Here’s where the primordial soup theory becomes implausible. A primordial soup is at chemical equilibrium. A cell is at chemical disequilibrium. Could a primordial soup form a cell? Sure, in the same sense that all the atoms in my body could quantum teleport themselves to the Restaurant at the End of the Universe. “Things randomly happened to violate thermodynamics” is so statistically unlikely it should be the last resort we fall back upon after every conceivable hypotheses fails.
Not if we already believe the world is Big. (And indeed one conclusion we could draw from all this is that the world is probably Big.) In a big world—say, one of infinite spatial or temporal extent—everything that can happen does happen infinitely often, no matter how unlikely it is to happen in any particular place. Indeed, given that we don’t see any aliens in the sky, there must be some sort of Great Filter, and so why don’t we just conclude that this is probably it? Solve two mysteries at once! (why we don’t see aliens & the Vital Question).
Relatedly, for a project I’m working on, it would be great to know precisely how unlikely it would be for a given puddle of primordial soup to spontaneously originate life in a given second. Are we talking 1/2^100, or 1/2^10000, or what?
One thing is: we’re in a universe where there are in fact geochemical features that concentrate reactive precursors. For example, the white smokers (a.k.a. alkaline hydrothermal vents) discussed here are expected to exist on lots of planets (IIRC, at least according to Nick Lane—I think he said white smoker = water + olivine, and those are both super common on planets).
So, if geochemical features that concentrate reactive precursors do in fact exist all over the place, and if such features make it 10100× more likely (or whatever) for life to arise, then the question of whether life could arise even in the absence of such features would appear to be an irrelevant question, right?
Anyway, if you want the emergence of life to be the great filter (as opposed to the emergence of eukaryotes, for example, which did in fact take longer on earth), you can still have that in an alkaline hydrothermal vent. For example, maybe it has to be just the right sub-sub-type of alkaline hydrothermal vent, a sub-sub-type which can only form on certain types of very young planets, or whatever. Or maybe you have to win the lottery with the reactive precursors coming together into just the right configuration (although I guess the speed that it happened on earth would vote against that one).
BTW the book has a good discussion of why the emergence of eukaryotes would be plausibly “hard”—like all the things that needed to go right in the first few generations after the merge. It seems to have happened only once in earth’s history, despite there being a gazillion coexisting archaea and bacteria, all around the planet, all the time.
I’m confused by this—even if everything happens infinitely many times, there are still different sizes of infinity and we select the biggest as the most likely. How does a big world shift this perspective?
What do you mean by different sizes of infinity here? Your ultimate claim is that a hypothesis in which life appears on every 10^40th planet is more probable than a hypothesis in which life appears on every 10^400th planet, even if both hypotheses have infinitely many instances of life appearing. (Perhaps you are using SIA, instead of SSA? SIA rewards hypotheses that have larger population sizes.)
If this claim is true, wouldn’t it also be true that a hypothesis in which life appears on every planet is more probable than a hypothesis in which life appears on every 10^40th planet? Shouldn’t we be looking hard to find an explanation for why aliens might be hiding from us, since it’s so improbable that aliens just don’t exist?
I was speaking to the probability of life appearing in different parts of the (finite) ocean, so possibly I misunderstood what you were addressing. But since the reasoning should generalize:
I mean different sizes of infinity like these:
the set-of-all-sets is infinite, but the set-of-all-sets-including-that-set is one set larger, and so on
given an infinite length, 1⁄2 the length is still infinite, and so is 1⁄3 the length, but the 1⁄2 length is larger than the 1⁄3 length
an infinite tube of soap foam and an infinite rod of steel of the same diameter both contain infinitely many atoms, but there are more atoms in the steel, because it is much more dense
there are infinitely many numbers between 0 and 1. But there are twice as many numbers between 0 and 2.
More germane to the life example, if process 1 generates life at 1 per unit time, and process 2 generates life at 2 per unit time, as the arrow of time extends infinitely, both process 1 and 2 will generate life infinitely many times, but process 2 genesis is twice as large as process 1 genesis.
I’m not familiar enough with either SSA or SIA to apply them, and my grasp of anthropic reasoning is shaky in the extreme, but the idea of not rewarding population sizes baffles me. Do you have a preferred breakdown for this point I should check out, or will google serve me well enough?
If this claim is true, wouldn’t it also be true that a hypothesis in which life appears on every planet is more probable than a hypothesis in which life appears on every 10^40th planet?
