Start watching the complexity of energy management in a cell or how the immune system works or how cancer cells implement countermeasures to bypass all of the evolved mechanisms against it (yet remain viable). It's beyond ridiculous
Honestly I just struggle with the overall statistical likelihood of life evolving. I've tried to get folks to engage on my thinking here but it never happens. I don't know where I'm off.
One example. The number of interactions per second in water is roughly the mean molecular velocity divided by the mean free path times the number of molecules. For a mole of water (~18g) that's approximately 590 m/s * 1/3.1e-10m * 6.02e23 or ~1.15e36 interactions per second. Per kg of water that's ~1.15e36 * (1000/18) = ~6.4e37. Given that most chemical reactions related to the formation of life are going to happen in solution in water, this seems like a reasonable place to start on this journey.
Expanded across all of the water and entire biomass of the earth (in kg) for 5 billion years, this is 1.15e36/kgs * (1.4e18kg + 6e14kg) * (360024365*5e9)s = 2.5e71 potentially life-forming molecular interactions in water on planet earth's history. We're just spitballing here so let's square that to account for any bullshit that I missed.
6.25e142 'interesting' molecular interactions on earth since it was a sphere of magma.
I know this is broken but just roll with the idea for a second.
Given 6.25e142 molecular interactions total, let's assume that's the upper bound of base pairs of DNA stacking on each other (reality is obviously many many many order of magnitudes less). A+C+A+G, then A+C+A+T, then A+C+T+A, etc etc. Let's further assume that the absolute minimum number of base pairs required for a critter to reproduce is 1/10 the smallest observed genome. The smallest i can find is Carsonella ruddi with 160k base pairs. Let's split that down to 16k base pairs just for shits and giggles and assume that most of it can be random and still result in a living reproducing organism. The only constraint i'm going to ask for is that there are one or more critical sections that need to be reasonably correct for the DNA to be viable.
The question is what is the likelyhood that all of these interactions are going to create one of those critical sections?
There are four nucleotides that make up dna, commonly labled as A, C, G and T. The possible combinations of these nucleotides grow exponentially with the length of the sequence. 4 base pairs have 4^4 combinations, 10 base pairs have 4^10 combinations. etc etc. What if we assume perfectly uniform distribution of random variations (obviously broken), what's the maximum length of 'critical section' that we can guarantee will get at least one shot at life.
log₄(6.25e140)=237
237 nucleotides is the limit to where our mythical mix master could explore all possible combinations and any one of those has a 100% chance of being realized. (Keep in mind that this is individual nucleotides, not genes. A single gene typically starts at 1000 nucleotides, our 'most simple organism' Carsonella ruddi has 160,000 base pairs, and the human genome has ~3 billion base pairs.)
What this tells me is that you cannot have a 'critical section' in ANY gene or DNA sequence that is more than 237 nucleotides long and attribute the source of that sequence as some kind of statistical certainty...from there it starts to become a function of chance. The degree to which it is chance depends on what kind of sequence you need to find to get life moving. (ANd this is completely ignoring the fact that DNA is not living, it needs all sorts of mechanisms around it to actually function) This is all true also if you limit the scope of your investigation to the planet earth.
If you expand it to all of the atoms across the entire universe assuming it's one giant blob of water, just expand the middle term from ~1.5e18kg to 1.5e53kg and the length...
The weakness may be down to your assumption that the simplest form of life then is equivalent to the simplest form of life you personally can find now.
. Life has to do a lot of things now that maybe the first reproducing cells needn't, like defending itself from other life.
. There may be simpler forms of life that exist currently that we are unaware of...
. ...or maybe not, because each time it gets created it gets eaten by much more advanced life around it.
. We don't even know what life really is.
etc. etc. I get the feeling you're perhaps paving the way for a theistic argument. If you don't believe that god the creator exists, the proof that life is possible regardless of your stats is all around you.
Thanks for the reply. I definitely feel that my idea of where it started is polluted by where we are now. However, the minimum viable product for evolution is essentially 'trait inheriting chemical structure' and even the simplest examples that we have of that today are incredibly complex.
