Why do people believe in evolution?

[Gene Parmesan]

Well all the good folks here can debate all day long, & I've never seen anyone concede defeat !(except by that most graceless form; insult) But there is something very humbling about arguing with a computer.

I had a friend who was a surgeon & skeptic of Darwinian evolution, I thought I could objectively prove to him the power of the Darwinian algorithm in a program - again I'm not claiming to have single handedly debunked ToE - I just proved to myself that it absolutely did not work the way it did in my thought experiments- which led me to investigate further

I think it is due to a fundamental bias- everything we do is in anticipation of a future consequence, that's how we are wired, and it's almost impossible to remove this 'forethought' in our thinking, but a computer can.

Thank you so much for the thoughtful response. It is really interesting. I do a bit of programming myself, though mostly simple web-based stuff for work, but I'm curious about the program you were working on to support evolution. And if it did not align with your thought experiments, how did you determine your thought experiments weren't faulty?

EDIT: You answered some of this in a previous post, so don't worry about that part.

 

[pitabread]

natural selection acting upon random mutation, very heavily biased towards a favorable outcome- it just amazed me how much you have to tilt the playing field for this to be effective, to the point you're essentially providing the outcome.

I mean specifically, what was the actual algorithm? IOW, what was the program and how did it work?

You don't have to post the actual code, although you can if you have it. But at least a pseudocode representation of what you wrote is what I'm looking for here.

I want to see how what you wrote actually works.

(I have a computer science background so feel free to be technical.)

 

[Guy Threepwood]

I mean specifically, what was the actual algorithm? IOW, what was the program and how did it work?

You don't have to post the actual code, although you can if you have it. But at least a pseudocode representation of what you wrote is what I'm looking for here.

I want to see how what you wrote actually works.

(I have a computer science background so feel free to be technical.)

I wrote it to be flexible so you could set different starting conditions, small, large populations, numbers of distinct (non interbreeding) species, genetic traits, mutation rates, harshness of environment etc

but basically using a 2D spatial environment, with variable assignments for up to I believe 100,000 individual life forms, each with individual genetic traits, e.g. metabolism, speed, vision, -which could be randomly mutated at various rates. Each individual was put to the test- i.e. not just given a mathematical 'survivability' variable- they had to compete, move, seek out food, survive till maturity, mate etc- kinda fun

The reproduction was assumed to be sexual-i.e. genes of two healthy/mature surviving individuals combined to produce new offspring. The very generous part here is that the random mutations had a 50/50 chance of being beneficial v deleterious - far better than in the real world.

The aim initially wasn't so much to show new emergent properties developing, rather just to show that superior genetic lines would naturally be selected over time. You could observe in real time, or cycle through millions of generations. What was surprising to me is how readily advantageous genetic lines simply die out in even a fraction of the chaos that exists in the real world- and/or how unrealistically great an advantage must be, for the algorithm to select it.

As someone put it here, evolution progresses in 'microsteps' and that works in our minds, even small advantages will eventually pay off over time. In the algorithm, without that foresight, they never get the chance, an insignificant advantage is simply that- insignificant

What tends to happen is what we see in reality, e.g. horseshoe crabs, populations remaining in virtual stasis for 100's of millions of years, or dying out completely

 

[pitabread]

The reproduction was assumed to be sexual-i.e. genes of two healthy/mature surviving individuals combined to produce new offspring. The very generous part here is that the random mutations had a 50/50 chance of being beneficial v deleterious - far better than in the real world.

How was sexual reproduction modeled in your simulation?

Likewise, how were beneficial or deleterious mutations modeled? In nature, whether mutations are beneficial or deleterious (or neutral) can be environmentally dependent. Did you take that into account in how you modeled things?

The aim initially wasn't so much to show new emergent properties developing, rather just to show that superior genetic lines would naturally be selected over time. You could observe in real time, or cycle through millions of generations. What was surprising to me is how readily advantageous genetic lines simply die out in even a fraction of the chaos that exists in the real world- and/or how unrealistically great an advantage must be, for the algorithm to select it.

What was the specific determining factors for survivability? Was it strictly finding food? Or were there other factors?

Were there changing environmental factors to consider relative to the 'genotypes' of your organisms? This could include something like a pathogen being introduced, changing climate conditions impacting mobility, food availability, etc.