If planets and stars lasted infinitely long, and were sufficiently constrained in their composition, then I would say yes. But this chain of reasoning ignores the local information we have about the problem. Returning to the primordial soup quote you were responding to, the argument is that what we know of thermodynamics doesn’t allow a causal mechanism to work. By contrast, the white-smoker vent hypothesis does allow a causal mechanism to work; therefore we should prefer it as the explanation for the origin of life (as we know it).
When I try running the intuition of your example of different frequencies of planetary genesis in reverse, and on primordial soup:
We should keep primordial soup on the table because a big world predicts this will still work an infinite number of times, then surely we must also keep a less-complex primordial soup (say, a primordial cocktail) on the table for the same reason; and then bare rock, with no soup at all; and then a complex cell springing into existence with no causal history; etc. This may be true, but doesn’t really seem helpful in terms of what to expect.
An alternative framing: would it be fair to say that under a big world, every prediction happens an infinite number of times? If I accept that all infinities should be treated the same, that still leaves us with the ability to compare the number of infinities, which should lead us to favor the hypotheses according to how many predictions they allow us to make.
Regarding the appearance of aliens, have you checked out the grabby aliens post yet? I recommend it.
If density/population size is so important, you may not like the conclusions you end up with! In particular, you probably end up concluding that this whole world is a simulation run on a different substrate, and that different substrate is infinite in extent and also extremely densely packed with life. I think...
SSA vs. SIA: I don’t have a favorite link, sorry. It’s been a while since I thought about this stuff. I remember liking Katja Grace’s posts on the subject, and Bostrom’s Anthropic Bias. Plus googling and the philosophy literature should work pretty well. Also some Stuart Armstrong posts but that’s more advanced.
Yeah, thinking about grabby aliens is the thing I need to know the probability number for. :)
Before seeing any evidence, we should indeed expect that life has high density in the universe. We just have enough data to rule that out. More generally I think UDASSA is probably the best framework for approaching problems like this, and it would hold that, in situations where our existence is contingent on an anthropically-selected unlikely event, we should still expect that this event is as likely as possible while being consistent with the evidence. So 10^-40 likelihood origination events more probably than 10^-400 likelihood events.
Also, primordial soups don’t necessarily have to be in chemical equilibrium. If the soup is sitting in sunlight, then it could very easily be out of chemical equilibrium, and probably would be. (Soups could be out of equilibrium for a variety of other reasons too, like a consistent influx of chemicals coming out of the ground.) Sunlight comes from a surface with a temperature of 5700K. Compared to a puddle with a temperature of 300K sitting on the Earth’s surface, those photons are absurdly energetic, and could easily bump up the prevalence of some molecules that would otherwise have too much free-energy to exist.
Once you have excess free energy, copying information becomes possible, and it becomes conceivable that you could have self-copying information, aka life.
Not if we already believe the world is Big. (And indeed one conclusion we could draw from all this is that the world is probably Big.) In a big world—say, one of infinite spatial or temporal extent—everything that can happen does happen infinitely often, no matter how unlikely it is to happen in any particular place. Indeed, given that we don’t see any aliens in the sky, there must be some sort of Great Filter, and so why don’t we just conclude that this is probably it? Solve two mysteries at once! (why we don’t see aliens & the Vital Question).
Relatedly, for a project I’m working on, it would be great to know precisely how unlikely it would be for a given puddle of primordial soup to spontaneously originate life in a given second. Are we talking 1/2^100, or 1/2^10000, or what?
One thing is: we’re in a universe where there are in fact geochemical features that concentrate reactive precursors. For example, the white smokers (a.k.a. alkaline hydrothermal vents) discussed here are expected to exist on lots of planets (IIRC, at least according to Nick Lane—I think he said white smoker = water + olivine, and those are both super common on planets).
So, if geochemical features that concentrate reactive precursors do in fact exist all over the place, and if such features make it 10100× more likely (or whatever) for life to arise, then the question of whether life could arise even in the absence of such features would appear to be an irrelevant question, right?
Anyway, if you want the emergence of life to be the great filter (as opposed to the emergence of eukaryotes, for example, which did in fact take longer on earth), you can still have that in an alkaline hydrothermal vent. For example, maybe it has to be just the right sub-sub-type of alkaline hydrothermal vent, a sub-sub-type which can only form on certain types of very young planets, or whatever. Or maybe you have to win the lottery with the reactive precursors coming together into just the right configuration (although I guess the speed that it happened on earth would vote against that one).
BTW the book has a good discussion of why the emergence of eukaryotes would be plausibly “hard”—like all the things that needed to go right in the first few generations after the merge. It seems to have happened only once in earth’s history, despite there being a gazillion coexisting archaea and bacteria, all around the planet, all the time.
I’m confused by this—even if everything happens infinitely many times, there are still different sizes of infinity and we select the biggest as the most likely. How does a big world shift this perspective?
