5,454 karma · joined August 11, 2009
Engineering Manager at Gemini, formerly at Sana, Justworks, Etsy
will . dewind at gmail.com
It absolutely says that. Absent other factors changing, a reduction in caloric intake will result in a weight loss. It does not disagree with there also being other inputs to CO.
> Humans, difference in body composition: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2645638/
So this study is actually not bad. It has three small flaws that I can find, but is still compelling regardless. It is one of the only studies I have ever seen that did a decent job controlling the food. I'd like to see it repeated and at a larger scale, of course. The flaws are:
1. Malabsorption is a real issue when consuming nearly 2400 calories in a small window. This could be a real confounder with regards to metabolic differences. The malabsorption could be resulting in fewer calories being consumed.
2. Massive withdrawal rate in comparison to other dietary studies. The authors address this, I think reasonably, by concluding that IF is actually really difficult for a lot of people to comply to. This isn't necessarily a strike against IF being metabolically better, but if we dig into exactly why the diet is hard to comply to and force the people who didn't want to comply to comply, we might get different metabolic results. There is a large potential for survivorship bias here.
3. Sample size is tiny blah blah. This doesn't disregard the study but it does limit its applicability.
Good find, though, thanks!
As for the other two, I don't think they meet the requirements really.
No, we can reason about this fairly easily knowing the extraction costs of calories from macronutrients. You're correct in that macronutrient content can affect CO, but you're vastly overstating the amount. There is no model of digestion we've come up with that accounts for much more than 10% of calories going to digestion, so even if you max that out, you still end up with pretty similar levels of obesity. We cannot fix this problem without radically reducing the amount we eat, regardless of the macronutrient content.
> I seem to recall a bunch of animal models where compliance was controlled for and still showed a strong correlation between insulin and set weight.
Please send links, all I can do is respond to the specifics of an actual study.
But to steel man your argument: yes, macronutrient composition obviously matters, but you can lose weight with a low calorie diet of any composition and you cannot lose weight with a high calorie, macronutriently balanced diet.
That is, unfortunately, exactly why I am so acquainted with this stuff, and yeah, it is pretty frightening to have doctors tell you they just don't know. But to be fair, I'm talking about being wrong about fundamental biology, not medicine, in which massive shake ups like that are far rarer. It's also not like this stuff hasn't been studied.
> Anyways, how does what you're saying mesh with the fact that extra glucose is stored in your body's reserves while ketones are secreted with urine and breath?
Not sure what you mean but fwiw glucose is also secreted in urine and breath.
Maybe they do matter somewhat, but here's the deal: if everyone who was obese in this country all of a sudden started eating a diet that was macronutriently ideal and consumed at the perfect times, but maintained the same number of calories, there would be no difference in the prevalence of obesity. I think the science really confidently supports that position, and I'd be happy to look at anything you think denies it.
> Depends on what you call epidemiological study, but using the diabetic population, the causal link between insulin and weigth gain is pretty strong. The book "obesity code" make of good case for that.
I am very familiar with Dr Fung's work and don't think that any of it makes the case you are trying to make. All of his work provides proof for what I've been saying over and over in this thread: IF and Keto can be great ways to create compliance to a diet, but there is no evidence, including the work of Dr Fung, that shows these diets behaving differently from simple calorie control outside of the compliance aspect.
That's way too broad a statement for me to blanket agree or disagree with. Short answer: no I definitely don't think that. Long answer: basically everything else I said in this thread.
Right, this is the point: it's about compliance, which is real and important, but very different from metabolic change.
> Please don't tell me there is conclusive evidence on anything related to nutrition. The second statement is not true, differences have been shown.
No, there is no study showing the same number of calories consumed over different feeding schedules have any different outcomes. I am 100% confident about this. To be clear: there are studies showing people on IF eat fewer calories, and that eating few calories has net benefits. That's great, but that's compliance not metabolism.
> Yes, it's not binary, but the easy way to increase lipolysis is to decrease insulin.