Was anything like predator/prey relationships modeled or other dynamic interactions between organisms?

Was there any modeling of speciation events (e.g. sympatric or allopatric speciation)? Or was it assumed that all organisms were the same species?

Was their any population geography modeled? For example, cutting off portions of the population geographically and having them evolve independently, including things like founder effects? Conversely, was there any modeling of geographically distinct populations being reintroduced?

As someone put it here, evolution progresses in 'microsteps' and that works in our minds, even small advantages will eventually pay off over time. In the algorithm, without that foresight, they never get the chance, an insignificant advantage is simply that- insignificant

With what you've described so far, this doesn't sound unsurprising. What I'd be interested in is determining the extent of selection pressures in your simulation. In an environment with relatively low selection pressure, random genetic drift would be more prominent and you would see exactly what you described.

Did you happen to measure selection coefficients in your simulation?

 

[Guy Threepwood]

Thank you so much for the thoughtful response. It is really interesting. I do a bit of programming myself, though mostly simple web-based stuff for work, but I'm curious about the program you were working on to support evolution. And if it did not align with your thought experiments, how did you determine your thought experiments weren't faulty?

EDIT: You answered some of this in a previous post, so don't worry about that part.

That's interesting, I do some web stuff for work also and I find that much more difficult and frustrating, compatibility with platforms etc.. you finally get something to work the way you want and the next browser/windows update wrecks it *$^&$

I started on a 16k machine in about 1982? it couldn't do much but at least you had complete control of it!

 

[Guy Threepwood]

How was sexual reproduction modeled in your simulation?

well it wasn't all that fun , but again individuals had to be 'fit' and 'mature' I figured if they have to 'physically' peruse mates and outpace rivals, that would help amplify the selection process (again I was trying to prove the mechanism, not debunk it)

Likewise, how were beneficial or deleterious mutations modeled? In nature, whether mutations are beneficial or deleterious (or neutral) can be environmentally dependent. Did you take that into account in how you modeled things?

I had the mutations occurring at the point of reproduction, I didn't get into epigenetics no!
I certainly admit this was obviously nothing close to the real world in complexity or resolution, but again the basic mechanism people are taught and run through their heads is a very simple principle that seems intuitively very compelling - without a lot of extra mechanisms required.

What was the specific determining factors for survivability? Was it strictly finding food? Or were there other factors?

finding food (random vegetation)- and also predation from carnivorous species if included

Were there changing environmental factors to consider relative to the 'genotypes' of your organisms? This could include something like a pathogen being introduced, changing climate conditions impacting mobility, food availability, etc.

Sure, I figured some sort of dynamic producing periods of 'stress' would also help speed the selection process. So the vegetation would go through periods of slower growth (as a simple emulation of climate variation), it could also be over-eaten by a thriving population

Was anything like predator/prey relationships modeled or other dynamic interactions between organisms?

as above yes, carnivorous species could be introduced at the beginning or at any time (divine intervention!) which would only eat other species (and would have to be fast enough to catch them)

Was there any modeling of speciation events (e.g. sympatric or allopatric speciation)? Or was it assumed that all organisms were the same species?

That was actually included by a 'size' gene, if the sizes of 2 individuals had grown too far apart, they could not mate and hence were effectively a new species- or you could introduce your own

Was their any population geography modeled? For example, cutting off portions of the population geographically and having them evolve independently, including things like founder effects? Conversely, was there any modeling of geographically distinct populations being reintroduced?

That was one interesting thing that did happen 'naturally'- the vegetation reproduced from itself, so where it was 'overgrazed' you'd get 'deserts' separating distinct populations. Each individual had a sight horizon that would allow it to head towards vegetation or prey only within a range

With what you've described so far, this doesn't sound unsurprising. What I'd be interested in is determining the extent of selection pressures in your simulation. In an environment with relatively low selection pressure, random genetic drift would be more prominent and you would see exactly what you described.

Did you happen to measure selection coefficients in your simulation?

