Most “predictions of evolution” that can be found online are more about finding past evidence of common descent (e.g. fossils) rather than predicting the future path that evolution will take. To apologize for that, people say that evolution is hard to predict because it’s directionless, e.g. it doesn’t necessarily lead to more complexity, larger number of individuals, larger total mass, etc. That leads to the question, is there some deep reason why we can’t find any numerical parameter that is predictably increased by evolution, or is it just that we haven’t looked hard enough?
Plenty of people predict that increased antibiotica use will lead to a raise in antibiotica resistance among bacteria.
Organisms like bacteria that have much more iterations behind them then humans also tend to have less waste in their DNA.
Grasses beat trees at growing in glades with animals that eat plants. Why? Grass has more iterations behind them and is therefore better optimized for the enviroment than the trees.
A tree has to get lucky to survive the beginning. If it surives the beginning it can however grow tall and win.
Let’s say you keep the enviroment stable for 2 billion years. Everything evolves naturally. Then you take tree seeds and bring them back to the present time. I think there a good chance that such a tree would outcompete grass at growing in glades.
Most “predictions of evolution” that can be found online are more about finding past evidence of common descent (e.g. fossils) rather than predicting the future path that evolution will take.
Fossils don’t really get used as the central evidence of common descent anymore. These days common descent usually get’s determined by looking at the DNA.
In my experience people who discuss evolution online that do focus on fossils are usually atheists who behave as if their atheism is a religion. They think it’s important to defend Darwin against the creationists. On the other hand they aren’t up to date with the current science on evolution.
Organisms like bacteria that have much more iterations behind them then humans also tend to have less waste in their DNA.
Grasses beat trees at growing in glades with animals that eat plants. Why? Grass has more iterations behind them and is therefore better optimized for the enviroment than the trees.
You seem to be predicting that grasses have smaller genomes than trees, but wheat is famous for having a huge genome. Here’s a table of a few plants. Maybe wheat is an outlier and I’d be interested if you had documentation of some pattern, but I’ve always heard that there is none.
Do you have evidence that the variation in genome size among multicellular organisms is not variation in waste? Added: As far as I know, the consensus is that it is. If you disagree with the consensus, you should acknowledge that’s what you’re doing.
Do you have evidence that the variation in genome size among multicellular organisms is not variation in waste?
I haven’t made a claim that strong. To the extend I made a claim it’s not all variation in genome size between multicellular organisms is due to different amount of waste.
And no I don’t intend to claim something that’s out of consensus in this topic. To the extend I might differ on this topic from consensus consider that to be errors.
If I remember right then one reason for plants like grasses to have long genomes was to have multiple copies of genes to speed up protein production.
What do you mean, “predict”? It has been empirically observed, a lot.
cousin made the claim that we can only say something about evolution that happened in the past. I say that we can confidently predict that increasing antibiotica resistance among bacteria will continue in the future.
Huh? It doesn’t work like that at all. For one thing, the “environment” isn’t stable.
Firstly describing complex system in a ew words is seldom completely accurate. The question is whether it’s a useful mental model for thinking about it.
In this case the idea I wanted to communicate is that it’s very useful to think about the speed of iterations and the competitive advantage that a specis gets by having as advantage of hundred of millions of iterations over their competitors.
The enviroment doesn’t have to be stable for the argument that I made. In changing enviroments a spezies with faster iterations adapts faster. A lot of genetic adaptions are also about housekeeping genes that are useful in most enviroments.
Bacterial cultures taken of sick people introduce microevolutions by trying to grow bacteria in several different antibiotic environments. Doctors try to use antibiotics that cause dead ends in those evolutions. I’m not sure how much time mutation has to happen in vitro, though.
Evolution leads to a higher level of fitness in the environment, but the problem is that the environment itself is constantly changing in unpredictable ways. It’s like an optimization process where the utility function itself is contantly changing. That’s why it’s very hard to reliably quantify fitness. For instance, billions of years ago, the increase in oxygen in the atmosphere killed a lot of existing organisms and forced aerobic bacteria on to the scene.
Replies to comments that attempted to point out a numerical parameter that’s increased by evolution. (I’d be more interested in comments pointing out a deep reason why we can’t find such a numerical parameter, but there were no such comments.)
lmm:
Life “wants” to spread, so perhaps an increase in the volume in which life can be found?
That’s been steady for awhile now.
ChristianKl:
Organisms like bacteria that have much more iterations behind them then humans also tend to have less waste in their DNA.
Evolution can both add and remove junk DNA. Humans are descended from bacteria.
David_Gerard:
Total number of species (including extinct).