Again, time windows. Please consider time windows. In a 24 hour period this stuff is all neutralized. Or maybe it's not, but we don't have any science showing it's not, and quite a bit showing it is.
Yep, I would not suggest anyone take up a diet of pure sugar, only that if you are trying to use science to set your diet we aren't really confident, at the level of macros, about what makes a difference, but we are very confident about excess calories.
Most IF studies do not show any metabolic difference, and the ones that do do not control for calories. I'm sorry to shift the burden of proof, but I feel like CICO is the default position and additional nuance requires proof, not the opposite. Also I am lazy.
> Again any claim to this ? Most reference i see have this causation relationship reversed....
Yeah totally, so the claim is not that insulin does not cause damage, it is that where we have recorded that damage is directly to cells in labs, but we have not translated this to epidemiological studies. This is sort of like the Schroedinger's Cat of nutrition: we see these effects in the lab at micro scale, but we don't actually see them translate to real world effects when we look for them at the macro scale. One explanation is that with the same way we see insulin causing damage, we see caloric deficit causing repairs. Maybe these cancel each other out when calories are controlled for? I don't have an explanation, but the key is that we don't observe this stuff on the macro level.
My conclusion is that but slightly more conservative: more that science is really hard, and we have not been able to show that macro nutrient composition matters nearly as much as we thought. It may still matter quite a bit, though you'd expect it to be easier to measure if it did, but again science is hard.
> Yet, I struggle to entertain the possibility that the guy eating a lot of pizza and ice cream and candy is going to encounter health problems a lot sooner than someone of the same CICO eating avocados and chicken and broccoli.
I can show you small scale examples of people doing this and not having problems (http://www.cnn.com/2010/HEALTH/11/08/twinkie.diet.professor/...) but I can't show you a strong epidemiological study that confirms this, unfortunately.
> What about skinny people who get heart attacks?
This does happen, though of course far more rarely. Lots of stuff impacts this (genetics etc.), and I don't know nearly enough about heart attacks to comment, but I will say broadly that one of the worst things you can do for your heart, and most of the rest of your body, is to be fat, and the only way to be fat is to eat excess calories.
So the main problem would be nutrient absorption, and there would be a number of problems that would impact it, from literal surface area of your digestive system, to current bacterial populations etc. etc. These problems matter less in a 24 hour window when you are far less likely to hit the same thresholds eating a reasonable amount of food. But basically, you could not effectively break down and absorb all the nutrients quickly enough, and you would have pretty nasty shits/vomiting.
But, let's say somehow you could, and somehow your body ended up storing a bunch of fat. In general, the second half of your statement is totally doable because the human body absolutely kicks ass at energy storage and expenditure. https://www.sciencealert.com/the-true-story-of-a-man-who-sur...
I agree, I'm trying to more accurately state what we confidently know based on the available science, rather than what might be true based on our hunches. Insulin spikes being bad are 100% in hunch land.
> The complex kinetics depends both on rate as well as average / basal dosages.
I agree but, again, very little science that disambiguates effectively.
> Spikes can trigger different biological responses that respond to thresholds and equilibria.
Right, this is what the basic science has shown, but what we don't know is what difference this makes within a 24 hour window, or epidemiologically over time.
> It's a differential problem, and the answer isn't easy.
Definitely agree.
I have repeatedly said in this thread this doesn't apply to diabetics, who do have an insulin pathology. Diabetics do need to worry about insulin, non-diabetics do not.
> Your body is very bad at burning fat in the presence of insulin. Meal time is only important as to not keep your insulin level elevated all the time.
There is no science that disambiguates between large spikes followed by zero food in take (IF), and small spikes continually at the same calorie levels. Please feel free to send any study you think does.
> It's reversible by either supplementing more food
Again this is only relevant to the post-prandial window. If you are overweight, and have a slow metabolism, and respond by eating a lot of food, you will not successfully "speed up" your metabolism outside of the prandial window. If you are overweight and start dieting, your metabolism with slow down, but the evidence overwhelmingly shows exercise counteracts this, and that the slowdown is reversible when your body fat percentage lowers.
