The world is complex. There are thousands of basic foods that are combined, processed and cooked in countless ways. You can't figure out which of these are good or bad individually. You have to make categories to simplify the problem.
There are dozens of categorizations that are studied in literature. Vegan/Veg/Carnivore is one. Each diet becomes its own category to be studied, such as keto.
Level of processing is another such categorization. It won't give you the final answer on what is good or bad, only a partial answer, that will indeed have to be combined with results of other categorizations to get a final answer.
Please understand that nutritional science is a few decades old at this point. Demanding physics (a 500 year+ academic discipline) level of clarity from a discipline which is many times more complex is not going to happen in our lifetimes. The only thing we are going to get are bad partial answers; just like you point out. Everyone knows they are partial answers, but they are partial answers because of complexity, not because of malice.
My problem with picking ultra-processed as a food category is that it has very little to do with the nutritional value. And where we decide the starting line is for that processing is also arbitrary, as far as the result goes.
I mean we could say that beef is ultra-processed grass. And lettuce, ultra-processed sunlight.
Our biological ancestors evolved over 10s-100s of millions of years to consume raw plant matter and raw meat. Only in the last 100k years or so did we start consuming cooked food, which gives some time for evolution to adjust our genes. Our bodies are wet chemical machines that we are confident can consume such food.
These are the unprocessed and moderately unprocessed foods in the NOVA categorization. Processed foods is something that began to be consumed a few thousand years ago (eg. Tofu is 2k+ years old in China), and ultra-processed foods, as defined above, became largely consumed a few decades ago. So there are very good reasons to define these categories as they are. The lines dividing them are fuzzy but that is how all non-physics sciences work.
> ultra-processed as a food category is that it has very little to do with the nutritional value.
Now, if you have a complex system like the human body, it's reasonable to say that the more you drive it outside the domain that it was designed and tested in, the more risk you are taking on the something bad will happen. So a priori you would be mindful of consuming ultra-processed foods. It is a testament to biological evolution that nothing radically bad happens when we consume ultra-processed foods. But it seems very reasonable to investigate such foods more carefully for smaller but long term impacts on health.
> little to do with the nutritional value.
The digestive system doesn't just care about raw nutrition - the number of proteins/fats/carbs - of the incoming food. There are a number of physical and chemical processes occur during digestion, and these processes can be sensitive to what combination of things go in and in what physical form, etc.
It's based on fear of change -- that something bad might be in newer foods. Taking this advice to heart would mean limiting what we eat to just a few well-known, well-tested foods.
And yet the common best-practice nutritional advice is to eat a wide variety of foods. At the same time we know of plenty of natural raw foods, and even some simple cooking processes, that are harmful.
I'd actually advocate for both practices. It's how animals survive best: some members of a species are risk takers and others are cautious, so that one or the other is likely to survive any change in environment.
There are dozens of categorizations that are studied in literature. Vegan/Veg/Carnivore is one. Each diet becomes its own category to be studied, such as keto.
Level of processing is another such categorization. It won't give you the final answer on what is good or bad, only a partial answer, that will indeed have to be combined with results of other categorizations to get a final answer.
Please understand that nutritional science is a few decades old at this point. Demanding physics (a 500 year+ academic discipline) level of clarity from a discipline which is many times more complex is not going to happen in our lifetimes. The only thing we are going to get are bad partial answers; just like you point out. Everyone knows they are partial answers, but they are partial answers because of complexity, not because of malice.