Hacker News Reader

Comment by fktnktofk | Hacker News Reader

Aren't those electrical field patterns still somehow encoded in the DNA?

They don't need to be.

I'm not sure what the conclusion of biologists is today, but I have a feeling a lot of important mechanisms aren't encoded in the DNA - or, at least, not anymore. The key insight here is that cell reproduction doesn't happen out of nothing, it's always an existing mechanism working on a new mechanism.

The seminal paper of Ken Thompson, "Reflections on Trusting Trust" (1984)[0], demonstrates how you can encode a backdoor in a compiler in such a way that the compiler embeds this backdoor in every program it builds, and embeds the backdoor-embedding machinery in every new compiler binary it builds. You can use it to build a "vanilla" compiler and make it backdoor-embedding, even though the source code doesn't have backdoor-embedding instructions in it.

I suspect similar things happen in biology - there is computationally relevant information embedded in the runtime state, that's not explicitly encoded in the DNA - but it carries on, because every new living cell is built by a previous living cell.

[0] - https://www.cs.cmu.edu/~rdriley/487/papers/Thompson_1984_Ref...

The issue is you’re going to get massive mutations building up in anything that doesn’t have the significant error correcting capability of DNA. So while the coding is not direct, it still has to somehow be in DNA so that the trait will be conserved over the generations. Shorter term changes could conceivably be stored in other molecular structures, though. (Think patterns of methylation in epigenetics or even the fringe idea of RNA memory.)

I would suspect really basic things like replication would be stored somehow in DNA (nuclear DNA, mitochondrial DNA, etc) as they must be highly conserved and mutation resistant in order for organisms to survive and reproduce.

I get your point and I agree, the DNA runs on hardware, which is why you can't take cat DNA and put it in a salamander cell and expect to get a cat. The resulting cell will just die since the hardware is not compatible with the DNA.

My point was that there is only so much information capacity in the cell if you exclude DNA, surely much less than needed to store electrical fields patterns for a whole embryo.

A cell has a lot of atoms, which may store a lot of information?

But not error corrected like DNA.

https://en.wikipedia.org/wiki/Tuatara

https://asunow.asu.edu/20200806-dragon-dna-sequencing-genome...

https://phys.org/news/2020-08-dinosaur-relative-genome-linke...

I really recommend Anton's segment on the Tuatara, https://www.youtube.com/watch?v=PWywZyzmBDE MindBlown!

Can single atom store more information than a bit?

DNA is made up of 4 main components (G,C,A and T). Very generally these encode instructions for making proteins.

Those blocks are very small, less than 20 Atoms.

https://en.m.wikipedia.org/wiki/Adenine

They’re working on using it for storage. (very slow read write times). Duplicating shouldn’t be a problem.

https://en.m.wikipedia.org/wiki/DNA_digital_data_storage

Each electron present in one of the atom's orbitals is a bit, so yeah.

More realistically, the simplest atom- one proton, one electron- has 3 integer numbers associated with the state of the electron. You need more than a bit to describe the state of that system, so yeah.

Computationally relevant information is sometimes embedded in runtime state (for example, yeast prions), but this is very rare. For small bacteria, you can switch out the entire genome with a new genome and have them continue living just fine. For all intents and purposes on an engineering level, everything is/can be encoded in DNA.

Speculating for fun based on what scientist people have told me at parties :) I think the idea is that DNA will directly and literally encode things like "the genes for proteins A and B tend to get activated together". But the next thing that happens is that protein B catalyzes some chain of chemical reactions that winds up activating a different gene for protein C. And then maybe protein C causes some sort of mechanical change in the cell, and the resulting mechanical stress triggers the production of protein D in adjacent cells. Mapping the mechanical relationships between cells in an animal with developing bones and muscles gets super complicated. So the question of "does the DNA encode the relationship between A and D" ends up depending on what you mean by "encode".

Everything is deterministic if you're omniscient.

In our more limited experience, certain levels of complexity mean unpredictable outcomes.

As for why-not-DNA, I think about it in terms of "If there were some way that was faster able to adapt organisms than DNA changes, wouldn't most organisms accumulate substantial functionality via that pathway over time?"

> Everything is deterministic if you're omniscient

Is this true, given https://en.wikipedia.org/wiki/Uncertainty_principle ?

It is indeed not true, even omniscient beings can only produce probability distributions.

However it is not clear if quantum effects are relevant at a biomolecular scale. Some experts[1] suggest so but these ideas are considered fringe by the mainstream.

1. Sir Roger Penrose

I'm of the mind that randomness is only an approximation of complexity, if you're omniscient then randomness doesn't exist and everything is deterministic.

However it seems to me the deterministic vs randomness debate is of the same order as the existence of god(s) : we will never be able to prove or disprove it and we aren't able to make useful predictions out of the information one way or the other so we might as well just agree to disagree and move on.

DNA is basically digital information. Every rung on the helix ladder is either an AT or GC. Nature found the simplest say to store information in our chemistry set. Couldn't really store any field patterns though.

Reply on news.ycombinator.com