>I get the feeling you're perhaps paving the way for a theistic argument. If you don't believe that god the creator exists, the proof that life is possible regardless of your stats is all around you
Nope. It's really not. There's no way for me to prove it, and I'm glad you said it out loud so it could be addressed, but it's really not. This honestly came to me out of my career in information security and tangents into encryption. If you think of those 'critical sections' as passwords or encryption keys to life, our current thinking just seems to be totally fucked by impossibly small probabilities.
The place where the ball settles in the saddle for me is panspermia OR that there is something that we don't quite understand in general about emergent phenomena that drives 'chance' in a certain direction.
Read Nick Lane’s The Vital Question. He goes into a somewhat related line of thinking about energy transfer in hydrothermal vents and what could have led to primitive structures that lead to life.
Will do, thank you. At this point I just set it aside because there's no real way for me to answer the question and there seems to be even less that it will do to drive my decisions, but it's still there.
The evidence that life started almost as soon as it was physically possible indicates to me that there's probably some kind of strong inevitability behind it. Here's an interesting theory about emergence of life being driven by thermodynamics. [0] In short, physical systems naturally evolve in a way that maximizes dissipation of energy, and life is really good at dissipating energy.
Genes probably came after metabolism in my view. Certainly, a "critical gene sequence" is no good without the other chemical infrastructure to make use of it, in much the same way code is no use without a runtime environment. Yes, RNA can also form enzymes, but it's not really the same thing because in that role they don't have to fully define a life form; it can just be part of a life-like mess. I also like droplet/membrane-first theories of abiogenesis [1].
My guess is a lot of stuff co-evolved before "cells" were really a firm concept. Proto-life would have just barely worked at first, hardly distinct from the environment, with a lot of useless or counterproductive material along for the ride. "Living" systems would have replicated by simple physical phenomena (like the droplet division in [1]), with genes taking over as emerging competition justified the complexity.
Side note: probably one of the reasons, if not the main one, that people don't want to engage on the improbability of abiogenesis is that it smells like Creationism.
Great points, thank you. I'm going to steal the runtime analogy, because it very succinctly expresses (at least to a given audience) an idea that I struggle to get out concisely.
Also.
>Side note: probably one of the reasons, if not the main one, that people don't want to engage on the improbability of abiogenesis is that it smells like Creationism.
I totally agree and it's also super frustrating. Will take a look at the links you provided, thanks!
> Again i'm sure something is wrong with my thinking. But my mental model is that we are operating somewhere along an infinite stream of events of trillions of universes and we are unbelievably special/unique. OR the universe is much more bizarre than we even remotely understand.
If you were to apply that napkin math to a petri dish solution of bacteria, you'd come up with a disappointing small number of possible evolutionary 'outcomes'.
Yet, in a few hours, with the right environmental pressure, a petri dish will happily evolve a strain of, say, antibiotic-resistant bacteria.
Edit - Actually, maybe not. Let me give this a shot - just to do napkin math again.
1 Staphylococcus aureus is ~1um in diameter or ~5e-15 liters in volume. They reproduce let's say once per hour, so if you had a liter of them after an hour you would have (1 / 5e-15) * 2^24 or 3.3e21 reproduction events.
Let's say immunity is transferred at a specific spot in the DNA, first three nucleotides in gene X are flipped to ACT. Spontaneous DNA replication errors happen roughly between every 10e5 and 10e9 replication events.
Staphylococcus aureus has ~2.8M base pairs and we need to set a specific three of them to ACT. My stats are a little rusty but the likelihood of any error happening on the low end is 1:10e5. The likelihood that this error occurs in the right spot while still leaving a viable organism has to be greater than 1:2.8e6 but not 1:1 or you're just spitting out random shit. Let's cut it to 1:1e5. The likelihood that the output of three base pairs being ACT under random chance is 3^4 or 81. All of these together i'm getting ~8e12. So i'm being pretty generous here but the likelihood that you'll get a resistant strain would then be ~1:8e12. With 3.3e21 replication events that gives you almost 2 billion resistant bacteria over 24 hours.
Moving to more reasonable values you start with 10µL of staph, spontaneous replication errors every ten million splits and let's split the chance of the mutation being in the right area to 5e5. You end up with 3.3e16 replication events and roughly 1:4e15 chance of getting a resistant mutation or approximately 20 resistant bacteria after 24 hours.