> If you are eating all the time, the short term situation becomes the long term situation, not barely on 24 hour periods but for years on end. This is especially true for diabetics and pre-diabetics, as their insulin level will lower way slower.
No, this is only true if you are eating all the time and also eating too many calories. . Again, please send a link to a study you think says otherwise.
> While I'm not a fan of isocaloric studys, this should work, yes.
Why are you not a fan of the only kind of study that could possibly prove your point? And why hasn't this study ever been successfully completed? Why have studies that attempted to do this been unable to show any difference in outcomes?
> I'm genuinely interested as to what your understanding of insulins / glucagons effect on your body's ability to access fat is.
I don't really know what you're implying here, but I suspect you are still getting hamstrung by the prandial window. Yes, insulin can block your ability to access fat in the short term, no it will not in the long term, and it does not do so entirely either. If you are eating the same number of calories and thus constantly releasing insulin you are blocking some, but not all, lipolysis. Lipolysis is not a binary, nor is insulin release. You can have both of them happening at the same time, even though insulin is blocking some of the lipolysis.
Via gluconeogenesis, but it is, for sure, less effective than protein based gluconeogenesis.
> If you don't have elevated blood sugar from eating a bunch of carbs, what effect does insulin have?
You have elevated blood sugar from having an excess of calories. Dietary sugar is most rapidly converted to blood sugar, but all excess calories are eventually converted via gluconeogenesis.
> Does it cause blood sugar to go even lower?
Yes, and in extremes can cause death.
> On gluconeogenesis: I thought this was demand driven rather than supply driven. Insulin apparently is an inhibitor for this. So how does this work?
Gluconeogenesis first causes the breakdown of protein into available glucose. What the body does with it after that has a lot to do with the body's current energy demands. Insulin is an inhibitor for this, but insulin is also only elevated post-prandial, and most gluconeogenesis happens off schedule from eating (or at the very least after the initial insulin response).
The major error people make when trying to understand this stuff is to focus on too short of a time frame, in this case the very small period right after a meal is consumed, rather than something longer, like a 24 hour period. As soon as you zoom out to 24 hours a lot of these ideas stop making sense because your body has a large amount of resources dedicated to effectively regulating energy expenditure.
Interestingly context can totally reverse the implications here too. For instance: as a weightlifter I want a heavy insulin response and I want a heavy inflammatory response because both of those help build muscle. So even saying that inflammation is bad requires a bit more nuance. It's very complicated, which is why Calories In Calories Out is such a powerful simplification. It doesn't tell you everything, but everything fits into that explanation, and if you can figure out how to control calories a ton of other bad things stop happening too.
Right but what happens when you no longer inputting calories and thus no longer have high insulin? Then you start burning the fat again. You can only keep your insulin levels pathologically elevated if you are eating enough to support that over the long term, regardless of when you eat. Again, your body is very good at supplementing energy needs with fat when it needs to. Again there is no study showing meal time having any impact when calories are balanced, and several confirming the opposite.
> It absolutely does matter, because your body does not burn a static amount of calories but adapts to the situation. Slowing metabolism is real. If you can't burn energy (low caloric input but frequent meals), your body will do what it needs to survive.
Your body adapts to a short term situation (a meal and the small post-prandial period) but in the long term (24 hour period) it makes zero difference.
> Slowing metabolism is real.
Yes but it's also reversible by, shockingly, increasing caloric demand.
To be very clear, the proof required for what you are suggesting is a study with a minimum of 2 groups eating the same exact meals but at different times, and being forced to finish the entire meal and experiencing different outcomes. This has never happened, and based on our current understanding of metabolism never will. Maybe it would change, but this would be a massive massive change in our understanding of how the human body deals with energy.
The problem is that there isn't any real data, on an epidemiological level, that says this matters. I don't mean to be offensive, it's just a weird obsession with the Keto community I don't really understand. Same with insulin spikes. They just don't matter. It's averages over a long period of time that matter. If you're eating two large meals, you're still releasing a lot of insulin, you're still doing the same amount of fat storage, you're just timing it differently. If that works for you, great.
