It's interesting how some diet advice sticks around longer than others. Your older contemporary Jean Brillat-Savarin wrote in Physiologie du goût (1825) that a primary cause of obesity, aside from variance in people's constitutions and lack of exercise, was eating way too much starchy food. He sounds like he's advocating the Atkins diet sometimes ("Avoid everything starchy in whatever form it appears. You still have the roasts, salads, and herbaceous vegetables."). He also notes the problem of eating when we aren't actually hungry.
Sugar and other carbs drive food cravings in "most" people.
Those cravings result in over eating. Which is why you gain weight.
That's why low carb. And keto work for so many people. They reduce cravings which means you eat less junk food which means less total calories which means you lose weight.
Also please note the word "most" studies have shown that people are different. There is no one perfect diet for everyone
Especially for things like exercise and physical health, which are multi-factorial and massively confounded with things like your existing healthiness, diet, stress, and genetics, even RCT's are noisy, and the endpoints we care most about are far off and rare events (like all cause mortality, or morbidity), RCT's generally can't run long enough to show statistically valid effects, and would be too expensive to do, and wouldn't ever actually control for everything that was impacting the endpoints.
But when you think about topics like diet and health and weight, the endpoints are multiply confounded in ways that RCT's don't really address.
Let's consider KD Halls' (excellent) study on ultra processed foods - Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake (2019):
He actually took people into the study center for 4 weeks so he could completely control their diets to compare 2 weeks of eating UPF versus two weeks of eating "real" food - both admirable and expensive.
But the health endpoints measured, and things like weight, are a factor of LOTS of different things. Diet, naturally, and that's what we're testing - but also activity levels, exercise, smoking, stress from jobs, commutes, and family and relationship, and more.
So you've suddenly brought people into your study center for 4 weeks. Now you're controlling their diet, but all the stresses in their daily life are totally mitigated - there IS no commute or crappy bosses or fighting at home. Are they allowed to smoke in the rooms in the NIH Clinical Center Metabolic Ward? I don't know, but it seems likely that they're not allowed? Well, there goes smoking.
Physical activity is definitely different too - patients are typically either confined to their rooms or the Metabolic Ward floor overall, and at least one day a week they were confined to a sealed respiratory room that monitors air exchange and consumption to measure metabolism, so they're almost certainly moving and walking less than usual.
Moreover, each room in the Metabolic Ward has a stationary bike and treadmill, so people can keep up with their usual activity - but they're here in a much different daily routine, with a lot more free time than usual. Does exercise go up compared to their regular life? Down? Who knows! It's a total wild card. And what if they didn't have bikes and treadmills? Then anyone fit would suddenly be a lot more sedentary.
So we're still confounded on a lot of other variables that we know matter.
Still, would it be any better if they went about their usual daily lives, and we just asked them to photograph or otherwise measure whatever they ate? Not necessarily - now people would get up to the usual tricks of forgetting or not reporting snacks, under-reporting calories or weight, or even changing what they eat in noticeable ways. They may change other aspects of behavior too - the Hawthorne Effect is a real thing, where people positively change their behaviors *because* they're being studied.
So RCT's are still the best we have, but they're certainly not foolproof, especially for complex and multi-factorial things like health, diet, and exercise.
Do all those other factors and major changes in the specific RCT I broke down NOT confound the results? I don't see how they could fail to, but maybe I'm missing something.
At the minimum, they'd have to change the effect size.
There is a mountain of evidence against seed oils being healthy, though. (Animal experiments, human experiments, quasi-random designs, aggregate food supply trends, ….) Your readers might be interested in a recent working paper of mine on the topic: https://tyleransom.github.io/research/obesity-seed-oils.pdf
Yes, I have seen Dynomight's take. I wasn't convinced. I do agree that the cross-country evidence is not as strong as other points of evidence. Genetics could be part of it, and East Asia has above-expected diabetes rates. It's not crazy to think that seed oil damage would manifest differently across different gene pools.
Here are a couple of good counterpoints to Dynomight, if you're interested. The first one at the very end tackles the cross-country question.
Diet culture is also ingrained into the cynical western (especially anglosphere) political culture. It tells people that they’re unhealthy because other people put “shit” in their food, or because academics lied to them for a quick buck. It has to be the fault of some shadowy cabal, or faceless corporations. Fat Cats and Freemasons. It can never be so simple as “you eat more calories than you burn”. Accountability is always shifted
Obesity rates have tripled over the last 60 years. *Something* has changed. The problem with virtue ethics is that it tends not to produce actionable solutions to emergent problems.