The book What is life? How chemistry becomes biology by Addy Pross is an interesting read. Here is an article of the same author in Aeon that I just found:
Summary of the idea: there are self-replicating chemical systems (e.g. start with a single type of molecule that shows autocatalysis) which will undergo variation (replication mistakes / 'mutations'). These will have to compete with the original systems and might take over if they show a higher replication rate (exponential growth!) in the environment that they are in, i.e. there will be selection and different niches. Complex systems seem to be better suited for this according to him as there seems to be more leeway to work in non-ideal conditions. So there is a sort of darwinian selection process that leads to complexification which in turn leads to what we consider as life.
It does not claim to explain how life on Earth historically came to be but why life might be sort of inevitable.
Your assumption that the start of life was like what we see today is the problem.
We know that RNA molecules form spontaneously, and many such molecules exhibit enzyme activity. We know that an astonishingly short molecule suffices to copy other RNA sequences it encounters.
All we need, really, is for two of those to encounter one another in the right conditions, and in short order the seas are filled with really excellent copiers. (Better copiers, as they arise, reproduce faster, and soon dominate.) So, the starting condition of life is an environment brimming with RNA replicators, and numerous copies of whatever aids them. Any sort of encapsulation event is likely to include one or more replicators.
Encapsulation is necessary to preserve improvements in interaction between molecules, so we need spontaneous closed-membrane formation.
Curiously, there is no known instance I know of where a cell produces a membrane de novo. It always starts with what it got from a mother cell. Arguably it is not the nucleic acids that define life, but the membrane.
Sure there are many plausible series of events that lead from a dead planet to one shimmering with life. My main question isn't if it's possible...we have direct evidence that it is. My question is more related to the probability and how much of a 'happy path' that our little spot in the universe had to have in order to lead us to where we are today. Going back to the origin of my question, it's easy to say 'all we need is the right key and we can decrypt this file' and wave away the fact that all of the atoms in the universe could be recruited for the cause and not have a chance in hell to guess the correct key in all known time.
describes an RNA enzyme (I count 63 base pairs) that replicates itself, in the absence of other things to replicate, so you don't even need two of them, to start with.
The point is that we don't need to assume spontaneous conjunction of all the particles that make a complete system. We should instead guess that before there was what we would recognize as life, many of the constituent parts, at a scale much larger than atoms or amino acids, had arisen, many a result of pre-life analogs of living processes such as replication, driven to high populations by pre-life statistical analogs of natural selection.
Then life becomes not a miraculous coincidence, but an inevitability.
The thing I wonder about, is that presumable we can control temperature, pressure, magnetism, electric currents, elements and pretty much everything we know and understand in the physical to a relatively fine degree. We can simulate most environments that have existed on the planet over its existence, and yet with all that control, we cannot create even the simplest of life. We can manipulate it, alter it, destroy it - but we cannot yet create it. So either we don’t understand things nearly as well as we think we do, or there are some things that we don’t even know about yet.
17 comments
[ 3.6 ms ] story [ 31.6 ms ] threadIt truly is humbling.
Honestly I just struggle with the overall statistical likelihood of life evolving. I've tried to get folks to engage on my thinking here but it never happens. I don't know where I'm off.
One example. The number of interactions per second in water is roughly the mean molecular velocity divided by the mean free path times the number of molecules. For a mole of water (~18g) that's approximately 590 m/s * 1/3.1e-10m * 6.02e23 or ~1.15e36 interactions per second. Per kg of water that's ~1.15e36 * (1000/18) = ~6.4e37. Given that most chemical reactions related to the formation of life are going to happen in solution in water, this seems like a reasonable place to start on this journey.
Expanded across all of the water and entire biomass of the earth (in kg) for 5 billion years, this is 1.15e36/kgs * (1.4e18kg + 6e14kg) * (360024365*5e9)s = 2.5e71 potentially life-forming molecular interactions in water on planet earth's history. We're just spitballing here so let's square that to account for any bullshit that I missed.
6.25e142 'interesting' molecular interactions on earth since it was a sphere of magma.
I know this is broken but just roll with the idea for a second.
