Evolutionary Gibbering.
Sep. 30th, 2004 07:58 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
swestrupI'm still feeling sick, and my brain is stuttering, so I doubt that what I'm going to post will make any sense, but I wanted to try and put down some thoughts so they don't get forgotten. I watched (and taped) a 2-hour NOVA yesterday about the origins of life and the prospects of life on other planets. I liked it, and pretty much thought they gave an even-handed explanation.
They took the trouble to point out the face problem and said that faces would be very unlikely on aliens. They also mentioned that amino acids and proteins are very likely parts of alien biology, but DNA isn't. In fact, I only noticed one factual error, but it showed up in a number of places.
That error, now that I think of it, is based on a false assumption, and is endemic to discussions about the way evolution works and the probabilities of intelligence evolving.
That assumption became clear to me when a biologist said 'There have been over 100 million species on this planet. Only one has proven intelligence. That makes for pretty long odds against it happening'. There are multiple bad assumptions here, but the big one is that the probabilities are independant, which is very far from the case.
Evolution can be (and often is) modelled as a randomly directed walk through a fitness landscape of all possible life forms. What I've never heard of anyone modelling is the fact that the presense of an existing life form in a niche makes that niche less valuable to other species. It is usually easier to evolve into an unoccupied space on the fitness graph, than to compete with existing organisms that are already there (and are usually better adapted to the niche).
So, those 100 million species aren't 100 million independant tries at intelligence. They are 100 million nodes on a tangled graph through the search space of all life forms. Somewhere (or somewheres) on that graph are regions where creatures are intelligent. Each new species starts at an existing space on the edge of the graph, and takes a random walk. The chance of that creature falling into a region of intelligence then depends on the chance that the starting node was close to one of those regions, and that depends on the total number of nodes in the graph. So, the more species that there are, the more of the non-intelligent niches that are filled, and the greater the likelyhood of intelligence arising.
Thus the biologists statement about the likelyhood of intelligence is completly wrong. Its as if they had said that the chance of a grain of sand being 3-feet from the floor was vanishingly small, because we had so far dropped 100 million grains into a pile, and so far only one has stayed above the 3 foot mark. The fact that there is now a 3-foot pile of sand, and that additional grains are very likely to end up that high, is never taken into consideration.
As a related fact (but don't expect me to be able to explain HOW its related till I feel better), I should mention that the history of that graph is partially encoded in the DNA of the creatures that the nodes represent. So, if all creatures in a niche are wiped out, I would expect that niche to become repopulated MUCH faster than it happened the first time. There are already creatures that have wandered into and back out of the space of the niche. If any of them survive the wipe-out, they will find it very easy to readapt to that niche and exploit the new empty space. So, when they say that an asteroid could hit the planet and destroy all multicellular life, it does NOT follow, that it would then take even a billion years to get back where we are, nevermind the 3 that is often implied.
This also relates to recent studies on the sizes of extinct giant dogs, and what that implies for the stability of the graph where it involves megafauna predators and prey.
Anyway, I think that all of the above means that a much better statistician than me might be able to use the above to get a much better estimate of the probability of intelligence arising in an ecosystem than has previously been done.
They took the trouble to point out the face problem and said that faces would be very unlikely on aliens. They also mentioned that amino acids and proteins are very likely parts of alien biology, but DNA isn't. In fact, I only noticed one factual error, but it showed up in a number of places.
That error, now that I think of it, is based on a false assumption, and is endemic to discussions about the way evolution works and the probabilities of intelligence evolving.
That assumption became clear to me when a biologist said 'There have been over 100 million species on this planet. Only one has proven intelligence. That makes for pretty long odds against it happening'. There are multiple bad assumptions here, but the big one is that the probabilities are independant, which is very far from the case.
Evolution can be (and often is) modelled as a randomly directed walk through a fitness landscape of all possible life forms. What I've never heard of anyone modelling is the fact that the presense of an existing life form in a niche makes that niche less valuable to other species. It is usually easier to evolve into an unoccupied space on the fitness graph, than to compete with existing organisms that are already there (and are usually better adapted to the niche).
So, those 100 million species aren't 100 million independant tries at intelligence. They are 100 million nodes on a tangled graph through the search space of all life forms. Somewhere (or somewheres) on that graph are regions where creatures are intelligent. Each new species starts at an existing space on the edge of the graph, and takes a random walk. The chance of that creature falling into a region of intelligence then depends on the chance that the starting node was close to one of those regions, and that depends on the total number of nodes in the graph. So, the more species that there are, the more of the non-intelligent niches that are filled, and the greater the likelyhood of intelligence arising.
Thus the biologists statement about the likelyhood of intelligence is completly wrong. Its as if they had said that the chance of a grain of sand being 3-feet from the floor was vanishingly small, because we had so far dropped 100 million grains into a pile, and so far only one has stayed above the 3 foot mark. The fact that there is now a 3-foot pile of sand, and that additional grains are very likely to end up that high, is never taken into consideration.
As a related fact (but don't expect me to be able to explain HOW its related till I feel better), I should mention that the history of that graph is partially encoded in the DNA of the creatures that the nodes represent. So, if all creatures in a niche are wiped out, I would expect that niche to become repopulated MUCH faster than it happened the first time. There are already creatures that have wandered into and back out of the space of the niche. If any of them survive the wipe-out, they will find it very easy to readapt to that niche and exploit the new empty space. So, when they say that an asteroid could hit the planet and destroy all multicellular life, it does NOT follow, that it would then take even a billion years to get back where we are, nevermind the 3 that is often implied.
This also relates to recent studies on the sizes of extinct giant dogs, and what that implies for the stability of the graph where it involves megafauna predators and prey.
Anyway, I think that all of the above means that a much better statistician than me might be able to use the above to get a much better estimate of the probability of intelligence arising in an ecosystem than has previously been done.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
no subject
Date: 2004-10-01 05:43 am (UTC)Most important of all, however, is that the benefit of intelligence is that it solves complex problems, and one of the characteristics of a rich ecosystem is that it poses complex problems for intelligence to evolve to solve!
no subject
Date: 2004-10-01 07:49 am (UTC)Given those, you get a formula for how long it takes intelligence to show up.
no subject
Date: 2004-10-02 09:03 am (UTC)no subject
Date: 2004-10-02 09:59 am (UTC)no subject
Date: 2004-10-02 10:07 am (UTC)