What do you mean by different sizes of infinity here? Your ultimate claim is that a hypothesis in which life appears on every 10^40th planet is more probable than a hypothesis in which life appears on every 10^400th planet, even if both hypotheses have infinitely many instances of life appearing. (Perhaps you are using SIA, instead of SSA? SIA rewards hypotheses that have larger population sizes.)
If this claim is true, wouldn’t it also be true that a hypothesis in which life appears on every planet is more probable than a hypothesis in which life appears on every 10^40th planet? Shouldn’t we be looking hard to find an explanation for why aliens might be hiding from us, since it’s so improbable that aliens just don’t exist?
I was speaking to the probability of life appearing in different parts of the (finite) ocean, so possibly I misunderstood what you were addressing. But since the reasoning should generalize:
I mean different sizes of infinity like these:
the set-of-all-sets is infinite, but the set-of-all-sets-including-that-set is one set larger, and so on
given an infinite length, 1⁄2 the length is still infinite, and so is 1⁄3 the length, but the 1⁄2 length is larger than the 1⁄3 length
an infinite tube of soap foam and an infinite rod of steel of the same diameter both contain infinitely many atoms, but there are more atoms in the steel, because it is much more dense
there are infinitely many numbers between 0 and 1. But there are twice as many numbers between 0 and 2.
More germane to the life example, if process 1 generates life at 1 per unit time, and process 2 generates life at 2 per unit time, as the arrow of time extends infinitely, both process 1 and 2 will generate life infinitely many times, but process 2 genesis is twice as large as process 1 genesis.
I’m not familiar enough with either SSA or SIA to apply them, and my grasp of anthropic reasoning is shaky in the extreme, but the idea of not rewarding population sizes baffles me. Do you have a preferred breakdown for this point I should check out, or will google serve me well enough?
If planets and stars lasted infinitely long, and were sufficiently constrained in their composition, then I would say yes. But this chain of reasoning ignores the local information we have about the problem. Returning to the primordial soup quote you were responding to, the argument is that what we know of thermodynamics doesn’t allow a causal mechanism to work. By contrast, the white-smoker vent hypothesis does allow a causal mechanism to work; therefore we should prefer it as the explanation for the origin of life (as we know it).
When I try running the intuition of your example of different frequencies of planetary genesis in reverse, and on primordial soup:
We should keep primordial soup on the table because a big world predicts this will still work an infinite number of times, then surely we must also keep a less-complex primordial soup (say, a primordial cocktail) on the table for the same reason; and then bare rock, with no soup at all; and then a complex cell springing into existence with no causal history; etc. This may be true, but doesn’t really seem helpful in terms of what to expect.
An alternative framing: would it be fair to say that under a big world, every prediction happens an infinite number of times? If I accept that all infinities should be treated the same, that still leaves us with the ability to compare the number of infinities, which should lead us to favor the hypotheses according to how many predictions they allow us to make.
Regarding the appearance of aliens, have you checked out the grabby aliens post yet? I recommend it.
If density/population size is so important, you may not like the conclusions you end up with! In particular, you probably end up concluding that this whole world is a simulation run on a different substrate, and that different substrate is infinite in extent and also extremely densely packed with life. I think...
SSA vs. SIA: I don’t have a favorite link, sorry. It’s been a while since I thought about this stuff. I remember liking Katja Grace’s posts on the subject, and Bostrom’s Anthropic Bias. Plus googling and the philosophy literature should work pretty well. Also some Stuart Armstrong posts but that’s more advanced.
Yeah, thinking about grabby aliens is the thing I need to know the probability number for. :)
Before seeing any evidence, we should indeed expect that life has high density in the universe. We just have enough data to rule that out. More generally I think UDASSA is probably the best framework for approaching problems like this, and it would hold that, in situations where our existence is contingent on an anthropically-selected unlikely event, we should still expect that this event is as likely as possible while being consistent with the evidence. So 10^-40 likelihood origination events more probably than 10^-400 likelihood events.
Also, primordial soups don’t necessarily have to be in chemical equilibrium. If the soup is sitting in sunlight, then it could very easily be out of chemical equilibrium, and probably would be. (Soups could be out of equilibrium for a variety of other reasons too, like a consistent influx of chemicals coming out of the ground.) Sunlight comes from a surface with a temperature of 5700K. Compared to a puddle with a temperature of 300K sitting on the Earth’s surface, those photons are absurdly energetic, and could easily bump up the prevalence of some molecules that would otherwise have too much free-energy to exist.
Once you have excess free energy, copying information becomes possible, and it becomes conceivable that you could have self-copying information, aka life.