“Processed”​ is a Useless, Empty Descriptor of Food

I posted this last week to LinkedIn: https://www.linkedin.com/pulse/processed-useless-empty-descriptor-food-karthik-sekar/

I’ve seen some profoundly twisted takes on alternative foods and how “processed” they are. In this article, I clarify some confusion, explain the chemistry and biology behind food digestion, and argue why “processed” is a meaningless food descriptor. 

If we genuinely want tasty, affordable, healthy, kinder, and more sustainable food, we should use the better metrics and concepts already out there. 

What do our bodies use protein for?

Note from February 9, 2020: I’ve advanced this analysis for the non-fiction book, The Future without Animal Products. Thanks to the help of some scientific peers, I found numbers and ways to quantify the “protein recycle”. Stay tuned for the more interesting, improved book version!

Note #1: I was wrong about fat being able to be converted to sugars. That chemistry exists in microbes but not in humans. Text has been adjusted accordingly. Thanks E. Noor!

One of the most cited reasons for eating meat is that it’s nutrient-rich, in particular that it contains lots of protein. With this post, I dig into the nutritional aspects of protein. The default reaction is to consider nutritional research, but I disagree. First, I share my concerns with nutrition research in order to motivate another approach.

Nutrition research primarily sits at an epidemiological level, meaning that the nutritional benefits or detriments are extrapolated from public health data. The most influential research perhaps comes from the Seven Countries Studies, where the the diets and health outcomes (e.g. obesity) for seven different countries were measured. The studies concluded that fats were detrimental and should be minimized in diets. The authors would likely not have drawn the conclusion had France or Switzerland been included in the study. French and Swiss people are relatively healthy with a high amount of fat in their diet. Nevertheless, the studies emboldened food manufacturers to replace fat with sugar in foods, and the obesity rate only exacerbated. Epidemiological studies succeed when a single factor has a pronounced, clear effect (e.g. vitamins, tobacco), but they resolve nuanced effects poorly. An epidemiological study will not easily gauge the optimal amount of protein in a diet or what we need it for.

Good thing I ate all that fat. It’s keeping me warm up here.

The next rung up the scientific rigor ladder sits controlled studies: a treatment and control group are chosen and given separate diets to elucidate the effect of the treatment option. Even though nutrition researchers are performing more controlled studies, many problems remain. Primarily, I find many of the studies highly susceptible to the reduction fallacy. Many of the studies hypothesize that something is bad (e.g. sugar, fat, junk food) and/or good (e.g. omega 3 fatty acids, a glass of red wine a day). How much fat is bad? Fat is essential to live. A little fat in our diet shouldn’t kill us. Other problems surface too: What is junk food? If I take a banana, mash it, dry it, and form it into a bar, is that junk food? What about if I add some xanthan gum to bind it better? Maybe a smidgen of sulfites as a preservative? Where is the line? Do these results hold for one population (e.g. children and adults) versus another?

Nutritional research is a poor starting point to answer the prompted questions. Instead, we can take a physics-like approach and consider what goes into the body and comes out in terms of protein. If we know exactly where protein goes (e.g. hair, energy, etc.), then we should be able to calculate the amount of protein that we need to consume. Metabolism research will prove helpful here. When we eat food, we break it down into chemical constituents. These constituents undergo chemistry (metabolism) to form everything we are right now (biomass) and to energize the machine (maintenance). Metabolism is highly fungible: Different biomolecules constantly interconvert depending on the demands of the body. Sugar convert to proteins. Fats convert to energy. Protein to fat. Et cetera. (By the way, this highlights another problem with nutritional reductionist approaches: Consumed fat doesn’t remain fat!).

So why do we need protein? Human biomass is mostly protein. Ignoring water weight, we are almost 50% protein! So does that mean that 50% of diet should be protein, which can be broken down and form our proteins? No. As mentioned before, the fungibility of metabolism convolutes this. Furthermore, we generate biomass differently for each period of life. A few things need to be explained.

How do we synthesize proteins? Proteins are polymers, or a long chain, of amino acids. Amino acids are molecules roughly the same size as a sugar molecule. Unlike sugar molecules, amino acids have nitrogen incorporated in them. Sugar can be metabolized and turned into amino acids when a nitrogen is available. Almost every food oozes nitrogen. In fact, you consume so much nitrogen, that a waste dispensing system is biologically necessary to remove the excess.

What about essential amino acids? Of the 20+ different kinds of amino acids, 9 are essential, meaning that our body does not have the chemistry to produce them metabolically from something like sugar. Therefore, these 9 amino acids are not fungible with the rest of metabolism. However, they are constantly recycled. When your body breaks down a protein into amino acids, the amino acids build into new protein; therefore, amino acids do not have to be replaced completely by food. Furthermore, you don’t have to get these essential amino acids only from protein. If you consume the amino acids directly, then your body would actually save the cost of having to break down the protein. Plants and nuts drip with such essential amino acids [1].

