The Latest Geneticist Bunk About the Human Genome

The popular American English expression or meme "you had one job" or "you had only one job" refers to examples of incompetence when someone has a relatively simple task to do. For example, if you hire someone to paint a wall in your house, and he gets paint all over your floor because he failed to use a drop cloth to protect the floor,  you might cringe at such incompetence and say, "You had only one job!" We can use such an expression to describe the misrepresentations of geneticists in describing DNA and the genes in it. 

Ever since the Human Genome Project was completed around 2001, and the contents of human DNA and genes have been well-mapped,  it's pretty much been something like a "you have one job" situation for geneticists, the scientists who specialize in studying genes and DNA.  That job is for them to correctly inform the public about what DNA and its genes are.  At this simple job our geneticists have failed time and time again. Over and over again geneticists have told us the most outrageous fictions about what DNA and its genes are. 

Some of these fictions are described in a new science paper that itself is guilty of giving us baloney about DNA.  The paper is entitled "The Genomic Code: The genome instantiates a generative model of the organism." The title itself makes a bunk claim. The human genome is not any model of a human body or any of its organs or cells, and the human genome (DNA) is not any specification of how to make a human body. Nor is the genome a "generative model," which is a type of advanced high-tech software utterly unlike the genome. 

The paper starts out fumbling by asking, "How does the genome encode the form of the organism?" There has never been any evidence that the genome encodes the form of the organism. When biologists start out their papers asking questions such as this, they are acting as foolishly as when they start out papers by asking, "How does the brain create the mind?" There is no proof that the brain creates the mind, and there are very many weighty reasons for thinking that the brain cannot possibly be the source of the human mind. Similarly, there are many weighty reasons for thinking that the genome cannot possibly encode the form or structure of an organism. Just as nature never told us that brains make minds, nature never told us that genomes encode the structure of an organism. It was merely scientists who told us that the genome (DNA) encodes the structure of an organism, and scientists who told us that brains make minds. In both cases such scientists were making claims that were never warranted, claims contradicted by many facts. 

The main body of the paper starts out by giving us this very bad example of non sequitur reasoning, stating this:

"Cats have kittens and dogs have puppies. This is despite, in each case, new individuals starting out as a single, undifferentiated cell – the fertilised egg, or zygote. The complex form of the parents is thus not transmitted to their offspring – instead, it must be reconstructed in each new individual. Clearly then, the zygotes of cats and dogs must contain some substance – the genetic material – that is inherited from the parents, which somehow directs the development of the growing organism so as to produce a new token of the feline or canine type. The question is: how? What is the nature of this genomic encoding of form?"

No, the fact that dogs have puppies and cats have kittens absolutely does not provide any warrant for the claim that "the zygotes of cats and dogs must contain some substance – the genetic material – that is inherited from the parents, which somehow directs the development of the growing organism so as to produce a new token of the feline or canine type." Such a claim cannot be justified by any a priori armchair arguments, but could only be justified by the discovery of something in the genome that consists of such incredibly complex instructions. Nothing like any such instructions have ever been found in any genome or DNA or in any genes.  All that exists in DNA (the genome) is low-level chemical information, not high-level information such as how to construct anatomical structures, or how to build cells or any of the organelles that make up cells. 

Before introducing its groundless and silly claim that a genome (DNA) is a type of AI technology called a "generative model," the paper describes how previous claims about the genome (DNA) were untrue. We read this:

"Over the years, various metaphors have been used to conceptualise the nature of what we will call the genomic code – i.e., how the genome specifies the form of an organism (Keller 2020; Nijhout 1990). These include, among others, a codescript (Schrödinger 1944), blueprint (Plomin 2018), program (Keller 1999; Peluffo 2015), recipe (Mitchell 2018), or resource that the developing organism can draw on (Oyama 2000). One of the most enduring of these, especially common in popular science treatments, is the idea that the genome constitutes a 'blueprint'  of the organism (e.g., Plomin 2018). This metaphor conveys the idea of a detailed but miniaturised plan that can in some way be referred to, in order to direct the development or construction of a pre-specified final product."

Noting "the metaphor quickly falls down in several ways," the next paragraph starts to explain why the blueprint metaphor is inappropriate, but does a very bad job of doing so. A much clearer explanation would go something like this:

(1) Blueprints contain images of how the final output of some construction process should look, but nothing like any such images or specifications are found in the genome (DNA). 