In each case of an obese person (regardless of the rate at which other people are obese) the actionable solution (eating in a calorie deficit) will result in weight loss. Whether that's easy to do or not is a different question. But it's actionable.
Conducting surgery without anesthetic is technically 'actionable' in a very materialist sense. But practically, offering anesthetic makes surgery much more likely. And offering anesthetic is probably going to be a lot more effective in encouraging surgery than just deriding a population for their lack of sheer willpower. Increasing social pressure potentially yields diminishing returns, past a point.
Practically speaking, if a person wants to achieve a particular goal yet can't reasonably be pressured into adopting a mechanistically adequate solution then the solution is not actually that actionable and some approach other than increasing social pressure (which has its own costs) should reasonably be considered. Brains are machines with very real limitations. One of those limitations might be voluntarily running a calorie deficit under a given set of conditions. Similarly, bodies are machines. Increasing metabolism through exercise can be difficult for some people for a variety of reasons which merit consideration.
There are lots of very real influences on obesity that we should also be discussing in addition to a calories in calories out model. Bisphenol A exposure. Stress. Opportunity for exercise. Gut flora. The impact of various medications on weight gain. Trans fats. Sleep quality and quantity. The impact of artificial sweeteners on caloric consumption. etc.
If the number of people who opt to run a caloric deficit and then follow through decreases as a portion of the population over time, it's very much worth pausing and asking just *what's going on* to effect that population change. To the extent that saying 'they could just eat less and exercise more' fails to reduce weight in people who want to lose weight then I question whether it's a productive observation, even if it is technically true, assuming a person is able or forced to carry it out.
There is science behind various diet beliefs. We just don’t have enough to prescribe a very good general diet for everyone.
In many ways, science is like a new religion. The scientific methodology is great. But the implementation of it is a problem, as we see from the falsifications of results, or just inconclusive experiments due to factors like gender etc not being accounted for.
Many people strongly believed in communism at a point but then implementation of sharing public goods didn’t really work out.
Science requires a lot of trust and verification, and investment into core topics isn’t always profitable.
Anti-vaxers are on the rise because people are struggling to trust that the vaccines are good for them, that childhood ailments manifesting at the same time are not just correlation and below all that is a lack of trust in Big Pharma and the government to check that science is being properly conducted.
Then there are other issues like race and gender underrepresentation in scientific studies.
I admit I've never fact checked it, but I was struck by the advice that mutually contradictory diet advice could (mildly) reduce obesity because the average restriction on diet, if followed, will make eating less impulsive and hedonic. I'll also have to make sure to look at longitudinal studies, if possible, of course.
I like peanut butter. Have since I was a kid. That is the peanut butter without sugar and emulsifiers which it didn't have when I was a kid so the oil would rise to the top leaving hard packed peanut butter below. I'd have this peanut butter with (churned cream) butter and jam.
In my 20s I learnt about aflatoxins derived from mouldy peanuts and its causal association with liver cancer (especially relevant in Africa and China where peanuts were promoted as a high protein food crop). After that, for a time I dropped my consumption from daily to a few times a week.
Occasionally, consumer organisations test and report aflatoxin levels found in retail brands. The levels appear to be mostly low to present in only trace amounts (at the limit of detectability) but are independent of brand and price, which is frustrating as I would prefer to be confident that I am eating a product with consistently minimal hazard.
As far as I can make out all the peanut butter available in NZ is grown in China and most is simply packaged from peanuts ground in China though one brand grinds Chinese sourced raw peanuts.
I could give you a similar story about rhubarb and, perhaps even, field mushrooms.
Aflatoxins are produced by Aspergillus fungus that grows during storage when, I presume, the nuts aren't fully dry or when the air is humid (thank you Perplexity.ai). I see Valencia peanuts are described as sweeter so perhaps they are less prone to the fungus infection during storage.
Valencia peanuts are usually grown in dryer climates. The argument is that there's less risk for fungal contamination while growing. And they're usually processed in the same dryer area.
Repeat after me: If it *can* be selection effects, it *is* selection effects.