> I can tell you this would not happen if I ate 6 small meals per day. And I know because I measured myself. But it’s also how the body works ... ketosis happens after the liver’s glycogen stores are depleted, which takes 10-12 hours and the switch to burning fat isn’t very efficient and getting less so with age. Your body needs a break from food in order to mobilize the energy in your adipose tissue.
This is a semantic construction that is not supported by the science. You are correct that you would not measure nearly as high blood ketone levels, but you are incorrect in the idea that you wouldn't consume essentially the same amount of fat over the same amount of time given the same caloric inputs. Most people also go into ketosis while they are sleeping.
> With 6 small meals what actually happens is constant hunger because you’re interrupting lipolysis every time you eat.
This doesn't matter. Given the same energy requirements and the same energy inputs you're going to burn the same amount of fat over a 24 hour period regardless of when those inputs come. Anything else is an extraordinary claim that requires extraordinary evidence.
> This happens even with the ketogenic diet, because even meat triggers a hormonal response, small but not insignificant.
Yes, keto is actually a low protein diet as much as a high fat diet. Most people ignore the first part.
> Also speaking from my own experience, I suffered from hunger at night, which made me binge. Switched to IF, hunger is now gone. Sounds like a paradox but it really isn’t.
This is great, and a very real reason for people to try IF/Keto/Other diets. But it does not suggest anything other than a compliance hack that is working well for you.
> The IF people (along with the Keto people) use a variety of tricks to try and prove their point. There is absolutely not a single paper showing calorically balanced food intake has clinically significant outcomes. Period, full stop etc. There are tons of papers saying things like if you force people to eat in constrained windows they eat fewer calories (or if you force people to eat high fat diets they eat fewer calories etc.). Those are all useful outcomes: they are "compliance hacks." But they idea that any of these are significantly metabolically different is not at all supported by the science.
1. These papers do not show different outcomes with calorically controlled inputs, as I mentioned. One of the tricks used by the IF scientific community is to serve the same number of calories and so it looks like the same number of calories were consumed in the abstract. They never are, because if this did happen it would have massive implications for our understanding of the human body. It would require us to be wrong about a lot of things.
2. There is a difference between statistical significance (which could be, but has not been, shown in calorically controlled diets), and clinical significance (which definitely has not been shown in calorically controlled diets, and likely will not).
I assume you mean low glycemic load. I don't think so. I think we should all focus primarily on eating the right number of calories, and the behaviors that get us to do that. If that is Keto, or IF, or Low GI, even something random like afternoon naps then great.
Also, for some populations, like athletes, a low-gi diet would end up resulting in decreased performance. I think it's really difficult to make sweeping dietary recommendations because there are just so many confounding factors.
We can't answer this confidently, the way I would say it is: there isn't strong evidence to support the idea that either is better.
> Given my current understanding from all the intermittent fasting hype lately, it is indeed better to eat big meals rarely rather than small meals frequently. For some reason, your post got me slightly confused because I lack the necessary knowledge to have a proper model about nutrition in place.
The IF people (along with the Keto people) use a variety of tricks to try and prove their point. There is absolutely not a single paper showing calorically balanced food intake has different clinically significant outcomes. Period, full stop etc. There are tons of papers saying things like if you force people to eat in constrained windows they eat fewer calories (or if you force people to eat high fat diets they eat fewer calories etc.). Those are all useful outcomes: they are "compliance hacks." But they idea that any of these are significantly metabolically different is not at all supported by the science.
As for yourself: nutritional science is weak, and if the nutritional science says do something, and you observe a bad outcome in yourself from doing it, I'd take your own personal data over the nutritional science, frankly. For instance: fiber is a great modulator. If you go to a GI specialist and say "I'm having dietary issues," increasing fiber is usually the first thing they recommend. But also, sometimes that doesn't work and then the next thing is decreasing fiber (FODMAP diet). So it's really hard to make sweeping recommendations because the body is so complex and nutrition involves so many different cooperating systems that can all go slightly wrong.