That can’t decrease by definition, and will increase under any mechanism that gives nonzero chance of speciation, e.g. if God decides to create new species at random.
Lumifer:
The chances of successful transmission of genes across generations given a stable environment.
Evolution can both add and remove junk DNA. Humans are descended from bacteria.
More particularly, the equilibrium size of the DNA is very roughly inversely correlated with population size. A larger population size is better at filtering out disadvantagous traits. It’s not linear—there are discontinuities as decreasing population size eliminates natural selection’s ability to select against different things. And those things sometimes can even go on to be selected for for other reasons—there are genomic structures that are important for eukaryotes that could probably never have evolved in a bacterium because to get to them you need to go through various local minima of fitness.
Soil bacteria can have trillions of individuals per cubic meter of dirt and they actually experience direct evolution towards lower genome size—more DNA means more sites at which something could mutate and become problematic and they actually feel this force. Eukaryotes go up in volume by a factor of ~1000 and go down in population by at least as much, and lose much of the ability to select against introns and middling amounts of intergenic DNA and expanding repeat-based centromere elements.
Multicellular creatures with piddlingly tiny population sizes compared to microbes lose much of the ability to select against selfish transposon DNA elements, gigantic introns and gene deserts, and their promoter elements get fragmented into pieces strewn across many kilobases rather than one compact transcriptional regulation element of a few dozen to a few hundred base pairs (granted, we’ve also been able to make good use of some of these things for interesting purposes from our adaptive immune system to the concerted regulation of our hox gene clusters that regulate our body plans). They also become very sensitive to the particular character of the transposons or DNA repair machinery of their particular lineage and wind up random-walking like crazy up and down an order of magnitude or two in genome size as a result.
Thanks! I was hoping you’d show up, it’s always nice to get a lesson :-)
Going back to the original question, are there any “general purpose adaptations” that never disappear once they show up? Does evolution act like a ratchet in any way at all?
Closest thing I can think of from what I know without going through literature is the building up of chains of dependencies. Once you have created a complex system that needs every bit to function, it has a tendency to stay as a unit or completely leave.
You can see that in a couple contexts. One is ‘subfunctionalization’. Gene duplications are fairly common across evolution—one gene gets duplicated into two identical genes and they are free to evolve separately. You usually hear about that in the context of one getting a new function, but that’s actually comparatively rare. Much more likely is both copies breaking slightly differently until now both of them are necessary. A major component of the ATP-generating apparatus in fungi went through this: a subunit that is elsewhere composed of a ring of identical proteins now has to be composed of a ring of two alternating almost identical proteins neither of which can do the job on its own. Ray-finned fish recently went through a whole-genome duplication, and a number of their developmental transcription factors are now subfunctionalized such that, say, one does the job in the head end and the other does its job in the tail end.
Another context is the organism I work in, yeast. I like to call yeast “a fungus that is trying its damndest to become a bacterium”. It lives in a context much like many bacteria and it has shrunk its genome down to maybe 2.5x that of an E. coli and its generation time down to 90 minutes. But it still has 40 introns hanging out in less than 1% of its genes so it needs a fully functional spliceosome complex to be able to process those transcripts lest those 40 genes utterly fail all at once, and it has most of the hallmarks of eukaryotic genome structure and regulation (in a neat, smaller, more research-friendly package). That being said it has lost a few big eukaryotic systems, like nonsense-mediated RNA decay and RNA interference, and they left relatively little trace behind.
Sure, but mostly because evolution’s so good at it. The fact that evolution so quickly filled a tidal pool, so quickly filled all the tidal pools, so quickly filled the oceans, so quickly covered the land, is evidence of strength rather than weakness.
There does seem to be a “punctuated equilibrium” effect here; life fills a region, appears static for a while, but then makes a breakthrough and rapidly fills another region. It could be argued that this is also true of things that humans optimize for: human population growth has abruptly rapidly accelerated at least twice (invention of agriculture, industrial revolution). Slavery was everywhere in the ancient world, then eliminated across most of it in the space of a century. Gay marriage went from hopefully-it-will-happen-in-my-lifetime to anyone who opposed it being basically shunned. Scientific and technological breakthroughs tend to look a lot like this.
Generalizing this to all optimization processes would be very speculative.
Evolution can both add and remove junk DNA. Humans are descended from bacteria.
From bacteria that lived a long time ago. Not from those that live today that had many iterations to optimize themselves. Different bacteria species can also much better exchange genes with each other than vertebrates that need viruses to do so.
Implying that humans evolved from the kind of bacterias that are around today might be more wrong than saying that the bacteria we see know evolved from humans. There more evolutionary distance between todays bacteria and those from which humans descended and humans and those bacteria from which they descended.