Once you have internalized this sentence, you will understand how fucked all social sciences currently are. Nutrition science is just exhibition #13 or so.
That doesn't mean science in these fields is pointless, but it means it needs to be much more carefully designed and interpreted. Which almost nobody does; I regularly even catch myself arguing from some paper or another, only to realize how easily it could be selection effects yet again.
So, keep up the good work; This is something that needs to be hammered into people's heads over and over again until they get it.
I’m wondering..for the people who are actually healthy, who pick up these fads and make them appear to play a casual role in health - what exactly is making the actually healthy people healthy? I’m sure it correlates with class and IQ. But.. do they just eat less calories due to higher conscientiousness or lower hunger signals or something? Or is exercise actually causal? Or something else?
I think the bottom line is that we don’t really know. “Everything good is correlated”, as they say, almost certainly some but not all is genetics, and of the rest it’s probably a mix of environmental effects, lower stress levels or similar, and also some of the behaviors are probably helpful.
I've often been concerned that studies don't seem to differentiate between synthetic DL-alpha-tocopherol and mixed D-tocopherols and tocotrienols found in foods. These are both 'vitamin E' but they are not promised to have the same effects.
Most studies showing that 'vitamin E' is not beneficial probably use the synthetic form. But many don't state their methods clearly in this regard.
I’m with you on most of what you said, except Europe part. Aldo, read the thread invisible rainbow which describes the effect of electrification on our health.
It's interesting how some diet advice sticks around longer than others. Your older contemporary Jean Brillat-Savarin wrote in Physiologie du goût (1825) that a primary cause of obesity, aside from variance in people's constitutions and lack of exercise, was eating way too much starchy food. He sounds like he's advocating the Atkins diet sometimes ("Avoid everything starchy in whatever form it appears. You still have the roasts, salads, and herbaceous vegetables."). He also notes the problem of eating when we aren't actually hungry.
Sugar and other carbs drive food cravings in "most" people.
Those cravings result in over eating. Which is why you gain weight.
That's why low carb. And keto work for so many people. They reduce cravings which means you eat less junk food which means less total calories which means you lose weight.
Also please note the word "most" studies have shown that people are different. There is no one perfect diet for everyone
Very true, great post.
Especially for things like exercise and physical health, which are multi-factorial and massively confounded with things like your existing healthiness, diet, stress, and genetics, even RCT's are noisy, and the endpoints we care most about are far off and rare events (like all cause mortality, or morbidity), RCT's generally can't run long enough to show statistically valid effects, and would be too expensive to do, and wouldn't ever actually control for everything that was impacting the endpoints.
But when you think about topics like diet and health and weight, the endpoints are multiply confounded in ways that RCT's don't really address.
Let's consider KD Halls' (excellent) study on ultra processed foods - Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake (2019):
He actually took people into the study center for 4 weeks so he could completely control their diets to compare 2 weeks of eating UPF versus two weeks of eating "real" food - both admirable and expensive.
But the health endpoints measured, and things like weight, are a factor of LOTS of different things. Diet, naturally, and that's what we're testing - but also activity levels, exercise, smoking, stress from jobs, commutes, and family and relationship, and more.
So you've suddenly brought people into your study center for 4 weeks. Now you're controlling their diet, but all the stresses in their daily life are totally mitigated - there IS no commute or crappy bosses or fighting at home. Are they allowed to smoke in the rooms in the NIH Clinical Center Metabolic Ward? I don't know, but it seems likely that they're not allowed? Well, there goes smoking.
Physical activity is definitely different too - patients are typically either confined to their rooms or the Metabolic Ward floor overall, and at least one day a week they were confined to a sealed respiratory room that monitors air exchange and consumption to measure metabolism, so they're almost certainly moving and walking less than usual.
Moreover, each room in the Metabolic Ward has a stationary bike and treadmill, so people can keep up with their usual activity - but they're here in a much different daily routine, with a lot more free time than usual. Does exercise go up compared to their regular life? Down? Who knows! It's a total wild card. And what if they didn't have bikes and treadmills? Then anyone fit would suddenly be a lot more sedentary.
So we're still confounded on a lot of other variables that we know matter.
Still, would it be any better if they went about their usual daily lives, and we just asked them to photograph or otherwise measure whatever they ate? Not necessarily - now people would get up to the usual tricks of forgetting or not reporting snacks, under-reporting calories or weight, or even changing what they eat in noticeable ways. They may change other aspects of behavior too - the Hawthorne Effect is a real thing, where people positively change their behaviors *because* they're being studied.
