Toward a unified theory of weird internet diets

(and an old-timey one, for good measure)

Weight-loss diets are, it is known, generally pretty bad at their jobs. They don’t help people lose much weight easily, and when people do painstakingly lose weight on them, they tend to gain it all back when they stop. The internet has gotten fed up with this state of affairs of late, though, and has stumbled upon some pretty interesting exceptions to the rule. They are, at first glance, pretty dissimilar.

Slime Mold Time Mold ran an extensive Potato Diet Trial, on which participants lost an average of 10.6 pounds in four weeks by eating, roughly, as many potatoes as they liked¹ and nothing else. They’re now running a Potato Diet Riffs Trial, in which they ask people to change something about that protocol to see whether it halts or preserves the weight loss effect, to help figure out why and how it works.

ExFatLoss has been doing the notional opposite of that: “ex-150”, a (mostly) heavy cream and (just 150g/day of) ground beef diet, which has them down 60 pounds, from 292 to 232, in a year or so. (But they’ve been doing it super inconsistently, switching up protocols out of, as far as I can tell, a general feeling that they must need more protein than that. They initially dropped 43 pounds in 5 months.)

Brad from FireInABottle, formerly of “Croissant Diet” fame, is currently trialing an “Emergence Diet”, based on his theory that modern metabolic dysregulation somewhat parallels “torpor” in hibernating mammals. Having seen studies that suggest branched chain amino acid (BCAA) metabolism is dysregulated in obese subjects, he is strictly curtailing his BCAA consumption by consuming glass noodles (which are 100% starch) as a staple, with very little meat, and supplementing that protein with low-BCAA gelatin. He’s restricting his fat consumption to <10% of his total calories per day. (Also he’s taking supplements, some of which he sells.) He’s purportedly down 14.5 pounds in 28 days.

(The above diets are not intended to function by way of calorie restriction; they are ad libitum diets.)

Walter Kempner isn’t some guy blogging on the internet, but I did recently find out about him on the internet and his studies seem relevant, so I hope you’ll forgive me for mentioning him here without changing the title. He was a doctor treating hypertension in the 1930s. There was no good way to do that (patients with high blood pressure generally just died), and people thought he was crazy to try and treat it with diet. The foods allowed on his diet protocol were: rice, fruit and fruit juices, and table sugar. This reliably made his kidney/hypertension patients not die, and improved a number of their metabolic and cardiovascular markers. He was later successful in using the diet for weight loss… though he admitted to literally whipping some patients into compliance. Still, the weight loss results for his “massively obese” patients were, on average, 140 pounds. He took before and after photos, which are dramatic:

Each of these people lost more than 110 pounds in 11 months or less. It is my understanding that Kempner did restrict calories for the weight loss trials. Still, it’s impressive. And it consisted of very little fat or protein:

Meditating on all these staggeringly effective diet protocols at once (something I’m doing because recent bloodwork came back with very high LDL cholesterol levels), it’s conspicuous that they all match on the low-protein angle. Ex-150 is very high-fat and low-carb, and the others are high-carb and low-fat, but they all restrict the protein (and BCAAs, in case those matter) to a considerable extent.

How considerable? Helpful redditor Suitable-Attempt2577 posted a spreadsheet pulled from myfooddata.com with some BCAA data for various foods, which I sorted by total BCAA content and supplemented with ExFatLoss’ new nutrition search tool to get a sense of these diets’ total daily BCAA.

Based on that myfooddata spreadsheet, 2,000 calories per day of potatoes has between 42 and 55 g of protein, and between 5.13 and 7.98 g of BCAAs. With the allowed sweet potatoes (lower in protein and BCAAs) and oil (containing neither) included, the amount would likely be lower still. 2,000 calories of white rice would come in at 6.84-7.72g of BCAAs, but with low-protein fruit, juice, and sugar making up some of the diet, that number could be averaged down considerably. The 25g of protein listed in Kempner’s example 2400kcal diet above suggests a potential BCAA consumption of <5g, as rice appears to have less than 20% of its protein from BCAAs (an all-rice 2000 calorie diet would clock in with more like 35-40g of protein).

Ex-150’s 150g of 80/20 ground beef has 25g of protein, and 4.41g of BCAAs. But 2000kcal of heavy cream would add an additional 17g of protein and 3.23g of BCAAs, for a total of 42 grams protein and 7.64 grams BCAAs. (He generally eats more cream than that, I think, but the cream is the less protein/BCAA-heavy part of the diet. Even with 3000kcal of cream, he’d only get up to a total of ~50g of protein and ~9g of BCAAs.)

Brad’s Emergence Diet is targeting under 8g of BCAAs. The specified 1oz of gelatin brings 24g protein/1.59g BCAA off the bat, and he’s sometimes mentioned eating 6oz of lean pork along with that, which might bring it up by up to an additional 40g protein/7.28g BCAA or so (if very lean). Let’s say under 65g protein, 9g BCAA for sure.