Yeah, and there are often bacteria in a single flower pot that are less related to each other than you are to the potted plant. But both bacteria still have a much smaller genome than you or the plant, maybe because genome size matters for reproduction speed for them, but is insignificant for us.
That can’t decrease by definition, and will increase under any mechanism that gives nonzero chance of speciation, e.g. if God decides to create new species at random.
Just apply Occam.
That seems to be contradicted by the possibility of evolutionary suicide.
Possibility wouldn’t contradict anything, a high enough probability would.
That seems to be contradicted by the possibility of evolutionary suicide.
Evolutionary suicide seems to be someone’s theoretical idea. Is there any evidence that it happens in evolution in reality?
In any case, are you basically trying to find the directionality of evolution? On a meta level higher than “adapted to the current environment”? There probably isn’t. Evolution is a quite simple mechanism, it just works given certain conditions. It is not goal-oriented, it’s just how the world is.
However if I were forced to find something correlated with evolution, I’d probably say complexity.
Damn it. It was going to be a better example because I was going to give the actual genera (Aspidoscelis and Cnemidophorus) of whiptail lizards whose species keep going down this path and then I got distracted and didn’t do that. Oops.
Depends on your time frame. Looking at the whole history of life on Earth evolution certainly correlates with complexity, looking at the last few million years, not so much.
I understand the argument about the upper limit of genetic information that can be sustained. I am somewhat suspicious of it because I’m not sure what will happen to this argument if we do NOT assume a stable environment (so the target of the optimization is elusive, it’s always moving) and we do NOT assume a single-point optimum but rather imagine a good-enough plateau on which genome could wander without major selection consequences.
But I haven’t thought about it enough to form a definite opinion.
I think you just don’t give an amoeba much credit because it’s no multicellular organism. It’s genome is 100-200 times the size of the human. As it’s that big it seems like we haven’t sequenced all of it so we don’t know how many genes it has.
We also know very little about amoeba. Genetic analysis suggests that the do exchange genes with each other in some form but we don’t know how.
Amoeba probably express a lot of stuff phenotypically that we don’t yet understand.
Why should there be a numerical parameter predictably increased by evolution? Why not look for a numerical parameter predictably increased by continental drift? or by prayer? by ostriches?
One of the key pieces of justification for FAI is the idea of “optimization process”. Evolution is given as an example of such process, unlike continental drift or ostriches. It seems natural to ask what parameter is optimized.
Just FYI, I interpret that question very differently than your original.
Why don’t you start with a simpler example, like a thermostat? Would you not call that an optimization process, minimizing the difference between observed and desired temperature?
Most of your rejections of suggestions in this thread would also reject the thermostat. An ideal thermostat keeps the temperature steady. Its utility function never improves, let alone monotonically. A real thermostat is even worse, continually taking random steps back. In extreme weather, it runs continually, but never gets anywhere near goal. It only optimizes within its ability. Similarly, evolution does not expand life without bound, because it has reached its limit of its ability to exploit the planet. This limit is subject to the fluctuations of climate. But the main limit on evolution is that it is competing with itself. Eliezer suggests that it is better to make it plural, “because fox evolution works at cross-purposes to rabbit evolution.” I think most teleological errors about evolution are addressed by making it plural.
Also, thermostats occasionally commit suicide by burning down the building and losing control of future temperature. (PS—I think the best example of evolutionary suicide are genes that hijack meiosis to force their propagation, doubling their fitness in the short term. I’ve been told that ones that are sex-linked have been observed to very quickly wipe out the population, but I can’t find a source. Added: the phase is “meiotic drive,” though I still don’t have an example leading to extinction.)
OK, if we replace “evolution” with “fox evolution”, and “increase” with “try to keep steady”, what parameter does fox evolution try to keep steady? Or am I missing the point of your analogy?
Do you mean to say that the expected inclusive fitness of a randomly selected creature from the population goes up with time? Well, if we sum that up over the whole population, we obtain the total number of offspring—right? And dividing that by the current population, we see that the expected inclusive fitness of a randomly selected creature is simply the population’s growth rate. The problem is that evolution does not always lead to >1 population growth rate. Eliezer gave a nice example of that: “It’s quite possible to have a new wolf that expends 10% more energy per day to be 20% better at hunting, and in this case the sustainable wolf population will decrease as new wolves replace old.”
While I don’t know of any simple or convenient numerical parameter, I’d note that we do have some handy non-retrospective pieces of evidence for evolution by natural selection, such as the induced occurrence of evolutionary benchmarks such as multicellularity.
In general, there are some adaptations which are highly predictable under certain circumstances, but there may not be any sort of meaningful measure we can use for evolution of organisms over time which aren’t a function of their relationship with their environment.