Right, and the intermittent fasting crowd uses the same tricks as Gary Taubes but in reverse: look what happens when you fast! All these great reparative things happen. But it's all basic science and ignores a long term (even 24 hour) view of consumption. Ultimately there might be small, statistically significant metabolic differences between consuming the same number of calories in 1 meal as spread out over multiple meals (ie: 1 big insulin spike vs multiple smaller ones), but we've had a hell of a time showing them, and and even harder time showing clinically significant differences.
The problem is not dietary sugar, it is blood sugar levels, which are elevated by many things (including protein and fat), but more than anything by excess calories. The problem is not peak blood sugar levels, it is average blood sugar levels.
Edit: to quickly reply to your TLDR edit:
> Thus, protein- and fat-rich foods may induce substantial insulin secretion despite producing relatively small bloodglucose responses.
This is only true if you completely ignore everything else going on in the body. Gluconeogenesis is a thing, and just because dietary fat or protein doesn't immediately increase blood sugar levels does not mean long term they don't. This stuff can't just be looked at as "input/output" you have to take into consideration the existing state of the person before the input. /edit
You can reason about this fairly easily: take a caloric intake of 2000 calories consumed all at once vs in 10 meals over the day. The all at once crowd gets a huge insulin spike. Oh no! Fat storage, right? Sure, but what happens the rest of the day when there is no dietary input? Fat burn! The 10 meals/day crowd probably never gets as high blood sugar levels, but also has much higher throughout the day, and accesses fat stores less. But guess what? Net the body is really pretty good at smoothing this stuff out, it's sort of the entire point of fat storage.
Where this gets tricky is that high insulin can eventually lead to low blood sugar dips. People without diabetes can simply wait this out, and the body will start burning fat again and raise the blood sugar. But this also makes many people feel hungry, and so they eat more, which raises their average blood sugar levels, and the cycle continues. Some people have a hard time with this, others do not, but it's really important to understand that the insulin is causing people to consume more calories which is causing damage, not the insulin itself.
This is really overstated. The paper is a paper on round worms, which is really one step above basic cellular science (which is important but only a building block of the whole picture). We do not have large scale, epidemiological studies about sugar, and the ones that we do don't point to sugar behaving differently than other calories with regards to health outcomes. The problem is blood sugar not dietary sugar, and it turns out high blood sugar can come from many places that are not dietary sugar, and a calorically controlled diet that is high in sugar does not result in high average blood sugar levels.
I don't know how this is different from what I said and my point is that I don't think they are a good tool to detect problems. In most cases I would consider burn down graphs to be trailing indicators of problems, not leading, which is why I consider them more appropriate for reporting up the chain than down.
> Updating a burndown chart should be a grand total of 20sec. What takes time is the discussion you have when you're late. Yes, solving problems takes more visible time than burying them :)
Right, my point is not about checking off a ticket and letting an automatic burndown chart update. My point is about the kind of project management where you have an intense focus on the how instead of the what. It's very inappropriate for early stage startups because of how quickly requirements change. It also represents a communication overhead.
> It's a terrible tool for reporting up the chain. No reason to report the details of the sprint up the chain
We can disagree about whether or not it has value for reporting up the chain, but why wouldn't the status of a sprint be something you'd want reported up the chain?
> It sometimes is (usually when the situation is dire), but most of the time really not. I don't know if the burndown chart is the best tool to identify bottlenecks though, at best it shows that there is some problems that might be bottlenecks. It's just a very lightweight tool so make problems immediately visible. If you find a burndown chart to be too much overhead... I don't know what tools you'll ever use
We're getting semantic, it depends on how you identify bottlenecks. My experience has been it typically looks something like "a bunch of tickets are stuck in design, design is a bottleneck" which isn't really helpful when the core issue is, for instance, that you don't have enough designers. That's typically a reality most people know and understand, so I don't think it's super valuable here, nor do I think a different project management or organizational system fixes it. Interested in what you consider to be a bottleneck and how you would identify it.