Given all of these issues, we need some way of integrating them into a holistic view; I turn to metabolic models. Metabolic models describe the input (food, nutrients), how the inputs convert into intermediates (e.g. amino acids, fatty acids), and how the intermediates form into biomass and maintenance. The models are constantly updated and, most importantly, quantitative. They explicitly quantify the amount of nutrients needed to make biomass and/or maintenance. Furthermore, the models account for the fungibility of metabolism.

The latest such model for humans was released recently [2]. So how much protein do we need? Well, it depends on how much we convert food to maintenance versus biomass. Infants allocate food moreso to biomass compared to adults. Adults are mostly using food for maintenance. According to the model, the amount of amino acids or protein needed for maintenance is tiny. Therefore, let’s focus on biomass generation. Humans add proportionally the most biomass as babies, and therefore, they need relatively more protein then, so let’s zoom in there. According to the model, 1 gram of biomass requires about 0.8 grams of amino acids/protein. (You can download my calculations here). Babies, when they grow their fastest, add roughly 25 g of biomass per day (or 1.7 pounds per month) [3]. This means that babies need about 20 g protein/day when they grow fastest. This seems to jive somewhat with current nutrient recommendations.

Unlike babies, we have two sinks of protein: the hair we grow (about 0.17 grams per day) and the skin we shed (about 1 grams per day). Not much and doesn’t explain where the majority protein goes. (See Appendix for calculations.) We soon diverge between the physics-like approach and the nutritional recommendations. For example, the other period of rapid growth is puberty. At the peak, voice-crackling males add ~450 g of biomass per month (requires 12 protein grams per day). This falls well short of the recommendation (52 g per day).  As adults, we synthesize zero net biomass (per the ontogenetic model), unless you’re a fastidious body builder. Yet, the current dietary recommendation is 56 g per day.

I do not know where all this net protein must go assuming the recommendations are correct. (Are they?) Of course my calculations do not account for protein loss due to incomplete digestion, inefficient recycle, damage, or poor uptake. Those certainly needs to be quantified and may be significant. However, we’re missing a lot of protein (more than +90%). Please let me know if you have ideas. Hopefully, we complete the protein balance. This will boost our nutritional understanding.


Where does protein go in our body? Taking a physics-like, mass balance approach, we find protein goes mostly to biomass and little into hair and skin. This discords with the dietary recommendations for protein intake (unless you’re a baby). Completing this protein balance will deeply enhance our understanding of protein-based nutrition.

Other notes

  • Thanks to E. Noor for helpful discussions for this post.
  • Vitamins are not as fungible with metabolism as proteins, fats, and sugars are. I anticipate a tradeoff between fungibility and how easily its health effects can be understood.
  • Protein will be very important for pregnant and nursing mothers. They must impart protein to their young one(s).

Key words

  • Epidemiological – A scientific approach that seeks to draw conclusions from measured quantities without controlling the pieces involved the study. There is nothing inherently wrong with epidemiological approaches (works well for climate science); however, I suspect that it’s limited when it comes to nutrition.
  • Controlled – A scientific approach where (at least 2) groups are separated and given different treatments.
  • Fungible – A property of an entity meaning it can easily be exchanged with another entity (e.g. molecules in metabolism). Money is another good example of a fungible entity. It can be exchanged for other currencies, goods, property, and numbers on a screen.


Protein per day needed for hair was calculated as follows:

(~100,000 growing hairs) × 1/2 π (50 um radius)^2 × (1 cm/28 days) × (0.9 g protein/1 g hair) (1.3 g/mL) ≈ 0.17 g/day

Density of keratin from [4].

We shed about 2 g of skin per day. I assume 50% of the mass to be roughly protein.


  1. John McDougall, ‘Plant Foods Have a Complete Amino Acid Composition’, Circulation, 105.25 (2002), e197; author reply e197
  2. Elizabeth Brunk and others, ‘Recon3D Enables a Three-Dimensional View of Gene Variation in Human Metabolism’, Nature Biotechnology, 36.3 (2018), 272–81
  3. Avlant Nilsson, Adil Mardinoglu, and Jens Nielsen, ‘Predicting Growth of the Healthy Infant Using a Genome Scale Metabolic Model’, NPJ Systems Biology and Applications, 3 (2017), 3
  4. P Mason. Density and Structure of Alpha-Keratin. Nature. 197.  (1963). 179-180. 10.1038/197179a0.