I think whatever numerical parameter evolution raised generally (not always) in respect to its environment, it would have to do with meaningful complexity , however that can be numerically expressed, and local decrease in entropy. Design would cause those too, but hypothesizing it would violate occam’s razor.
Different environments and different substrates for mutation cause different kinds of evolutions.
One main thing that happens with a long enough period of selection in a simple, stable environment on a microorganism is a shrinking of the genome.
You quite simply will not find a simple parameter perpetually increased by evolution. Whatever works better for that base organism in that particular environment will become more common. One thing being selected for under all circumstances and showing up all the time is just not the reality.
we can’t find any numerical parameter that is predictably increased by evolution
The chances of successful transmission of genes across generations given a stable environment. The number of offspring surviving to reproductive age is a good first-order approximation.
If you want something more tangible, predictions what features evolution would lose are rather easy—those that are (energy-)expensive and are useless in the new environment.
There have been plenty of evolutionary simulations, surely they provide some testable predictions. I vaguely recall one of them: that new adaptations tend to propagate first in small isolated groups and only then spread through the rest of the species. I don’t recall if this has been tested through the fossil records. I am sure there are many more testable predictions. Like how fish locked in a dark cave or murky water tend to lose eyesight. But the exact path is probably too hard to predict. For example, marine mammals did not develop gills. Or that mammals develop intelligence by growing Neocortex, while birds use DVR (dorsal ventricular ridge) or maybe Nidopallium for the same purpose.
Huh? Even if you accept the estimates that your link points to, the amount of information in mammalian genome and optimization power of evolution are VERY different things.
If you can narrow down the number of possible lifeforms to one in 2^n, that’s n bits of optimization power, and n bits of information as to what the final lifeform is.
If life is getting more and more optimal, then we can simply wait until we know that less than one in 2^25 million lifeforms are that optimal, and we have more than 25 megabytes of information as to what that lifeform is.
then we can simply wait until we know that less than one in 2^25 million lifeforms are that optimal
You go and wait. I’ll do other things in the meantime :-) Do you have any intuition how large that number is?
and we have more than 25 megabytes of information as to what that lifeform is.
You’ve spent all that 25Mb for an index into the lifeform space but you have not budgeted any information for the actual description of the lifeform.
Imagine the case where there’s one bit. It tells you whether creature-0 or creature-1 is optimal. But it doesn’t tell you what these creatures are.
In any case, all these numbers are based on the resistance of Earth mammals to genetic drift. That really doesn’t limit how evolution can optimize with different creatures in different places.
Do you have any intuition how large that number is?
It’s not going through them one at a time.
You’ve spent all that 25Mb for an index into the lifeform space but you have not budgeted any information for the actual description of the lifeform.
It’s not a simple English description, but narrowing down the possibilities by a factor of two is always one bit of information. It doesn’t matter whether it’s “the first bit is one”, “the xor of all the bits is one” or even “it’s a hash of something starting with a one using X algorithm, which is a bijection”.
Imagine the case where there’s one bit. It tells you whether creature-0 or creature-1 is optimal. But it doesn’t tell you what these creatures are.
It’s the one with a higher inclusive genetic fitness. That’s what evolution optimizes for.
If evolution has n bits of optimization power, that’s equivalent to saying that if you order all possible lifeforms based on how optimal they are, this is going to be in the top 1/2^n of them. (It’s actually somewhat more complicated, since it’s more likely to be higher up and there’s some chance of it being lower, but that’s the basic idea.)
In any case, all these numbers are based on the resistance of Earth mammals to genetic drift. That really doesn’t limit how evolution can optimize with different creatures in different places.
It does vary based on what lifeform you’re looking at, since they all have different mutation rates and different numbers of children, but there’s always a limit to the information, and I’m pretty sure that it’s pretty much always a limit that’s already been hit.
By my calculations, if you had the entire earth’s surface covered by a solid meter-thick layer of bacteria for 4.6 billion years and each bacterium lived for 1 hour, that would be approximately 2^155 bacteria having lived and died.
You can massively increase genetic information (inasmuch as that actually means much in biology) very quickly with very simple genetic changes. It’s not a case of searching through every possible 1 bit change.
narrowing down the possibilities by a factor of two is always one bit of information
Provided, of course, that your space of possibilities is finite and you know what it is. In the case of evolution you don’t.
that’s equivalent to saying that if you order all possible lifeforms
I don’t understand what does “all possible lifeforms” mean. Does not compute.
but there’s always a limit to the information, and I’m pretty sure that it’s pretty much always a limit that’s already been hit.
