Comment by jcims | Hacker News Reader

Comment by jcims | Hacker News Reader

Parent Full thread

I think so. Four billion years doesn't seem like enough time to evolve the the sophisticated encoding mechanism of DNA and the level of complexity in individual organisms and the interdependencies they have on each other.

I really wish I could find a sort of statistical breakdown of the odds of this all working purely by chance (I'm definitely a fan of evolution and i'm not necessarily invoking woo or God here, just thinking that there are mechanisms at play we don't understand, and a head start would explain a lot for me)

This is how my hopelessly naive mind thinks about the problem:

Let's round up to say there is 10^15kg of biomass on the planet[1], and there are 10^12 cells per kg[2]. Let's assume that those cells have reproduced every hour for the last 4 billion years (or 35x10^12 hours). That's ~35x10^39 opportunities to pass evolved genetic material to the next generation of cell.

One of the organisms with the simplest known genome is mycoplasma genitalium[3], with ~588 genes expressed over 580,070 base pairs. Each base pair has four possible configurations (A,C,G or T).

I don't know much of anything, and I definitely don't know the rules around how the base pairs come together. I'm going to naively assume that these base pairs can come together any way they like, and that mutations in DNA express themselves as changes in these sequences. This leaves the DNA of mycoplasma gentalium with 4^580070 or 8.7*10^349236 possible configurations of their DNA.

Obviously only a small portion of those are viable. But how do you get from nothing to something of the complexity required to reproduce, and then how do you get from something of that level of simplicity to the genetic diversity that we have on the planet. How many generations from the first living organism are there to current people. Let's say it's a trillion generations. How does random mutation bring you from say a half a million base pairs at the beginning to 3 billion finely tuned base pairs[3] at the end? It just doesn't seem like enough time to me.

1. https://en.wikipedia.org/wiki/Biomass_(ecology)

2. https://en.wikipedia.org/wiki/Cell_(biology)

3. https://www.saylor.org/content/BIO_Kimball/users.rcn.com/jki...

It's definitely not "totally random". DNA (and more importantly, RNA since that's probably where replication started) have their own chemistry and physics - i.e. it's not like toggling bits in a computer, particular combinations act very different to others. In the case of RNA this is notable because RNA sequences can fold and work as enzymes (ribozymes) on their own.

This means you're definitely not throwing random combinations together until something works - there are favored paths all the way down to the base-pair level.

It's worth reading about the RNA world hypothesis (https://en.wikipedia.org/wiki/RNA_world) to get a feel for this idea. It's unproven (and hell, panspermia still might speed everything up since dumping a truckload of ribozymes from the dessicated remnants of another planets biosphere would be a hell of a kickstart) but it's based on different observed properties of these compounds today.

Nature doesn't use arbitrary random sequences of base pairs, any more than we use random sequences of English letters. We can make them, of course, but that's as meaningful as 'bthdkekd'.

I am not sure we know enough about the correlation structure of the DNA code to do the calculation you propose.

In a way it is arbitrary, because anything else would imply meaning or intent. Mutations are, as far as we know, arbitrary.

Your last point is the thing that has stymied me to date. What is the range of mutation that would allow for a successful offspring and convey (mal)adaptive changes.

I use 'arbitrary' in the sense of uniformly random. The point is it's not uniformly random. You can think of reproduction as a random walk in the space of DNA combinations, but it's not a uniform random walk. It has a structure that comes from the biochemical processes which perform the recombinations of DNA. Focusing on mutations ignores this point. (Also, as an aside, it's not settled that mutations occur at random positions. That may be a reasonable starting point, but it's likely that the way DNA folds into nucleosomes/chromatine/chromosomes biases the location of mutations.)

I'm not a biologist, but I'd be surprised that genetic reproduction would be as stable 3 billion years ago than now. I would not be surprised that monocellular beings, which don't need DNA replication to survive, had a more defective cloning system, thus more subject to whichever DNA replication bug.

Plus, life took about 1 billion years to emerge. This is huge. The solar system took about the same time to form. Also, life may be extremely slow to appear, but once it's there, progression is quicker. Since the beginning of phanerozoic eon (and Cambrian explosion), a geologic period (Pleistocene, Jurassic, Cambrian...) only lasts a few tens of millions years. I find it very plausible that at the beginning of Cambrian era, a critical step in evolution was reached, allowing a better and faster evolution.

4 billion years, but also massively parallel. Trillions of simultaneous random walking hill-climbers.

This is one heck of a hillclimb:

https://www.youtube.com/watch?v=aoCzZHcwKxI

Pulling this completely out of my ass, lets say the ability to camoflage is 30 genes, ones to modify skin color, signalling to do it on command, mapping to specific parts of the body to create a specific pattern, all of the above for skin morphology, mental ability to map this to prey avoidance, all of the above for ink synthesis (glands, storage, ability to express on command). Let's further assume that those genes are expressed over 500 base pairs (woefully short) for better or worse.

Somehow those genes need to be introduced into the genome to create a ninja octopus out of whatever its predecessor was. To date, all i've heard is that mutation is the source of this new genetic material. So somehow we need to mutate our way to these 500 base pairs. Let's divide that by ten as some kind of poor man's compensation for the flexibility of dna to handle noise and whatnot.

So we need to introduce 50 essential base pairs. There are 4^50 possible combinations of these base pairs, or 10^29 in more recognizable numbers. Do we have 10^29 generations to introduce those? That's a lot of octopus.

> Somehow those genes need to be introduced into the genome to create a ninja octopus out of whatever its predecessor was.

Well no one is saying that a ninja octopus just appeared one day by chance.

> lets say the ability to camoflage is 30 genes, ones to modify skin color, signalling to do it on command, mapping to specific parts of the body to create a specific pattern, all of the above for skin morphology, mental ability to map this to prey avoidance, all of the above for ink synthesis (glands, storage, ability to express on command). Let's further assume that those genes are expressed over 500 base pairs (woefully short) for better or worse.

Think about the things that you said above.

If one octopus out of the hundreds of millions of generations had one of the 30 gene changes that did one thing to its skin that was beneficial, then in short order (say a few thousand generations) that would be the dominant type of octopus. Then one-by-one the items that you listed could easily do the same. Some individually, others maybe at the same time.

I mean there have only been a little over 100k generations since humans split from apes. So a lot can happen in the time spans you are talking about.

Octopuses as a species are supposedly 296 million years old, hatch by the thousands, and only live 3-5 years, so 100 million generations of thousands of octopuses or so to get to where they are today.

The other issue is that usually seemingly useful and complex systems are not "suddenly" created - i.e. they don't just appear. In living organisms almost every feature is an extremized or repurposed variant of some other system.

Something like a bacterial flagellum for example is actually a special-case configuration of proteins which allow the bacteria to secrete compounds. (see: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4632597/ and other resources - it's really interesting)

So all the individual systems in the octopus camoflage system didn't just have to suddenly appear - they all would trace their function to other systems the octopus already had - i.e. its not "chance of this exact thing appearing" its "chance a mutation doesn't make the old system totally useless, while maybe providing a slight benefit".

Reply on news.ycombinator.com