So RCT's are still the best we have, but they're certainly not foolproof, especially for complex and multi-factorial things like health, diet, and exercise.
Unless you have a randomization failure, RCT results should not suffer from confounding.
Do all those other factors and major changes in the specific RCT I broke down NOT confound the results? I don't see how they could fail to, but maybe I'm missing something.
At the minimum, they'd have to change the effect size.
There is a mountain of evidence against seed oils being healthy, though. (Animal experiments, human experiments, quasi-random designs, aggregate food supply trends, ….) Your readers might be interested in a recent working paper of mine on the topic: https://tyleransom.github.io/research/obesity-seed-oils.pdf
Have you read Dynomight’s take?
https://dynomight.net/seed-oil/
I’m persuaded there’s no problem with seed oils; and those who disagree need to explain why China isn’t obese.
Yes, I have seen Dynomight's take. I wasn't convinced. I do agree that the cross-country evidence is not as strong as other points of evidence. Genetics could be part of it, and East Asia has above-expected diabetes rates. It's not crazy to think that seed oil damage would manifest differently across different gene pools.
Here are a couple of good counterpoints to Dynomight, if you're interested. The first one at the very end tackles the cross-country question.
https://www.exfatloss.com/p/a-reply-to-dynomights-thoughts-on
https://tuckergoodrich.substack.com/p/response-to-gary-taubes-on-omega
I *do* say that sugar is good for you, or at least neutral, as long as you don't drink it: https://www.unaging.com/diet/diet-myths-surprises-from-my-food-research/
Diet culture is also ingrained into the cynical western (especially anglosphere) political culture. It tells people that they’re unhealthy because other people put “shit” in their food, or because academics lied to them for a quick buck. It has to be the fault of some shadowy cabal, or faceless corporations. Fat Cats and Freemasons. It can never be so simple as “you eat more calories than you burn”. Accountability is always shifted
Obesity rates have tripled over the last 60 years. *Something* has changed. The problem with virtue ethics is that it tends not to produce actionable solutions to emergent problems.
In each case of an obese person (regardless of the rate at which other people are obese) the actionable solution (eating in a calorie deficit) will result in weight loss. Whether that's easy to do or not is a different question. But it's actionable.
Conducting surgery without anesthetic is technically 'actionable' in a very materialist sense. But practically, offering anesthetic makes surgery much more likely. And offering anesthetic is probably going to be a lot more effective in encouraging surgery than just deriding a population for their lack of sheer willpower. Increasing social pressure potentially yields diminishing returns, past a point.
Practically speaking, if a person wants to achieve a particular goal yet can't reasonably be pressured into adopting a mechanistically adequate solution then the solution is not actually that actionable and some approach other than increasing social pressure (which has its own costs) should reasonably be considered. Brains are machines with very real limitations. One of those limitations might be voluntarily running a calorie deficit under a given set of conditions. Similarly, bodies are machines. Increasing metabolism through exercise can be difficult for some people for a variety of reasons which merit consideration.
There are lots of very real influences on obesity that we should also be discussing in addition to a calories in calories out model. Bisphenol A exposure. Stress. Opportunity for exercise. Gut flora. The impact of various medications on weight gain. Trans fats. Sleep quality and quantity. The impact of artificial sweeteners on caloric consumption. etc.
If the number of people who opt to run a caloric deficit and then follow through decreases as a portion of the population over time, it's very much worth pausing and asking just *what's going on* to effect that population change. To the extent that saying 'they could just eat less and exercise more' fails to reduce weight in people who want to lose weight then I question whether it's a productive observation, even if it is technically true, assuming a person is able or forced to carry it out.
Very decent points all. One small gripe I have is the body is not a machine. Bodies adapt; machines do not.
They’ve actually increased over the past century+. There’s a kirkegaard article on this.
There is science behind various diet beliefs. We just don’t have enough to prescribe a very good general diet for everyone.
In many ways, science is like a new religion. The scientific methodology is great. But the implementation of it is a problem, as we see from the falsifications of results, or just inconclusive experiments due to factors like gender etc not being accounted for.
Many people strongly believed in communism at a point but then implementation of sharing public goods didn’t really work out.
