The Delusion of Darwinian Natural Law

Acton Institute ^ | 12/27/03 | Marc D. Guerra

[Iris7]

The difference is that the computer is designed by intelligent entities. (At least in theory!) Darwinism as it is usually expressed postulates change from random variation instead of intelligent design, that is, random inputs followed by natural selection. The point I am making is that random inputs do not seem to provide adequate variation to model real world outputs.

[Iris7]

I think I covered that. If the one path to the existing situation observed has a probability of 1x10to the -100th power, and then if there are 1x10 to the 50th other paths possible, then the odds of any path showing is 1x10 to the -50th.

Please remember I don't grind any axes here. I find professions of faith based on anything but personal revelation unaesthetic.

[Iris7]

If there is no randomness, then doesn't this define predestination?

[donh]

Your objections are covered in the text under "irreducibly complex system".

No they aren't. Behe's foundational arguments, particularly with respect to the flagellum example, have been torn to shreads by subsequent (and some not-so-subsequent) scientific findings. See Miller Op. Cit. for the gruesome details.

[donh]

The point I am making is that random inputs do not seem to provide adequate variation to model real world outputs.

However, continuous random distributions, followed by winnowing through a selection criteria, to create a mean distribution with a central tendency, is all you need. It doesn't take rocket science to realize that an environent sculpts a plastic population in a direction to better survive in that environment.

[Iris7]

Very interesting. However, I do not depend on miracles, I believe. Instead I come from the direction that human desire to know how things work leads people to believe that they know how things work. I see most belief as hubris. I am not a nihilist nor a cynic; I just think that humans confuse how smart they want to be with how smart they really are.

[donh]

If there is no randomness, then doesn't this define predestination?

If it's impossible for you to calculate the future, in any far-reaching sense, why should you care if it's predestined or not? You can presently predict where the earth will be in respect to the sun in 25 years--do you think that that will prevent you from enjoying the fall leaves in 25 years?

[donh]

I think I covered that. If the one path to the existing situation observed has a probability of 1x10to the -100th power, and then if there are 1x10 to the 50th other paths possible, then the odds of any path showing is 1x10 to the -50th.

At the end of each path which has creatures, said creatures can all look back and say "Goodness--look at the odds against our particular outcome".

[Iris7]

I agree with your last. However, how true is it that continuous random distributions model the living world? The world I see does not appear continuously variable at all. Perhaps you are refering to variation within a breeding population? Also I have no problem with natural selection, as I stated earlier. Breeding populations show less variation over time as natural selection pressure increases, for instance.

[donh]

Instead I come from the direction that human desire to know how things work leads people to believe that they know how things work.

I believe it leads them to put meat on the table more reliably. And that those who believe things that are sub-optimal, are sub-optimal meat-getters, who tend to die out along with their belief systems.

I see most belief as hubris.

I see it as human. Some are better, or luckier, at it than others.

[Iris7]

I also agree with your #140.

[Iris7]

Itoh is over my head. I have to work for a living, and have never had time for such things, nor much aptitude. But isn't Itoh talking quite a bit about spaces that are quite filled, if I am understanding what that means? I am talking about 10^31 points in a space with the number of dimensions equal to the number of possible variations in life so far on Earth. Are those filled when one point per space even is unlikely?

[donh]

I agree with your last. However, how true is it that continuous random distributions model the living world?

We just went through this, didn't we? The important word you left out is "uniform". I think you are confusing yourself with the word continuous, and we ought to leave it off. A uniform distribution (which is what, to the casual observer, mutation seems to provide us with) would, indeed, not be an engine for change in a population, as you suggest. However, the uniform distribution which we get at birth, is not the distribution that survives to the point of procreation. What we see upon procreation, is a mean distribution with a central tendency (a bell curve). If the survival rate has been small, the central tendency will be large, and the resulting new population will manifest a rapid change in the direction of the central tendency. If the survival rate is large, than there will be a weak central tendency, and the resulting new population won't manifest a rapid change.

The world I see does not appear continuously variable at all. Perhaps you are refering to variation within a breeding population? Also I have no problem with natural selection, as I stated earlier.

