I wonder if saying that "it's all encoded in DNA (which is only 6 billion bits)" is accurate. Every time life (considering only humans for now, to simplify) arises, in addition to the DNA (the "code"), you also have the rest of the first cell (the zygote), which already has an "expression" of that code (which may contain additional information beyond what is already in the code), + the entire environment of the mother (which obviously provides additional information).
This makes me think that we need far more than just 6 billion bits to actually encode the entire process of new life formation (each instance always piggy-backs off the previous, and always provides processes/information).
He is studying exactly this. He found that when developing, cells move and grow in a particular direction. But when an obstacle is placed in their way, they move around it to somehow still end up where they’re supposed to.
DNA is not all of the encoding. Even cells appear to have some form of navigation or space search capabilities.
I imagine from an evolutionary perspective, a false positive of the fetus bringing harm to the mother is preferable over the long term for increasing long term reproductive survival.
Miscarriages must occur because our DNA is constantly mutating. Some of those mutations will be so harmful that the new organism cannot even survive until birth.
In addition to those unavoidable ones, there can also be things going wrong by accident (e.g. the pregnant woman suffer physical injury which kills the fetus).
for example (pre)eclampsia. we still don't know WTF is going on exactly (but it's basically abnormal blood vessel formation between the fetus and the uterus), and a few decades ago it was basically guaranteed loss of the fetus or the mother (or both), in about half a percent of pregnancies.
nowadays thanks to medical science it's a hundred-times more manageable.
Michael Levin has some awesome experiments where he screws around with the bioelectric fields of planetaria and he gets them to grow two heads and no tail, while being genetically identical to a normal one.
One of my old classmates' area of research is applying electric stimuli to various types of cells. Apparently the right amount at the right time can briefly allow foreign material to pass through the cell wall unimpeded. The complexity of a single cell is insane, that of a complex organism is way beyond insane.
I think you meant cell membrane rather than cell wall?
(There is a contrarian theory that says that the cell membrane does not exist at all, look up Gilbert Ling. So it's no surprise that things can move in and out of the cell under certain conditions.)
Planaria. (Yeah, I am a pedant sometimes.) Also, frogs, tadpoles, axolotls. Truly a wonderful mind. I admire him, not least for his ability to do really weird and important science outside the usual "don't do anything extraordinary" boundaries set by ossified grant committees.
We need purely functional nanomachinery to create every embrionic molecule from element atoms, and a declarative physical language to combine them in the right way to create a zygote.
What if, as human consciousness is an emergent property of complexity, so too are humans and DNA itself? Could it be that, just as consciousness arises from the intricate networks within the brain, life and its fundamental components—like DNA—emerge from an even deeper layer of complexity in the universe, suggesting that our very existence is a product of systems far beyond our current understanding?
I'm not really sure what you mean. Complex biological molecules are emergent from the laws of physics and an environment containing the right precursors.
We do, but it’s messy and usually involves a few years of courtship, followed by one to three decades of raising & educating the resulting human to anything resembling a useful model for comparison.
No, the claim is much stronger than that: that the zygote itself encodes necessary information that isn’t captured in DNA. Put another way, long after humans are extinct, the claim is that if aliens could download our DNA source code somehow, they still wouldn’t be able to build humans without replicating the zygote’s internal structures exactly. We’d come out as super deformed or something.
Perhaps the first generation would be deformed but second-gen (if they hypothetically get that far) should be much closer. You still need the mitochondrial DNA as well.
Well, first of all, they wouldn't be able to build a single functioning cell if all they had was DNA (and even if they had mitochondrial DNA too). Most organelles divide independently of the nucleus, and there is no reason to think that DNA encodes anything about their fundamental structure. Even if some changes in genes can effect some changes in the organelles, that doesn't mean that the genes specify every detail of the organelle.
Also, even if we accepted that DNA fully specifies how a cell can create an identical copy of the cell that contains it, that doesn't mean that it specifies how to create a cell from scratch. The "instructions" in DNA could very well depend critically on details of the current cell. For example, the DNA could specify se thing like "take 1% of the substance secreted in organelle A and mix it with 90% water and 9% the substance secreted by organelle B". This instruction is perfectly good for specifying a copy of the current cell, and perfectly useless if you don't have the original cell for which it is meant.
This sort of thing could very well apply at the level of the whole fetus. Details of the uterus and other parts of the mother organism may well be critical parts of the "program" described by the DNA. For example, it's easy to imagine that the early fetus follows instructions like "let this much fluid pass through the umbilical chord", or "grow horizontally until you find this much pH difference between the extremeties" or whatever other instructions that are only useful in the context of an existing functioning mother organism.