Which limit? The limit of information in the mammalian genome? Or the limit of evolution—whatever exists is the pinnacle an no better (given the same environment) can be achieved?
Most “predictions of evolution” that can be found online are more about finding past evidence of common descent (e.g. fossils) rather than predicting the future path that evolution will take. To apologize for that, people say that evolution is hard to predict because it’s directionless, e.g. it doesn’t necessarily lead to more complexity, larger number of individuals, larger total mass, etc. That leads to the question, is there some deep reason why we can’t find any numerical parameter that is predictably increased by evolution, or is it just that we haven’t looked hard enough?
Plenty of people predict that increased antibiotica use will lead to a raise in antibiotica resistance among bacteria.
Organisms like bacteria that have much more iterations behind them then humans also tend to have less waste in their DNA.
Grasses beat trees at growing in glades with animals that eat plants. Why? Grass has more iterations behind them and is therefore better optimized for the enviroment than the trees.
A tree has to get lucky to survive the beginning. If it surives the beginning it can however grow tall and win.
Let’s say you keep the enviroment stable for 2 billion years. Everything evolves naturally. Then you take tree seeds and bring them back to the present time. I think there a good chance that such a tree would outcompete grass at growing in glades.
Fossils don’t really get used as the central evidence of common descent anymore. These days common descent usually get’s determined by looking at the DNA. In my experience people who discuss evolution online that do focus on fossils are usually atheists who behave as if their atheism is a religion. They think it’s important to defend Darwin against the creationists. On the other hand they aren’t up to date with the current science on evolution.
You seem to be predicting that grasses have smaller genomes than trees, but wheat is famous for having a huge genome. Here’s a table of a few plants. Maybe wheat is an outlier and I’d be interested if you had documentation of some pattern, but I’ve always heard that there is none.
If you want to be exact I didn’t say genome size but waste. Through mutation inactived genes, retroviruses and so on. It takes time to remove them.
Do you have evidence that the variation in genome size among multicellular organisms is not variation in waste? Added: As far as I know, the consensus is that it is. If you disagree with the consensus, you should acknowledge that’s what you’re doing.
I haven’t made a claim that strong. To the extend I made a claim it’s not all variation in genome size between multicellular organisms is due to different amount of waste.
And no I don’t intend to claim something that’s out of consensus in this topic. To the extend I might differ on this topic from consensus consider that to be errors.
If I remember right then one reason for plants like grasses to have long genomes was to have multiple copies of genes to speed up protein production.
What do you mean, “predict”? It has been empirically observed, a lot.
Huh? It doesn’t work like that at all. For one thing, the “environment” isn’t stable.
cousin made the claim that we can only say something about evolution that happened in the past. I say that we can confidently predict that increasing antibiotica resistance among bacteria will continue in the future.
Firstly describing complex system in a ew words is seldom completely accurate. The question is whether it’s a useful mental model for thinking about it. In this case the idea I wanted to communicate is that it’s very useful to think about the speed of iterations and the competitive advantage that a specis gets by having as advantage of hundred of millions of iterations over their competitors.
The enviroment doesn’t have to be stable for the argument that I made. In changing enviroments a spezies with faster iterations adapts faster. A lot of genetic adaptions are also about housekeeping genes that are useful in most enviroments.
Bacterial cultures taken of sick people introduce microevolutions by trying to grow bacteria in several different antibiotic environments. Doctors try to use antibiotics that cause dead ends in those evolutions. I’m not sure how much time mutation has to happen in vitro, though.
Evolution leads to a higher level of fitness in the environment, but the problem is that the environment itself is constantly changing in unpredictable ways. It’s like an optimization process where the utility function itself is contantly changing. That’s why it’s very hard to reliably quantify fitness. For instance, billions of years ago, the increase in oxygen in the atmosphere killed a lot of existing organisms and forced aerobic bacteria on to the scene.
Replies to comments that attempted to point out a numerical parameter that’s increased by evolution. (I’d be more interested in comments pointing out a deep reason why we can’t find such a numerical parameter, but there were no such comments.)
lmm:
That’s been steady for awhile now.
ChristianKl:
Evolution can both add and remove junk DNA. Humans are descended from bacteria.
David_Gerard:
That can’t decrease by definition, and will increase under any mechanism that gives nonzero chance of speciation, e.g. if God decides to create new species at random.
Lumifer:
That seems to be contradicted by the possibility of evolutionary suicide.
Humans don’t have more offspring than bacteria in average conditions, and have much fewer offspring in ideal conditions.