Sure, I love stats- charts kept track of overall fitness of all genes within species and populations
Again pressure varied enough to often wipe out entire species, so I don't think that was the issue-
I just found myself having to tweak parameters to the point that I'm deliberately forcing good selection towards a desired goal - being the hand of God,

again, just not nearly as emphatic a proof of the simple Darwinian algorithm as I was sure it would be-

 

[Guy Threepwood]

This kind of piqued my interest in it again, I'll need to go but I'll see if I can get it to run on my system- and add some of your suggestions- it's about 12 versions of windows out of date by now!

 

[Gene Parmesan]

That's interesting, I do some web stuff for work also and I find that much more difficult and frustrating, compatibility with platforms etc.. you finally get something to work the way you want and the next browser/windows update wrecks it *$^&$

I started on a 16k machine in about 1982? it couldn't do much but at least you had complete control of it!

Haha! I'm currently addressing an issue that is only showing up on one type of device and ONLY when it is printed. I've found a solution but yeah. I feel that.

 

[Guy Threepwood]

Haha! I'm currently addressing an issue that is only showing up on one type of device and ONLY when it is printed. I've found a solution but yeah. I feel that.

Well it's good brain exercise I guess.. but it sometimes makes you which you were something like a lumberjack instead.. hmm that's probably paying pretty good right now!

 

[FrumiousBandersnatch]

I'm not saying I accurately modeled billions of years of evolution and disproved Darwin, but I proved to myself that it definitely did not work as in my thought experiments. Everything we do is in anticipation of a future beneficial consequence, and so this bias is very difficult to remove in the mind, but a computer can. It boils down to an information problem, without a generous external supply, the algorithm falls foul to entropy.

Again not the whole story but one of those moments that make you go 'hmmmm... '

It sounds like you're saying that your evolution program didn't work as you'd expected, therefore the ToE is wrong, therefore God... ?

That doesn't sound sensible - can you clarify?

 

[FrumiousBandersnatch]

Where is the repeatable experiment, observation, by which a single celled bacteria morphs into a human being through random errors in its genetic code? or anything else for that matter?

One can do science and make discoveries even when direct repeatable experiment and observation of hypothesised phenomena is not possible.

So far we have got as far as- more bacteria- that leaves quite a bit to the imagination.

The steps involve multicellular clumps, followed by cell specialisation. Steps exhibited by many contemporary organisms - and, as predicted, by the remains of ancient ones too.

 

[FrumiousBandersnatch]

It could.. and then again you could also credit the Bible and church with helping Europe establish that dominance, the church was the keeper of knowledge during the dark ages and propagated that learning.
Even just being able to read, write and copy the enormous text of the Bible kept those skills sharp!

But many Eastern and Far-Eastern scholastic traditions kept and propagated their knowledge through that and other periods of disorder without need or use of the Bible - I suggest it was the power and authority of the church that preserved the Bible.

 

[Guy Threepwood]

It sounds like you're saying that your evolution program didn't work as you'd expected, therefore the ToE is wrong, therefore God... ?

That doesn't sound sensible - can you clarify?

Well I wrote it around 2005, so it was about another 7 years or so before I would have called myself a theist. But it did give me pause for thought at the time, probably the first fracture in my being 'utterly convinced' of ToE

Like many people, I was taught the opposite of what you are saying i.e. 'ToE is obviously correct therefore no (designer) God (And the teaching of ToE is often heavily peppered with 'Bad Design' arguments- "what intelligent creator would design such and such...")

So I wrote it to try to back up my argument to a skeptic of Darwinism- that the basic principle of natural selection acting on random mutation was so demonstrably, logically powerful that there was practically no alternative. But it did more to back up what my skeptical friend was saying- without the benefit/bias of forethought, it did not work as effectively as I had conceived from thought experiments.

 

[Guy Threepwood]

One can do science and make discoveries even when direct repeatable experiment and observation of hypothesised phenomena is not possible.

Sure, but as in steady state, big crunch, phrenology, human evolution based around Piltdown man etc-
it leaves a lot more freedom for speculation, ideology, peer pressure, when a theory cannot be directly demonstrated.

