It was a well thought out post, so it deserves some sort of more thoughtful response, but really, the idea the one point is "the most important", is pretty silly.
Then we agree that the mere pretence of randomness does not entail that it would be helpful to label some process as random.
Assuming I understand the question, yes.
Right... but you do know that computers cannot be random right?
I'm not sure it matters, but just in case it does, you're wrong. In general, random computers wouldn't be all that useful for most of the things that computer scientists use them for, but if you wanted to make a random computer, I know how to do it. Some truly hard core mathematicians who studied automata theory might argue whether or not it was "really" a computer when you were done, but if I were so inclined I could make an electronic device that simulates processes, performs calculations, and all that sort of stuff that computers do, and where the results of certain calculations were totally unpredictable. So, if this is an important point, I disagree. I understand where you are coming from, though, I think. The "random" number generators used by computers that affect the calculations are not random at all. However, I don't know the relevance.
Any input to your genetic algorithm would be pseudo-random.
Again, in case it's important, I could create a truly random event if I wanted to. I'll use a measurement of radioactive decay to an external device to generate my random numbers, and it will be truly random, but pseudo-random is good enough in almost all cases.
Yet you told me that evolution couldn't work without randomness... well there's no randomness in what a computer does because everything is causal - deterministic.
What I said was that you couldn't model evolution accurately without modelling it as a random process (or words to that effect). In other words, if you want to answer the question, "Will this gene still be present in the gene pool after X generations?" An accurate answer will always be based on a probability. You will never be able to say, with certainty, that it will be present or will not be present.
So we need an infinite machine to predict a random event; but then the problem is that if we are trying to figure out whether or not something else is truly random, as in acausal, or just an infinite machine, as in causal, and we don't already know the internal mechanism then we can't ever decide whether or not there is a causal mechanism from examining the output of the box.
I'm not sure why you included parts of your explanation, but the bottom line, in the last sentence, is correct. You can't decide whether it's truly random or not from the outside.
So basically, if you wanted to, you could argue that everything is totally and utterly determined if you were happy with the idea that it is possible to build infinite machines.
I'm not sure how an infinite machine relates to radioactive decay.
As I said earlier - but no-one picked up on at all - something could be truly random or it could be simply unpredictable/hard to predict.
Actually, Schneibster and I both picked up on, and addressed this point. It is impossible to tell, and not just because we lack the technology to tell.
1) BOTH ways of describing something, acausal and causal CAN COMPLETELY DESCRIBE THE SAME THING.
Yes. Does anyone disagree?
2) You can therefore make a choice of which one to use when describing the mechanism of some process you are trying to figure out by experiment. (Uhh, which would be everything in this case: science is empirical remember?)
Good. Great. With you so far.
3) As such one should NOT be thinking random/non-random is the definition of the process but rather all processes are predictable/unpredictable with a spectrum inbetween and we simply decide to label things either random or non-random based on their relative predictabilities.
Cool. Good.
So is it really important that some process in evolution is truly random or does it only matter that it is unpredictable?
Only that it's unpredictable. I don't care why the snake ate one egg, and not the other, because I couldn't predict it regardless of the underlying cause.
See the above. Do you still think you know what randomness is?
Yes.
So in other words I am right but you aren't seeing it; it's a matter of choosing the best label.
Agreed.
As such I refer to my earlier definition of evolution for a far more accurate and unambiguous answer to the question:
Mutations are non-deterministic with respect to the genotype.
Selection of genotype is deterministic with respect to the phenotype.
Oops. You lost me here.
Let me try and illustrate. Give me one prediction about the future course of evolution that you can make that doesn't use a probability function. No arm waving here, I want one prediction about the future state of the world, predicted by evolution, and not involving probability. One experimental result that actually works.
You could say, "the most fit genes will survive", but when asked to provide a definition of fitness, you would have to say "the ones that survived".
Nevertheless, it may still be useful to discuss evolution in non-random terms, because the randomness really only shows up when trying to make specific predictions. When trying to describe what happened in the past, such as why a particular species survived or not, it's rarely useful to say, "It was random", even though there were random events. So, the best description will depend on exactly what you are trying to describe.
So you are familiar with it right?
Yes.
It is important for the reason I outlined in my argument above.
If nature used random number generators based on algorithms that computers use, I could see how autocorrelation would be relevant to evolution, but I'm having a hard time applying it to the real world.
Fine institutions but I doubt they would have covered the metaphysical implications computational analysis has for mathematics as I outlined above in most mathematical courses that weren't directly to do with computing.
You never met Professor Uribe, but in general, you are right.
But, suppose I don't give a hoot about the metaphysical implications. And, like a lot of metaphysical implications, it might be strongly influenced by philosophy, but not very useful to science. (Hey, didn't Schneibster just say that...again.)
So again; they don't adequately define randomness. See my earlier randomness test - each sequence is probabilistically describable but each one of them could be completely and utterly determined.
The problem here is that you are applying randomness to sequences of events, as if the test that matters is a random number generator. I fail to see the relevance of random number generators here. I'm about to flip a coin. It's a fair coin. All I care about is whether or not this flip will be heads or tails. I don't care about the next flip. I don't care about the last flip. This flip matters. How does autocorrelation apply?
A bird just laid an egg. I want to know whether the bird embryo in that egg will pass it's genes to the next generation. Do you know the answer, and how would you apply the concept of autocorrelation to this question?
I know how to apply probability density functions to those questions. I'm not seeing the relevance of autocorrelations.
Ah. I see you wish to get lost examining the forest missing the fact I'm talking about the trees.
You gave two possibilities. I think you left at least one possibility out.
ETA:
To summarize, there are two main points.
First, your issues are worthwhile, but they aren't the only worthwhile issues to discuss. Your points are not really the most important points. What is "most important" is a subjective judgement that varies from person to person.
Second, the common description of mutation as random and natural selection as nonrandom can be useful depending on the circumstances, but there are other times, especially when creating mathematical models to predict future events, when that description is not useful.
