Dangerous Mistakes of Reading the Internet

Zooko and I get a lot of questions here and on ketotic.org about our work, or about ketogenic diets. That’s encouraging, and we like hearing from you! However, we often don’t have time to answer questions in depth. I’ve decided to answer this one, because my answer to it has some common themes that transcend the particulars of the topic.

On Eat meat, not too little, mostly fat, Matthew Blanchard asked:

I have been reading through your blogs, and I find them interesting. I am considering doing the 30 day challenge. Then I found this webpage. A lot of this information seems to contradict your research. Have you seen this before? What are your thoughts on some of this?

Here are my thoughts.

The article is called Eight Common but Dangerous Mistakes of A High-Fat, Low-Carb Diet. The authors have done something frustrating, but not uncommon. They have placed most of the burden of research on the reader. They have made a swath of claims in an authoritative voice followed by over a dozen references of mixed origin: from peer reviewed papers to blogs. There is no way to tell which articles and which parts of those articles are supposed to support a given claim. That means that to verify a given point, you basically have to read all or most of the articles cited (and often further the articles those articles cite!), and then you still can only guess at why the authors think it supports the claim. On ketotic.org, we aim to show the full trace of every claim we make, which we call “end-to-end citations”. This allows the reader to know why we think something is true, and encourages them to look at that evidence critically. See the Apologia for more on our approach.

This approach also, unfortunately, takes much more work. I would say the vast majority of the time I spend writing an article comes from finding papers that clearly support what I’m claiming, even after I can find many sources that merely make the same claim. I try not to rely on authority, and I take into account the type of experiment used. On several occasions, I have found that I could not find clear support of something I assumed was true.

For example, as of last time I looked, there is no clear published research showing that lowering protein increases ketosis. I’m pretty sure that it’s true: clinicians in the field say it is true, many anecdotes report it, and I’ve experienced it myself! However, I don’t have the kind of evidence that meets my standard to claim it. Contrast my article Protein, Ketogenesis, and Glucose Oxidation with the glib statements made in Dangerous Mistake #2. One of those statements is “if you go overboard and eat too much protein than the body needs [sic], some of the amino acids in the protein will be turned into glucose via a process called gluconeogenesis”. I’m quite sure the causality implicit in this statement is wrong, and I’ve written a series of articles exploring that, starting with If you eat excess protein, does it turn into glucose.

All that said, the particular article of ours in question suggests eating an all-meat diet, which is not a well-studied variation of a ketogenic diet. We’ve tried to be quite explicit about that. We’re well aware that few ketogenic dieters do it this way all the time, and we don’t have anything even approaching a randomised controlled trial to back it up with. That’s why we’ve phrased it as an experiment; as a way to learn about how diet affects you by removing almost all confounds, including the variation in tolerance many people have to carbs. It is also true that I (and a handful of others I know) feel much healthier eating no plants, and eat essentially only meat all the time. Nonetheless, it is not very much different from the induction phase of most versions of low carb diets, which are extra restrictive for many of the same reasons as those we cite. See Wait, why eat only meat? for more on how this trial is similar to induction phases, and why we advocate it.

The main statements that I disagree with in the Dangerous Mistakes article are the ones about meat and vegetables in particular. I will briefly address “Dangerous mistakes” #3, #4, and #8.

#3. Poor gut health due to a diet high in animal protein and low in indigestible fiber

There are two claims here. First that “low-fiber, higher animal protein diets have been found to increase inflammatory gut bacteria.” I have been working on a post about the inflammatory gut bacteria hypothesis (which is all it is, the claims of the scientists who report it notwithstanding). So I don’t have much to say about it here, except that the claims vastly outweigh the evidence, and the scientists making them appear to assume that which they are trying to prove.

The second claim is that “these bacteria release a compound called TMAO after you eat animal protein, which increases plaque buildup in the arteries, elevating inflammation.” For this claim I will simply refer you to work by a researcher cited elsewhere in the Dangerous Mistakes article, Chris Masterjohn, who has shown the main problem with that claim: Meats produce no more TMAO than fruits and vegetables, or assorted grains and dairy, and the primary source of TMAO is seafood, not beef or other red meat. See his article for details about that, and more reasons to question that claim.

#4. Chronic inflammation from high intake of fat and protein and few fruits and veggies

The support for this comes in the form of the following logical chain:

  1. In one study from Tufts, the lean mass in young people was positively associated with increased oxidative stress and inflammation.
  2. The best explanation of the association the scientists could come up with was that these young people must have been eating fewer fruits and vegetables, because, the Dangerous Mistakes authors claim, “people who eat more animal protein tend to eat fewer plants”.
  3. Therefore, you should eat more of the particular plants that are fruits and vegetables (as opposed to, say, grains and legumes, which are also plants, I’ve heard…)

I didn’t look up the Tufts study, but this is terrible reasoning. The authors of the Tufts study presumably know that lean mass is usually associated with good health, which is why they were compelled to make something up after the fact to explain their results. (As an exercise for the reader, do an Internet search to figure out what lean mass is typically associated with.) Don’t get me wrong. Making up stuff is a fine activity—scientists are expected to offer an interpretation of unexpected results—but it is irresponsible to make strong claims based on untested hypotheses, especially when you are in a position of authority.

#8. A chronic acid load that degrades lean tissue and increases cancer risk

The paper in the reference list that seems to go with this claim is called Diet-induced acidosis: is it real and clinically relevant?I took the time to look at it briefly. The authors of this paper do point out that the human body keeps pH tightly regulated, and that diet can only effect pH to a small degree, which stays within normal range. They then speculate that this chronic small amount of acidity might lead to problems such as bone loss. They further cite some studies that associate acid-forming foods with bone loss. They also discuss the problem of kidney stones and its potential relationship to bone loss.

I don’t have time to thoroughly respond to this paper right now, but I’ll touch on two points.

First, diets with high-normal amounts of protein do not lead to bone loss [1], and in fact the physiological changes that happen under ketogenic conditions are lean mass preserving. These facts alone should lead you to think that the associations cited here may not apply to a carnivorous ketogenic diet.

Second, some children on ketogenic diets for epilepsy have had problems with kidney stones. However, the epilepsy diet has usually been based on KetoCal shakes (full of vegetable oil), not on meat. The best remedy for kidney stones appears to be high intake of potassium citrate. Plant advocates like to note that plants are high in potassium, which is true. However, it is also true that meat has potassium, but when we cook, it is mostly lost to the drippings. Eat the drippings. Drink the broth.


1. See for example, this review. It discusses this unsupported myth, and others. Note that it does reiterate the idea about alkalizing fruits and vegetables.

“Despite a widely held belief that high-protein diets (especially diets high in animal protein) result in bone resorption and increased urinary calcium, higher protein diets are actually associated with greater bone mass and fewer fractures when calcium intake is adequate. Perhaps more concern should be focused on increasing the intake of alkalinizing fruits and vegetables rather than reducing protein sources.”

but note that it is the potassium content of fruits and vegetables that leads them to that conclusion:

“[N]et renal acid excretion can be predicted from the ratio of dietary protein to potassium because the dietary intake of potassium occurs mainly as salts of weak organic acids and therefore has an alkalizing effect (30, 33). This relationship may explain the reported beneficial influence of fruit and vegetables, the major dietary source of potassium, on bone health”