 

[FrumiousBandersnatch]

I wrote it to be flexible so you could set different starting conditions, small, large populations, numbers of distinct (non interbreeding) species, genetic traits, mutation rates, harshness of environment etc

but basically using a 2D spatial environment, with variable assignments for up to I believe 100,000 individual life forms, each with individual genetic traits, e.g. metabolism, speed, vision, -which could be randomly mutated at various rates. Each individual was put to the test- i.e. not just given a mathematical 'survivability' variable- they had to compete, move, seek out food, survive till maturity, mate etc- kinda fun

The reproduction was assumed to be sexual-i.e. genes of two healthy/mature surviving individuals combined to produce new offspring. The very generous part here is that the random mutations had a 50/50 chance of being beneficial v deleterious - far better than in the real world.

The aim initially wasn't so much to show new emergent properties developing, rather just to show that superior genetic lines would naturally be selected over time. You could observe in real time, or cycle through millions of generations. What was surprising to me is how readily advantageous genetic lines simply die out in even a fraction of the chaos that exists in the real world- and/or how unrealistically great an advantage must be, for the algorithm to select it.

As someone put it here, evolution progresses in 'microsteps' and that works in our minds, even small advantages will eventually pay off over time. In the algorithm, without that foresight, they never get the chance, an insignificant advantage is simply that- insignificant

What tends to happen is what we see in reality, e.g. horseshoe crabs, populations remaining in virtual stasis for 100's of millions of years, or dying out completely

When I studied genetics as part of my biology degree, we spent many tedious hours calculating the time for genes with a certain selective advantage to go to fixation in populations of particular sizes, mutation rates, levels of selection pressure, etc. Even without the additional bells & whistles of real evolutionary processes, it was surprising to me how very small advantages could be successfully fixed given appropriate numbers. This, and the large number of successful computer evolutionary simulations produced over the years, inclines me to think that either your code was not emulating the process correctly or that your input parameters were unrealistic. There are certainly plenty of ways to get it wrong.

If my whole worldview was at stake, I would spend some time studying the mathematics of evolution (it's not that difficult) and then look at other evolutionary simulations to see how they were doing it - this way I would expect either to discover where I was going wrong or be able to demonstrate where tens of thousands of experienced biologists and statisticians had been getting it wrong for over a century...

In the '90s I spent some time playing around with an early evolution simulation, Tierra, which, like the simplest real-world evolution, has no specific selection criteria other than death from various causes. Given a few hundred thousand generations, an identical population of the simplest replicators would give rise to a whole ecosystem of parasites, symbiotes, predators, groups of co-operators, etc., with evolved changes that improved efficiency in the replicator code itself (unrolling loops by repetition of code, and so-on). Like many computer evolutionary simulations, it eventually went into stasis or total extinction - perhaps due to the lack of complexity and variation in the environment.

 

[VirOptimus]

Well I wrote it around 2005, so it was about another 7 years or so before I would have called myself a theist. But it did give me pause for thought at the time, probably the first fracture in my being 'utterly convinced' of ToE

Like many people, I was taught the opposite of what you are saying i.e. 'ToE is obviously correct therefore no (designer) God (And the teaching of ToE is often heavily peppered with 'Bad Design' arguments- "what intelligent creator would design such and such...")

So I wrote it to try to back up my argument to a skeptic of Darwinism- that the basic principle of natural selection acting on random mutation was so demonstrably, logically powerful that there was practically no alternative. But it did more to back up what my skeptical friend was saying- without the benefit/bias of forethought, it did not work as effectively as I had conceived from thought experiments.

It only proves you are a subpar programmer with an incomplete understanding of the ToE.

But I dont think you did program anything or that you are from an atheist background anyway.

 

[FrumiousBandersnatch]

So I wrote it to try to back up my argument to a skeptic of Darwinism- that the basic principle of natural selection acting on random mutation was so demonstrably, logically powerful that there was practically no alternative. But it did more to back up what my skeptical friend was saying- without the benefit/bias of forethought, it did not work as effectively as I had conceived from thought experiments.

The principle (heritable variation with natural selection) is effectively just an algorithmic summary, but it hides a wealth of important detail. It's ironic that many early attempts to get something interesting from evolution simulators and cellular automata involved many hours, days, and weeks of pseudo-evolutionary trial and error by the programmers

 

[Guy Threepwood]

When I studied genetics as part of my biology degree, we spent many tedious hours calculating the time for genes with a certain selective advantage to go to fixation in populations of particular sizes, mutation rates, levels of selection pressure, etc. Even without the additional bells & whistles of real evolutionary processes, it was surprising to me how very small advantages could be successfully fixed given appropriate numbers. This, and the large number of successful computer evolutionary simulations produced over the years, inclines me to think that either your code was not emulating the process correctly or that your input parameters were unrealistic. There are certainly plenty of ways to get it wrong.