More particularly, the equilibrium size of the DNA is very roughly inversely correlated with population size. A larger population size is better at filtering out disadvantagous traits. It’s not linear—there are discontinuities as decreasing population size eliminates natural selection’s ability to select against different things. And those things sometimes can even go on to be selected for for other reasons—there are genomic structures that are important for eukaryotes that could probably never have evolved in a bacterium because to get to them you need to go through various local minima of fitness.
Soil bacteria can have trillions of individuals per cubic meter of dirt and they actually experience direct evolution towards lower genome size—more DNA means more sites at which something could mutate and become problematic and they actually feel this force. Eukaryotes go up in volume by a factor of ~1000 and go down in population by at least as much, and lose much of the ability to select against introns and middling amounts of intergenic DNA and expanding repeat-based centromere elements.
Multicellular creatures with piddlingly tiny population sizes compared to microbes lose much of the ability to select against selfish transposon DNA elements, gigantic introns and gene deserts, and their promoter elements get fragmented into pieces strewn across many kilobases rather than one compact transcriptional regulation element of a few dozen to a few hundred base pairs (granted, we’ve also been able to make good use of some of these things for interesting purposes from our adaptive immune system to the concerted regulation of our hox gene clusters that regulate our body plans). They also become very sensitive to the particular character of the transposons or DNA repair machinery of their particular lineage and wind up random-walking like crazy up and down an order of magnitude or two in genome size as a result.
Thanks! I was hoping you’d show up, it’s always nice to get a lesson :-)
Going back to the original question, are there any “general purpose adaptations” that never disappear once they show up? Does evolution act like a ratchet in any way at all?
Closest thing I can think of from what I know without going through literature is the building up of chains of dependencies. Once you have created a complex system that needs every bit to function, it has a tendency to stay as a unit or completely leave.
You can see that in a couple contexts. One is ‘subfunctionalization’. Gene duplications are fairly common across evolution—one gene gets duplicated into two identical genes and they are free to evolve separately. You usually hear about that in the context of one getting a new function, but that’s actually comparatively rare. Much more likely is both copies breaking slightly differently until now both of them are necessary. A major component of the ATP-generating apparatus in fungi went through this: a subunit that is elsewhere composed of a ring of identical proteins now has to be composed of a ring of two alternating almost identical proteins neither of which can do the job on its own. Ray-finned fish recently went through a whole-genome duplication, and a number of their developmental transcription factors are now subfunctionalized such that, say, one does the job in the head end and the other does its job in the tail end.
Another context is the organism I work in, yeast. I like to call yeast “a fungus that is trying its damndest to become a bacterium”. It lives in a context much like many bacteria and it has shrunk its genome down to maybe 2.5x that of an E. coli and its generation time down to 90 minutes. But it still has 40 introns hanging out in less than 1% of its genes so it needs a fully functional spliceosome complex to be able to process those transcripts lest those 40 genes utterly fail all at once, and it has most of the hallmarks of eukaryotic genome structure and regulation (in a neat, smaller, more research-friendly package). That being said it has lost a few big eukaryotic systems, like nonsense-mediated RNA decay and RNA interference, and they left relatively little trace behind.
Sure, but mostly because evolution’s so good at it. The fact that evolution so quickly filled a tidal pool, so quickly filled all the tidal pools, so quickly filled the oceans, so quickly covered the land, is evidence of strength rather than weakness.
There does seem to be a “punctuated equilibrium” effect here; life fills a region, appears static for a while, but then makes a breakthrough and rapidly fills another region. It could be argued that this is also true of things that humans optimize for: human population growth has abruptly rapidly accelerated at least twice (invention of agriculture, industrial revolution). Slavery was everywhere in the ancient world, then eliminated across most of it in the space of a century. Gay marriage went from hopefully-it-will-happen-in-my-lifetime to anyone who opposed it being basically shunned. Scientific and technological breakthroughs tend to look a lot like this.
Generalizing this to all optimization processes would be very speculative.
From bacteria that lived a long time ago. Not from those that live today that had many iterations to optimize themselves. Different bacteria species can also much better exchange genes with each other than vertebrates that need viruses to do so.
Implying that humans evolved from the kind of bacterias that are around today might be more wrong than saying that the bacteria we see know evolved from humans. There more evolutionary distance between todays bacteria and those from which humans descended and humans and those bacteria from which they descended.
Yeah, and there are often bacteria in a single flower pot that are less related to each other than you are to the potted plant. But both bacteria still have a much smaller genome than you or the plant, maybe because genome size matters for reproduction speed for them, but is insignificant for us.
Just apply Occam.
Possibility wouldn’t contradict anything, a high enough probability would.
Evolutionary suicide seems to be someone’s theoretical idea. Is there any evidence that it happens in evolution in reality?