And even beyond the individual, you would have a big problem recreating the species to allow for a second generation to exist at all. In particular, even if you had a whole living healthy female mammal, you would have no information at all for how to create a male of the species, so no way to create sperm cells, so no way to perpetuate the species. So the DNA of a female mammal doesn't contain information for how to make more of the species. And if all you had was a male organism, you would lack the information probably encoded in the living female that I was discussing earlier.
As a side note, this problem would not exist for birds, where the female bird does have both male and female DNA.
Not only that, but you're not even given the compiler. If all you have is the DNA (even if you had the mitochondrial DNA too), you have source code for a language no one knows with no compiler.
Or, more accurately, it's like having binary machine code for an unknown ISA with no information about the CPU, and no example CPU.
Well based on compiler bootstrapping techniques, it seems the solution is simple: Start by hand-assembling the most basic RNA life, then run each step of evolution in sequence. All you need is the DNA (and know the living conditions) of every single organism between single-celled organisms and a modern human.
Mathematically it's true. Anything that can be enumerated (that is, has an ontological mapping with the set of whole numbers) can be interpreted as input to some function which can be evaluated to some result also with an ontological mapping with the set of whole numbers. In other words, the entire human population could be a part of some program calculating the answer to some universal question, which may be about life, the universe, and everything. In fact, the answer may already be known. We're just one big server with massively parallel operation.
What we don't know is what that function is. Put differently, what is the question?
I think they meant something even more basic, and common to all cells: you can't make a new cell without an existing living functioning cell. We have no firm idea of how the first cells developed, but all of the cells we know of today come from the division of another cell, and we have never observed anything remotely like a new cell forming pit of a puddle of all the required substances and DNA.
Yes, I recently read the book The Master Builder by Alfonso Arias which gets into some of the detailed pre-requisites provided by existing cells. (it’s also why restoring extinct species may be harder than expected even knowing their full genetic sequence)
Couldn't this be proven by growing mammals in petri dishes?
If the entire information is encoded in DNA, then the fetus would grow fine.
If however, the fetus needs to be in a womb to develop, then there must be information (and resources) that need to be sent from the mother to the baby in a sort of quine fashion.
Consider a Java compiler or python interpreter, they are themselves written in the target language, unless we are talking about early version. So you already need a java compiler to compile Java, the code for the compiler does not hold all of the necessary information, as you need a mother compiler to execute it.
Which part of the target is in the code and which in the interpreter/compiler?
A thought experiment would be a language where a built in statement compile() compiles code according to specifications. Thus the code for a compiler would simply be compile().
My best guess is that fetuses can only grow inside a functioning human, and that the information in DNA is not sufficient, you need a functioning specimen to grow another specimen.
It's unclear how much those non-genomic components affect organismal phenotypes; presumably, there is some minimal collection of factors that would suffice to allow us to construct an artificial egg cell that is viable for reproduction.
This is a very interesting observation. I wonder if there's a way to identify the contribution of that initial cellular environment to the phenotype. I guess you could do experiments where you tamper with the initial cellular environment or the DNA and see what happens.
A 27-state two value Busy Beaver can implement "Give a counter-example to Goldbach's Conjecture" - if there are none the program never halts. We can recognise that, and say "Oh dear" but it gets us no closer to an answer. Six billion isn't even in the right ballpark, our ability to understand the meaning of every self-acting machine extends maybe to six or seven bits if we're very smart.
I wonder if it's actually more similar to procedural generation, based on a seed value, that reliably recreates the same complex output that isn't entirely described by the seed alone. In this case, the DNA would just serve as input to a myriad of algorithms that yield a fully working cell. A single base pair change may lead to a mind-bogglingly long chain of side effects that result in a third arm, or really just nothing at all.
> somehow it's all encoded in DNA (only 6 billion bits in humans, which has to also build cells and their ultracomplex machinery)
One has to be cautious to not assume that the only data DNA holds is in base pairs.
We know that secondary factors like inter- and intra-strand interactions or DNA-histone interaction encodes information as well.
A good analogy is a book. While it is made up of individual letters, the combinations build to words, which build to sentences, which build to ideas, which can interact with other ideas (sometimes in other chapters!) to create a complexity beyond just the combination of letters would suggest.
That's not even getting into the temporal aspects of DNA expression.
I don’t think it’s true that DNA actually encodes an entire organism. Leaving all discussion of epigenetics aside, the proteins produced from DNA can basically do anything, including modifying DNA. Maybe we could think of large portions of DNA as simply bindings to shared libraries (i.e, the proteins). Or like a package manifest and build script for life rather than the code itself
A crappy, self-improving function with a decent feedback loop (survival) can get pretty far given 100s of millions of years with hundreds of million (or trillions) of concurrent experiments running at the same time. The time scale of life is the hardest element to comprehend.
* all the structures grow from a single cell
* somehow it's all encoded in DNA (only 6 billion bits in humans, which has to also build cells and their ultracomplex machinery)
* this all had to evolve using only a very noisy objective function