If my whole worldview was at stake, I would spend some time studying the mathematics of evolution (it's not that difficult) and then look at other evolutionary simulations to see how they were doing it - this way I would expect either to discover where I was going wrong or be able to demonstrate where tens of thousands of experienced biologists and statisticians had been getting it wrong for over a century...

In the '90s I spent some time playing around with an early evolution simulation, Tierra, which, like the simplest real-world evolution, has no specific selection criteria other than death from various causes. Given a few hundred thousand generations, an identical population of the simplest replicators would give rise to a whole ecosystem of parasites, symbiotes, predators, groups of co-operators, etc., with evolved changes that improved efficiency in the replicator code itself (unrolling loops by repetition of code, and so-on). Like many computer evolutionary simulations, it eventually went into stasis or total extinction - perhaps due to the lack of complexity and variation in the environment.

'successful computer evolutionary simulations'

^ I think that gets to the core of the issue.

I make a distinction between computer models and computer simulations

We can model a trebuchet, with known values for weight, potential energy, air resistance etc, and hence pretty accurately predict how far it can fling an object with a simple shape

Most flight 'simulators' on the other hand do not accurately model the physics of air molecules on control surfaces etc, they just simulate what you'd expect an aircraft to do when you fly it.

The crucial difference with the latter again is anticipation, you are working around a desired outcome, not an unknown one. if 'Tierra' had not been made to work to some extent, you would not have been playing around with it.

My program worked also, it just needed a lot more nudging towards desired goals than I would ever have thought - to the point that I did not present it as proof of the power of natural selection on random mutation, to my skeptical friend, because I knew it would be a cheat on my part.

 

[FrumiousBandersnatch]

Sure, but as in steady state, big crunch, phrenology, human evolution based around Piltdown man etc-
it leaves a lot more freedom for speculation, ideology, peer pressure, when a theory cannot be directly demonstrated.

Sure, science is a human endeavour, subject to human foibles. But steady-state was the default assumption in the absence of contrary evidence (until Hubble); the big crunch was a plausible hypothesis until contrary evidence was discovered; phrenology was pseudoscience; & human evolution wasn't built around Piltdown man.

Science works by testing ideas however it can; inference and interpolation are powerful tools, and untestable predictions of well-established theories carry greater weight than untestable speculations. Ideas that haven't been tested or can't be tested remain speculative or hypothetical, whether the popular media or scientists themselves claim otherwise. Science rightly has a degree of inertia to novelty and has evolved techniques and methodologies to minimise the influence of human bias and dishonesty and to self-correct.

 

[Guy Threepwood]

Sure, science is a human endeavour, subject to human foibles. But steady-state was the default assumption in the absence of contrary evidence (until Hubble); the big crunch was a plausible hypothesis until contrary evidence was discovered; phrenology was pseudoscience; & human evolution wasn't built around Piltdown man.

Science works by testing ideas however it can; inference and interpolation are powerful tools, and untestable predictions of well-established theories carry greater weight than untestable speculations. Ideas that haven't been tested or can't be tested remain speculative or hypothetical, whether the popular media or scientists themselves claim otherwise. Science rightly has a degree of inertia to novelty and has evolved techniques and methodologies to minimise the influence of human bias and dishonesty and to self-correct.

I think we agree, the ideological implications of a theory should have no bearing on it's scientific investigation, unfortunately not all scientists agree with us:

(Wiki)
[Hoyle] found the idea that the universe had a beginning to be pseudoscience, resembling arguments for a creator, "for it's an irrational process, and can't be described in scientific terms"

In the 1920s and 1930s, almost every major cosmologist preferred an eternal steady-state universe, and several complained that the beginning of time implied by the Big Bang imported religious concepts into physics; this objection was later repeated by supporters of the steady-state theory.[56] This perception was enhanced by the fact that the originator of the Big Bang theory, Lemaître, was a Roman Catholic priest.

Just one example of science v atheism, or 'methodological naturalism' - imposing arbitrary rules rather than following the evidence wherever it leads, whatever the implications