In any case, are you basically trying to find the directionality of evolution? On a meta level higher than “adapted to the current environment”? There probably isn’t. Evolution is a quite simple mechanism, it just works given certain conditions. It is not goal-oriented, it’s just how the world is.
However if I were forced to find something correlated with evolution, I’d probably say complexity.
Species of nightshade tend to evolve to become self-fertile, before dying out due to lack of genetic diversity.
Is this your source?
Link? Lots of plants are self-fertile and do quite well...
Better example: parthenogenic lizard species.
What makes that example better?
Damn it. It was going to be a better example because I was going to give the actual genera (Aspidoscelis and Cnemidophorus) of whiptail lizards whose species keep going down this path and then I got distracted and didn’t do that. Oops.
This doesn’t seem to be the case either
Depends on your time frame. Looking at the whole history of life on Earth evolution certainly correlates with complexity, looking at the last few million years, not so much.
I understand the argument about the upper limit of genetic information that can be sustained. I am somewhat suspicious of it because I’m not sure what will happen to this argument if we do NOT assume a stable environment (so the target of the optimization is elusive, it’s always moving) and we do NOT assume a single-point optimum but rather imagine a good-enough plateau on which genome could wander without major selection consequences.
But I haven’t thought about it enough to form a definite opinion.
Complexity in what way? Kolmogoroph complexity of DNA?
No, complexity of the phenotype.
How would you go about measuring that complexity?
I don’t know. Eyeballing it seems to be a good start.
Why do you ask? Do you think that such things are unmeasurable or there are radically different ways of measuring them or what?
I have a hard time trying to form a judgement about whether a human is more or less complex than a dinosaur via eyeballing.
Is a grasshopper more of less complex than a human?
Well, would you have problems arranging the following in the order of complexity: a jellyfish, a tree, an amoeba, a human..?
Yes.
I think you just don’t give an amoeba much credit because it’s no multicellular organism. It’s genome is 100-200 times the size of the human. As it’s that big it seems like we haven’t sequenced all of it so we don’t know how many genes it has.
We also know very little about amoeba. Genetic analysis suggests that the do exchange genes with each other in some form but we don’t know how.
Amoeba probably express a lot of stuff phenotypically that we don’t yet understand.
Sabre-toothed tigers and mammoths.
Huh? Sense make not.
Why should there be a numerical parameter predictably increased by evolution? Why not look for a numerical parameter predictably increased by continental drift? or by prayer? by ostriches?
One of the key pieces of justification for FAI is the idea of “optimization process”. Evolution is given as an example of such process, unlike continental drift or ostriches. It seems natural to ask what parameter is optimized.
Just FYI, I interpret that question very differently than your original.
Why don’t you start with a simpler example, like a thermostat? Would you not call that an optimization process, minimizing the difference between observed and desired temperature?
Most of your rejections of suggestions in this thread would also reject the thermostat. An ideal thermostat keeps the temperature steady. Its utility function never improves, let alone monotonically. A real thermostat is even worse, continually taking random steps back. In extreme weather, it runs continually, but never gets anywhere near goal. It only optimizes within its ability. Similarly, evolution does not expand life without bound, because it has reached its limit of its ability to exploit the planet. This limit is subject to the fluctuations of climate. But the main limit on evolution is that it is competing with itself. Eliezer suggests that it is better to make it plural, “because fox evolution works at cross-purposes to rabbit evolution.” I think most teleological errors about evolution are addressed by making it plural.
Also, thermostats occasionally commit suicide by burning down the building and losing control of future temperature. (PS—I think the best example of evolutionary suicide are genes that hijack meiosis to force their propagation, doubling their fitness in the short term. I’ve been told that ones that are sex-linked have been observed to very quickly wipe out the population, but I can’t find a source. Added: the phase is “meiotic drive,” though I still don’t have an example leading to extinction.)
OK, if we replace “evolution” with “fox evolution”, and “increase” with “try to keep steady”, what parameter does fox evolution try to keep steady? Or am I missing the point of your analogy?
Inclusive reproductive fitness.
Do you mean to say that the expected inclusive fitness of a randomly selected creature from the population goes up with time? Well, if we sum that up over the whole population, we obtain the total number of offspring—right? And dividing that by the current population, we see that the expected inclusive fitness of a randomly selected creature is simply the population’s growth rate. The problem is that evolution does not always lead to >1 population growth rate. Eliezer gave a nice example of that: “It’s quite possible to have a new wolf that expends 10% more energy per day to be 20% better at hunting, and in this case the sustainable wolf population will decrease as new wolves replace old.”
While I don’t know of any simple or convenient numerical parameter, I’d note that we do have some handy non-retrospective pieces of evidence for evolution by natural selection, such as the induced occurrence of evolutionary benchmarks such as multicellularity.
