sucrose molecule is just a glucose and a fructose chemically bound as a compound (so it necessarily has a 50/50 composition). But the fructose is HFCS is free, not bound to glucose. This is an interesting hypothesis for why HFCS might have such an effect compared to sucrose.

From the article:

"Second, as a result of the manufacturing process for high-fructose corn syrup, the fructose molecules in the sweetener are free and unbound, ready for absorption and utilization"

Interestingly, there's one difference between sucrose and a 50:50 glucose/fructose mix: the sucrose will have less sweetness per calorie, because it's only one molecule (a disaccharide) and can thus stimulate only half as many sweet receptors.

Given that the body attempts to measure current calorie intake by calibrating past calorie consumption versus past perceived sweetness, this means that the glucose/fructose mix will prime the body to crave sweets more intensely. This is pretty much the only biologically plausible argument against HFCS.

Compare: the studies showing that diet sodas actually lead to an increased risk of diabetes (doi:10.2337/dc08-1799).

Compare: the various diets encouraging the daily consumption of nearly-flavorless olive oil (example implementation, unsuccessful: http://lesswrong.com/lw/a6/the_mysteries_of_shangrila_dietin...).

I think he must not have been aware that article... pretty embarrassing for him after he does that long, snide blog post only to have a new study blow it out of the water.

The princeton study is very poor science. Here's a good summary of why: http://www.foodpolitics.com/2010/03/hfcs-makes-rats-fat/

And here's a longer article about why at Ars: http://arstechnica.com/science/news/2010/03/does-high-fructo...

He also left out the fact about the fibre, Lustig never said fruit was bad, but rather the fructose <b>without</b> the rest of the fruit

I think the point was that the Atkins diet allows for far less fruit consumption than the Japanese have on average, so grouping the two diets together as "effectively fructose-free" is incorrect.

Reply on news.ycombinator.com