Given 6.25e142 molecular interactions total, let's assume that's the upper bound of base pairs of DNA stacking on each other (reality is obviously many many many order of magnitudes less). A+C+A+G, then A+C+A+T, then A+C+T+A, etc etc. Let's further assume that the absolute minimum number of base pairs required for a critter to reproduce is 1/10 the smallest observed genome. The smallest i can find is Carsonella ruddi with 160k base pairs. Let's split that down to 16k base pairs just for shits and giggles and assume that most of it can be random and still result in a living reproducing organism. The only constraint i'm going to ask for is that there are one or more critical sections that need to be reasonably correct for the DNA to be viable.
The question is what is the likelyhood that all of these interactions are going to create one of those critical sections?
There are four nucleotides that make up dna, commonly labled as A, C, G and T. The possible combinations of these nucleotides grow exponentially with the length of the sequence. 4 base pairs have 4^4 combinations, 10 base pairs have 4^10 combinations. etc etc. What if we assume perfectly uniform distribution of random variations (obviously broken), what's the maximum length of 'critical section' that we can guarantee will get at least one shot at life.
log₄(6.25e140)=237
237 nucleotides is the limit to where our mythical mix master could explore all possible combinations and any one of those has a 100% chance of being realized. (Keep in mind that this is individual nucleotides, not genes. A single gene typically starts at 1000 nucleotides, our 'most simple organism' Carsonella ruddi has 160,000 base pairs, and the human genome has ~3 billion base pairs.)
What this tells me is that you cannot have a 'critical section' in ANY gene or DNA sequence that is more than 237 nucleotides long and attribute the source of that sequence as some kind of statistical certainty...from there it starts to become a function of chance. The degree to which it is chance depends on what kind of sequence you need to find to get life moving. (ANd this is completely ignoring the fact that DNA is not living, it needs all sorts of mechanisms around it to actually function) This is all true also if you limit the scope of your investigation to the planet earth.
If you expand it to all of the atoms across the entire universe assuming it's one giant blob of water, just expand the middle term from ~1.5e18kg to 1.5e53kg and the length...
. Life has to do a lot of things now that maybe the first reproducing cells needn't, like defending itself from other life.
. There may be simpler forms of life that exist currently that we are unaware of...
. ...or maybe not, because each time it gets created it gets eaten by much more advanced life around it.
. We don't even know what life really is.
etc. etc. I get the feeling you're perhaps paving the way for a theistic argument. If you don't believe that god the creator exists, the proof that life is possible regardless of your stats is all around you.
(None of my arguments are novel)
>I get the feeling you're perhaps paving the way for a theistic argument. If you don't believe that god the creator exists, the proof that life is possible regardless of your stats is all around you
Nope. It's really not. There's no way for me to prove it, and I'm glad you said it out loud so it could be addressed, but it's really not. This honestly came to me out of my career in information security and tangents into encryption. If you think of those 'critical sections' as passwords or encryption keys to life, our current thinking just seems to be totally fucked by impossibly small probabilities.
The place where the ball settles in the saddle for me is panspermia OR that there is something that we don't quite understand in general about emergent phenomena that drives 'chance' in a certain direction.
Genes probably came after metabolism in my view. Certainly, a "critical gene sequence" is no good without the other chemical infrastructure to make use of it, in much the same way code is no use without a runtime environment. Yes, RNA can also form enzymes, but it's not really the same thing because in that role they don't have to fully define a life form; it can just be part of a life-like mess. I also like droplet/membrane-first theories of abiogenesis [1].
My guess is a lot of stuff co-evolved before "cells" were really a firm concept. Proto-life would have just barely worked at first, hardly distinct from the environment, with a lot of useless or counterproductive material along for the ride. "Living" systems would have replicated by simple physical phenomena (like the droplet division in [1]), with genes taking over as emerging competition justified the complexity.
Side note: probably one of the reasons, if not the main one, that people don't want to engage on the improbability of abiogenesis is that it smells like Creationism.
[0] https://www.quantamagazine.org/first-support-for-a-physics-t...
[1] https://www.quantamagazine.org/dividing-droplets-could-expla... (edit: previously https://www.wired.com/2017/01/life-begin-dividing-droplets-h... which is a reprint)
Also.