Science requires a lot of trust and verification, and investment into core topics isn’t always profitable.
Anti-vaxers are on the rise because people are struggling to trust that the vaccines are good for them, that childhood ailments manifesting at the same time are not just correlation and below all that is a lack of trust in Big Pharma and the government to check that science is being properly conducted.
Then there are other issues like race and gender underrepresentation in scientific studies.
Re
most of what you saw was because Britain is just poorer than America.
Every time i have done a scatter plot of income per head (at PPP) vs life expectancy, i get the same pattern:
- as income per head rises, so does life expectancy
- the effect is strongest at low incomes (below world median) but persists, gently, at higher incomes
- the USA is a massive outlier from the peloton of other developed countries, withblife expectancy about 4yrs below that predicted by the trendline.
Britain has lower incone per head thsn the US... means... the US should be achieving higher life expectancy than Britain.
I admit I've never fact checked it, but I was struck by the advice that mutually contradictory diet advice could (mildly) reduce obesity because the average restriction on diet, if followed, will make eating less impulsive and hedonic. I'll also have to make sure to look at longitudinal studies, if possible, of course.
My nutrition journey:
I like peanut butter. Have since I was a kid. That is the peanut butter without sugar and emulsifiers which it didn't have when I was a kid so the oil would rise to the top leaving hard packed peanut butter below. I'd have this peanut butter with (churned cream) butter and jam.
In my 20s I learnt about aflatoxins derived from mouldy peanuts and its causal association with liver cancer (especially relevant in Africa and China where peanuts were promoted as a high protein food crop). After that, for a time I dropped my consumption from daily to a few times a week.
Occasionally, consumer organisations test and report aflatoxin levels found in retail brands. The levels appear to be mostly low to present in only trace amounts (at the limit of detectability) but are independent of brand and price, which is frustrating as I would prefer to be confident that I am eating a product with consistently minimal hazard.
As far as I can make out all the peanut butter available in NZ is grown in China and most is simply packaged from peanuts ground in China though one brand grinds Chinese sourced raw peanuts.
I could give you a similar story about rhubarb and, perhaps even, field mushrooms.
Aren't Valencia peanuts generally low aflatoxin? Costco's peanut butter pretzel snacks advertise Valencia peanut butter used.
Aflatoxins are produced by Aspergillus fungus that grows during storage when, I presume, the nuts aren't fully dry or when the air is humid (thank you Perplexity.ai). I see Valencia peanuts are described as sweeter so perhaps they are less prone to the fungus infection during storage.
Valencia peanuts are usually grown in dryer climates. The argument is that there's less risk for fungal contamination while growing. And they're usually processed in the same dryer area.
Repeat after me: If it *can* be selection effects, it *is* selection effects.
Once you have internalized this sentence, you will understand how fucked all social sciences currently are. Nutrition science is just exhibition #13 or so.
That doesn't mean science in these fields is pointless, but it means it needs to be much more carefully designed and interpreted. Which almost nobody does; I regularly even catch myself arguing from some paper or another, only to realize how easily it could be selection effects yet again.
So, keep up the good work; This is something that needs to be hammered into people's heads over and over again until they get it.
Makes sense. Almost all pro vegan health claims are just epidemiological studies with confounding and describing red meat as being McDonalds burgers.
I’m wondering..for the people who are actually healthy, who pick up these fads and make them appear to play a casual role in health - what exactly is making the actually healthy people healthy? I’m sure it correlates with class and IQ. But.. do they just eat less calories due to higher conscientiousness or lower hunger signals or something? Or is exercise actually causal? Or something else?
I think the bottom line is that we don’t really know. “Everything good is correlated”, as they say, almost certainly some but not all is genetics, and of the rest it’s probably a mix of environmental effects, lower stress levels or similar, and also some of the behaviors are probably helpful.
I've often been concerned that studies don't seem to differentiate between synthetic DL-alpha-tocopherol and mixed D-tocopherols and tocotrienols found in foods. These are both 'vitamin E' but they are not promised to have the same effects.
Most studies showing that 'vitamin E' is not beneficial probably use the synthetic form. But many don't state their methods clearly in this regard.
I’m with you on most of what you said, except Europe part. Aldo, read the thread invisible rainbow which describes the effect of electrification on our health.
The Carnivore diet works well as an elimination diet.