(2) Blueprints specify the appearance of particular visible parts and the physical arrangement of such parts within a complex architectural structure, but there is no such specification to be found in the genome (DNA). 

The paper then tells us how claims were made since at least 1961 that the genome (DNA) is a "program" for making a human body. We read this discussion of why such a metaphor is inappropriate:

"The word 'program' also carries some unsupported connotations. Regarding the means by which the steps of development are encoded, 'program' is in modern times also defined as: “a series of coded software instructions to control the operation of a computer or other machine” (Oxford Languages). The usage of the term 'genetic program' may thus seem to imply a regular, explicit, and interpretable set of instructions, logically laid out and executed in series. As with a blueprint, it can be taken as implying a kind of isomorphism, this time between elements of the genome and elements of the developmental program. Moreover, it again suggests a kind of algorithmic determinism, with all the details somehow spelled out in advance. Neither of these properties is observed."

The genome (DNA) is in no sense a program for producing anatomical structures. DNA does not have a series of instructions to be executed to make a human body or any of its organs or cells. 

The paper then discusses the metaphor that the genome (DNA) is a recipe for making a human body, giving us some weak criticism (such as a criticism of vagueness) that fails to explain what a huge lie it always was to claim that DNA is a recipe for making a human. We hear nothing of the main reasons why such a claim is deceptive, such as the fact that recipes are only simple one-page instructions that only specify how to make very simple non-living things such as salads, soups and cakes, with human bodies being states of internally dynamic organization 1,000,000,000,000 times more complex than what is produced by recipes.  The famous biologists who told us that DNA is a recipe for making a human were therefore huge liars telling us a lie as outrageous as the claim that a tricycle is an interstellar starship. 

There are two types of recipes:

(1) A recipe may specify how to make a completely unorganized thing such as a soup, salad or drink. Referring to that type of recipe, the human genome (DNA) cannot be called a recipe for making a human body, as the human body is something enormously organized, with  many levels of very complex organization. 

(2) Other types of recipes may specify exactly how to position things when creating an ordered structure (for example, a recipe for making a 3-layer chocolate cake may specify how to make the layers, how to position the layers, how to separate them with frosting, and how to cover the layers with frosting and put pieces of chocolate and marshmallow on the top). That type of recipe includes specifications of the positions of particular items to achieve the final structure; and the human genome (DNA) has no such specifications, not doing anything to specify either the structure of a human body or the structure of any cell. 

Although the genome might be crudely compared to a chef's shopping list of food ingredients, the genome cannot truthfully be described as a recipe for making a human body or any of its organs or cells. 

Having discussed some of the big fictions that biologists and geneticists have told about DNA, the paper co-authored by a geneticist then gives us the latest in the long series of scientist bunk statements  about DNA: a silly claim that DNA is a "generative model of the organism." DNA is not actually any type of model of a human being, and is certainly not the type of hi-tech software model that is called a "generative model."

The term "generative model" refers to so-called "artificial intelligence" software systems that are able to train on a large body of supplied examples, and provide two-dimensional output like the examples.  So, for example, you might have a "generative model" for making pictures of dogs. You might train the "generative model" by giving it 10,000 photos of dogs, telling the model that those were dogs, as well as 10,000 photos that were not dogs, telling the model that those photos were not photos of dogs. You might then ask the model to generate a picture of dog, and the model might be able to generate a flat two-dimensional image, something looking rather like a dog.  Such a "generative model" typically requires an advanced type of software facility called a "neural net."  Nothing like that exists in the genome or even in any cell of the body. Even the human brain is not at all a "neural net" as that term is used by software designers, because a software "neural net" involves various things and characteristics not found in the brain. 

The paper then gives paragraph after paragraph of gobbledygook trying to draw some similarity between genomes and the "generative models" of AI systems, with the result sounding like someone trying to argue that his tricycle is really like an interstellar spaceship. Don't be fooled by all the digressions and jargon. The human genome bears no resemblance to the "generative models" AI systems use nowadays.