>Side note: probably one of the reasons, if not the main one, that people don't want to engage on the improbability of abiogenesis is that it smells like Creationism.
I totally agree and it's also super frustrating. Will take a look at the links you provided, thanks!
If you were to apply that napkin math to a petri dish solution of bacteria, you'd come up with a disappointing small number of possible evolutionary 'outcomes'.
Yet, in a few hours, with the right environmental pressure, a petri dish will happily evolve a strain of, say, antibiotic-resistant bacteria.
Why is this the case?
Edit - Actually, maybe not. Let me give this a shot - just to do napkin math again.
1 Staphylococcus aureus is ~1um in diameter or ~5e-15 liters in volume. They reproduce let's say once per hour, so if you had a liter of them after an hour you would have (1 / 5e-15) * 2^24 or 3.3e21 reproduction events.
Let's say immunity is transferred at a specific spot in the DNA, first three nucleotides in gene X are flipped to ACT. Spontaneous DNA replication errors happen roughly between every 10e5 and 10e9 replication events. Staphylococcus aureus has ~2.8M base pairs and we need to set a specific three of them to ACT. My stats are a little rusty but the likelihood of any error happening on the low end is 1:10e5. The likelihood that this error occurs in the right spot while still leaving a viable organism has to be greater than 1:2.8e6 but not 1:1 or you're just spitting out random shit. Let's cut it to 1:1e5. The likelihood that the output of three base pairs being ACT under random chance is 3^4 or 81. All of these together i'm getting ~8e12. So i'm being pretty generous here but the likelihood that you'll get a resistant strain would then be ~1:8e12. With 3.3e21 replication events that gives you almost 2 billion resistant bacteria over 24 hours.
Moving to more reasonable values you start with 10µL of staph, spontaneous replication errors every ten million splits and let's split the chance of the mutation being in the right area to 5e5. You end up with 3.3e16 replication events and roughly 1:4e15 chance of getting a resistant mutation or approximately 20 resistant bacteria after 24 hours.
Still enough to get the party started.
So maybe this case is a little different?
https://aeon.co/essays/paradoxes-of-stability-how-life-began...
Summary of the idea: there are self-replicating chemical systems (e.g. start with a single type of molecule that shows autocatalysis) which will undergo variation (replication mistakes / 'mutations'). These will have to compete with the original systems and might take over if they show a higher replication rate (exponential growth!) in the environment that they are in, i.e. there will be selection and different niches. Complex systems seem to be better suited for this according to him as there seems to be more leeway to work in non-ideal conditions. So there is a sort of darwinian selection process that leads to complexification which in turn leads to what we consider as life.
It does not claim to explain how life on Earth historically came to be but why life might be sort of inevitable.
We know that RNA molecules form spontaneously, and many such molecules exhibit enzyme activity. We know that an astonishingly short molecule suffices to copy other RNA sequences it encounters.
All we need, really, is for two of those to encounter one another in the right conditions, and in short order the seas are filled with really excellent copiers. (Better copiers, as they arise, reproduce faster, and soon dominate.) So, the starting condition of life is an environment brimming with RNA replicators, and numerous copies of whatever aids them. Any sort of encapsulation event is likely to include one or more replicators.
Encapsulation is necessary to preserve improvements in interaction between molecules, so we need spontaneous closed-membrane formation.
Curiously, there is no known instance I know of where a cell produces a membrane de novo. It always starts with what it got from a mother cell. Arguably it is not the nucleic acids that define life, but the membrane.
describes an RNA enzyme (I count 63 base pairs) that replicates itself, in the absence of other things to replicate, so you don't even need two of them, to start with.
The point is that we don't need to assume spontaneous conjunction of all the particles that make a complete system. We should instead guess that before there was what we would recognize as life, many of the constituent parts, at a scale much larger than atoms or amino acids, had arisen, many a result of pre-life analogs of living processes such as replication, driven to high populations by pre-life statistical analogs of natural selection.
Then life becomes not a miraculous coincidence, but an inevitability.
we're not too far either
https://www.theguardian.com/science/2019/may/15/cambridge-sc...