The high-end numbers for these are all coming in very close together. By way of an interesting comparison, ExFatLoss has trialed an ex-225 diet, with 50% more ground beef, which he figured would basically work the same, but which didn’t lead to fat loss. That extra 75g of ground beef raises his theoretical high-end protein levels to around 65g and BCAAs to around 11g. (There were other factors involved that may have confounded the protein increase; I believe he mentioned he’d probably also been eating more cream at this time?)

So, to table this up:

The weird wonder-diets all clock in under 65 grams of protein (unless Brad has amped up the allowed non-BCAA protein in favor of directly limiting the BCAAs, in which case all but Emergence do), and under 9g of BCAA.

We can generate testable hypotheses about this!

If they all have the same mechanism—far from guaranteed—will any diet work, so long as BCAAs are limited to under 10g per day? Or, will any diet where either carbs or fats are limited to under 10% or so work, if BCAAs are limited to under 10g per day? (Or, for the seed oil-suspecting among us, that with the additional caveat that it avoids a great deal of polyunsaturated fats or linoleic acid, as these all do?) 

If only the high-carb, low-fat diets have the same mechanism, then can you make any high-carb low-fat diet work so long as you’re under 10% fat and ~9g of BCAAs?

Brad recommends the Emergence Diet as basically a glass noodle/isolated starch diet, where the staple starches are kept low in protein to allow room for gelatin supplementation (he thinks the increased glycine provided by gelatin is likely helpful) and some meat consumption for hedonistic purposes. But it could essentially be on a spectrum with the potato and Kempner rice diets; certainly they fit under the 8g BCAA and 10% fat limits. We’d expect a mostly rice- or potato-based diet, with additional calories from fruit or fruit juice, to allow enough headroom for the gelatin supplementation. (I have started testing this Frankenstein HCLF diet this week, with rice and potatoes as staples, allowing myself to have veggies and gelatin broths and only very tiny quantities of meat. I’m not sure if it counts as a potato diet riff, but I’ll ask Slime Mold Time Mold!)

---

If some form of this theory holds, I’d predict the following diets lead to similar effective weight loss:

Observed:

Potato mono-diet

Kempner

Emergence-style glass noodle and low-meat diet

(Ex-150, if the rule is just to restrict BCAA and pick carbs or fat?)

Theorized, testable:²

Sweet potato diet (with potential need to supplement with minor protein and fat?)

Gluten-free pasta in tomato sauce diet (wheat flour has high BCAAs relative to gluten-free alternatives)³

Taro or cassava diets supplemented with minor protein and fat

Rice noodles with veggies and extremely minor fat/protein supplement

(Potatoes and cream diet: if only the BCAA restriction is load-bearing, and you can still do high carbs and fats together?)

We would predict the following diets do not lead to weight loss:

Potatoes and steak (testable in Potato Diet Riff trial!)

Wild rice diet, white rice and beans diet (wild rice, beans probably too high in BCAAs/protein), unless protein/BCAA content heavily funged against with sugar and fruit.

---

I’m very interested in testing these hypotheses, as I find Slime Mold Time Mold’s attempts to study and explain the obesity epidemic through crowdsourced science admirable and just kinda fuckin’ rad. Knowing what elements are necessary for the diet to work will be very instructive for helping people lose weight, but it will also help us to figure out whether the effective diets can tell us anything about why people are getting so much fatter in the first place. SMTM thinks the potatoes may be low in some contaminant, but it may turn out to be the case that (something like) low protein just makes people drop weight irrespective of the cause of their weight gain, and that we learn nothing about the root cause and only figure out easy fast ways for people to lose weight. After all, pneumonia isn’t caused by a deficiency in penicillin, but penicillin can still treat it.

I will follow up soon with a post about takeaways from the BCAA spreadsheet for specific foods’ protein and BCAAs, for anyone wanting to try to stay under the emergence diet’s recommended limits, in case they accurately reflect the load-bearing elements of these HCLF diets. In the meantime, please comment if you’re interested in any help figuring out if you’d like to try one of the hypothesized confirming/disconfirming diets, or if you have any other interesting thoughts about these diets’ areas of overlap or discontinuity!

1. The diet protocol included sweet potatoes, and a bit of oil and spices–please don’t go off and just only eat potatoes without reading the linked protocol and disclaimers!

2. I have not yet vetted these for nutritional completeness, please don’t do them without responsibly checking whether you’ll need to supplement something or die.

3. This one is interesting, as Slime Mold Time Mold found tomato product consumption was inversely correlated with weight loss on the half-tato diet trial. This has led them to consider that potatoes may be low in some theorized contaminant (they think lithium) implicated in the obesity epidemic, and that tomatoes may be high in that contaminant. But their result is slightly confounded by the fact that they told participants they might want to avoid tomato products—this could be a proxy for participant adherence. If someone wants to try the gluten-free pasta diet and still loses weight, that would be good evidence that it’s just the protein and fat restriction, and that tomatoes don’t stop the effectiveness. If they don’t lose weight, maybe tomatoes are the problem!