The "generative models" used by AI systems produce outputs such as two-dimensional images that fit on a particular computer screen. They do not produce complex three-dimensional structures. There is not an AI "generative model" in existence that can produce very complex tangible three-dimensional objects after training on examples of such objects. If such AI software is ever developed, it would have to have very complex hi-tech features unlike anything found in the genome (DNA). Today's most widely-used generative models require huge server farms taking up very many times more space than a human body. 

All that we have in the new paper is the latest in a series of bunk statements geneticists have made while describing genomes.  Always be very suspicious when biologists offer any type of analogy, and be particularly suspicion when a biologist offers any analogy regarding either evolution or DNA. There is a very long trail of profoundly misleading analogies made by biologists about DNA, and an equally long trail of profoundly misleading analogies made by biologists about evolution. 

An imaginary conversation

What would we find if we looked at the programming code for some "generative model" used by a software company? We would find code making abundant use of the conveniences and language constructs of modern programming systems: things such as if/then logic, "for" loops, "do while" loops, switch statements, function calls, C++-style classes, instantiation, variable assignment, numerical return values of functions, and so forth. No such things exist in DNA, which consists of simple lists of chemicals. So DNA cannot possibly be a "generative model" like that used in AI systems.  The web page here shows some Python code for a generative model. We see function calls, parameter passing, for loops, if statements, variable assignment, object-oriented classes and various other aspects of software programming unlike anything found in the human genome (DNA and its genes). 

A generative model is a type of computer program, as shown in the link above, which lists the code for a Python computer program that is a generative model. Very strangely, the paper authors have first spoken against the claim that the genome is a program, and have then given a type of computer program as what they think is a good description of the genome. That's as goofy as someone saying, "Of course, my skateboard isn't an automobile; it's a Cadillac," as if he didn't know that a Cadillac is a type of automobile. 

The outputs of "generative models" like those used to make the image below are static two-dimensional things such as images. A human body is something vastly different: a three-dimensional structure in which there occurs constantly the most complex fine-tuned dynamic activity. The ongoing marvels of fantastically complicated human biochemistry and the continual construction of protein complexes and vastly complicated cells are some of the reasons why a human body is a far more complex and fine-tuned reality than a 747 jet.  A corporation knows how to construct 747 jets, but there is not a nation or corporation in the world that could construct a living human body from raw materials.   "Generative models" like those used in AI systems typically produce their outputs instantly, which bears no  resemblance to the construction of a human body, which requires nine months and then continues for years as a baby grows to an adult.  

Two of the words most misused by scientists are the words "plausible" and "model." Scientists very frequently use the word "plausible" to describe scenarios ranging from likely to unlikely to all-but-impossible to downright impossible. The word "plausible" is used so carelessly and "at the drop of a hat" in scientific papers that it means basically nothing when a scientific paper calls something "plausible." Scientific papers misuse the term "model" just as frequently. The term is very often used for the vaguest hand-waving, to try and make vacuous verbiage sound more substantial. 

The dictionary defines a model as "a three-dimensional representation of a person or thing or of a proposed structure, typically on a smaller scale than the original." The genome is no such model of a human body or any of its cells or organelles or protein complexes. There are also mathematical models, which are centered around equations. The genome doesn't have any equations, and does not have any math. The genome is not a model of the human body or any of its organs or cells. DNA specifies which amino acids make up a protein, but does not even specify the complex three-dimensional shape of a protein molecule. How such shapes arise is the unsolved mystery called the protein folding problem. DNA is an inert molecule containing only very low-level chemical information, so spinning grand fictions about it is rather like saying your broomstick is a transatlantic crossing device. 

Why do geneticists keep making glaring misstatements about DNA? It's largely because they don't want to face the shocking truth: that a speck-sized human zygote somehow progresses to become the enormous state of organization and vastly fine-tuned dynamism that is the human body, without having any set of bodily instructions for how to organize anything bigger than a protein molecule. If they realized this shocking truth, they might be on the doorstep of understanding the folly of all bottom-up explanations for the origin of human bodies and human minds. 

Postscript: Two other scientists say this:

"We see no valid use for definitions of the genotype

and phenotype in terms of blueprints, programs, or

sets of instructions, and their realizations or

manifestation....The program/manifestation metaphor is factually misleading, because it suggests that the genotype uniquely determines an organism’s phenotype. However, as is well known, all it does is specify an organism’s norm of reaction to environmental conditions (Rieger et al., 1991, Lewontin, 1992)."