In general, there are some adaptations which are highly predictable under certain circumstances, but there may not be any sort of meaningful measure we can use for evolution of organisms over time which aren’t a function of their relationship with their environment.
I think whatever numerical parameter evolution raised generally (not always) in respect to its environment, it would have to do with meaningful complexity , however that can be numerically expressed, and local decrease in entropy. Design would cause those too, but hypothesizing it would violate occam’s razor.
Different environments and different substrates for mutation cause different kinds of evolutions.
One main thing that happens with a long enough period of selection in a simple, stable environment on a microorganism is a shrinking of the genome.
You quite simply will not find a simple parameter perpetually increased by evolution. Whatever works better for that base organism in that particular environment will become more common. One thing being selected for under all circumstances and showing up all the time is just not the reality.
Shouldn’t directed evolution i.e. selective breeding count?
Are you looking for a parameter that is always increased by any kind of evolution? It doesn’t make sense to me why you would.
The chances of successful transmission of genes across generations given a stable environment. The number of offspring surviving to reproductive age is a good first-order approximation.
If you want something more tangible, predictions what features evolution would lose are rather easy—those that are (energy-)expensive and are useless in the new environment.
There have been plenty of evolutionary simulations, surely they provide some testable predictions. I vaguely recall one of them: that new adaptations tend to propagate first in small isolated groups and only then spread through the rest of the species. I don’t recall if this has been tested through the fossil records. I am sure there are many more testable predictions. Like how fish locked in a dark cave or murky water tend to lose eyesight. But the exact path is probably too hard to predict. For example, marine mammals did not develop gills. Or that mammals develop intelligence by growing Neocortex, while birds use DVR (dorsal ventricular ridge) or maybe Nidopallium for the same purpose.
Total number of species (including extinct).
Life “wants” to spread, so perhaps an increase in the volume in which life can be found?
Newly created islands may have “weird” biospheres initially, but evolve towards a more “normal” set of niches over time?
But why would life get more optimal? Evolution has finite optimization power, and it has long ago already reached this limit.
Huh? Even if you accept the estimates that your link points to, the amount of information in mammalian genome and optimization power of evolution are VERY different things.
How do you figure?
If you can narrow down the number of possible lifeforms to one in 2^n, that’s n bits of optimization power, and n bits of information as to what the final lifeform is.
If life is getting more and more optimal, then we can simply wait until we know that less than one in 2^25 million lifeforms are that optimal, and we have more than 25 megabytes of information as to what that lifeform is.
You go and wait. I’ll do other things in the meantime :-) Do you have any intuition how large that number is?
You’ve spent all that 25Mb for an index into the lifeform space but you have not budgeted any information for the actual description of the lifeform.
Imagine the case where there’s one bit. It tells you whether creature-0 or creature-1 is optimal. But it doesn’t tell you what these creatures are.
In any case, all these numbers are based on the resistance of Earth mammals to genetic drift. That really doesn’t limit how evolution can optimize with different creatures in different places.
It’s not going through them one at a time.
It’s not a simple English description, but narrowing down the possibilities by a factor of two is always one bit of information. It doesn’t matter whether it’s “the first bit is one”, “the xor of all the bits is one” or even “it’s a hash of something starting with a one using X algorithm, which is a bijection”.
It’s the one with a higher inclusive genetic fitness. That’s what evolution optimizes for.
If evolution has n bits of optimization power, that’s equivalent to saying that if you order all possible lifeforms based on how optimal they are, this is going to be in the top 1/2^n of them. (It’s actually somewhat more complicated, since it’s more likely to be higher up and there’s some chance of it being lower, but that’s the basic idea.)
It does vary based on what lifeform you’re looking at, since they all have different mutation rates and different numbers of children, but there’s always a limit to the information, and I’m pretty sure that it’s pretty much always a limit that’s already been hit.
By my calculations, if you had the entire earth’s surface covered by a solid meter-thick layer of bacteria for 4.6 billion years and each bacterium lived for 1 hour, that would be approximately 2^155 bacteria having lived and died.
You can massively increase genetic information (inasmuch as that actually means much in biology) very quickly with very simple genetic changes. It’s not a case of searching through every possible 1 bit change.
Provided, of course, that your space of possibilities is finite and you know what it is. In the case of evolution you don’t.
I don’t understand what does “all possible lifeforms” mean. Does not compute.
Which limit? The limit of information in the mammalian genome? Or the limit of evolution—whatever exists is the pinnacle an no better (given the same environment) can be achieved?
Something like “humans will have larger skulls and smaller teeth”?
But we know that isn’t true.