Breeding populations show less variation over time as natural selection pressure increases, for instance.

This is too oversimplified to be accurate:

If times are good for the species in question, it won't change much. If times are horrid, it will change a lot--in the direction of making the resulting population regard these times as not so horrid after all.

If the population is already well on the way toward multiple speciation, than extreme pressure will furrow down quickly whichever of these populations are less fit for the new environment. Thus a flowering of the remaining variations will be observed by paleontologists. If the population is fairly uniform, than extreme pressure will furrow under whatever recessives have been "going-along-for-the-ride", EXCEPT, for the recessives that are particularly well adopted for the new environment. Thus, a winnowing of variation will be observed by paleontologists.

[PatrickHenry]

The difference is that the computer is designed by intelligent entities. (At least in theory!)

We know. But but that's an irrelevant point. The flawed principle of "irreducible complexity" is well-illustrated by the computer model.

Darwinism as it is usually expressed postulates change from random variation instead of intelligent design, that is, random inputs followed by natural selection. The point I am making is that random inputs do not seem to provide adequate variation to model real world outputs.

And you can demonstrate this how? If you succeed, you'll be famous.

[Iris7]

My reply used "spaces" wrong. I meant the individual coordinant planes, intersections of two of the axes, not spaces. Shoot, my everyday math is Excel!

[tortoise]

Current physical theory still claims "random" for quantum mechanics. The problem is that any "deterministic" version so far has not been able to reproduce the experimental results.

I think we are talking apples and oranges here. I am not aware of any useful deterministic model of quantum mechanics. We have no ability to model it as anything other than "random" because we have no ability to measure the internal state of the process.

There are two very different things here. There are tests one can do to ascertain the "finite-state-ness" of a blackbox process e.g. quantum mechanics phenomena. Then there is knowing the state and specific nature of the process inside the blackbox. The first is easy but the latter is generally intractable for many types of processes. Crypto systems are built on very similar mathematical constructs.

Quantum mechanics reeks of being a deterministic process from a "finite-state-ness" standpoint; it is not even of the class of processes for which such things are difficult to ascertain. Unfortunately, this is a kind of zero knowledge proof. Knowing this does not give the slightest clue as to the state or structure of the process, but only tells you that it is deterministic. It does NOT allow you to create a deterministic model of the process, and in fact one can show that for many simple deterministic processes it is effectively impossible to ever build a deterministic model without looking inside the blackbox. We are therefore forced to treat many systems as non-deterministic computationally even if they are deterministic in fact. Nonetheless, knowing that something is deterministic is useful in and of itself even if it doesn't allow us to create a deterministic model.

In short, there are many processes that we must treat as random even if we know they are deterministic in nature. Quantum mechanics appears to be one of the myriad of things that fits this profile. Consider strong PRNGs as an example. Very low Kolmogorov complexity, but intractable to reverse engineer the internal state by sampling the output.

[tortoise]

If there is no randomness, then doesn't this define predestination?

Yes, but unless you can perfectly model future states (impossible as it happens), it won't seem deterministic to you. Therefore, you have no reason to act as though there is predestination because you are not capable of perceiving the extent of this fact. Or to put it another way, all of us would necessarily be living under an impenetrable illusion of non-determinism inside a purely deterministic system.

Determinism doesn't ruin all the fun.

[balrog666]

You do have a way with words, but you really need to suppress that urge to hold back and be polite!

[balrog666]

I think I covered that. If the one path to the existing situation observed has a probability of 1x10to the -100th power, and then if there are 1x10 to the 50th other paths possible, then the odds of any path showing is 1x10 to the -50th.

No, you didn't. As PatrickHenry and donh and jennyp (and everybody else) has explained, all paths aren't equal and that is where your hopes die.

[Doctor Stochastic]

You don't even need continuous variables. A finite variance is sufficient to get a central limit theorem. (Not necessary but the those conditions are rather complicated.) With a finite variance, one always gets a Gaussian. Only with distributions with infinite variance can one get a non-Gaussian result. Still there are often limit laws in this case too. Randomness has its own set of laws.