|
Source: (consider it)
|
Thread: The Death of Darwinism
|
Glenn Oldham
Shipmate
# 47
|
Posted
quote:
Originally posted by Alan Cresswell:
quote:
Originally posted by sharkshooter:
...I have been trying to understand the main argument here. I think it can be summarized:
"Similarity in design means evolution occurred."
Is that correct? ...
That certainly seemed to be Glenn Oldham's point.
No, the point was more that the same basic feature occurs even though it has no real reason to be the same.
So, in your house analogy, it could be something like an "ancestral house" having a fireplace then subsequent houses maintain that feature even though in many of them the fire becomes a couple of electric bars and there is no chimney, and that the introduction of central heating has made it superfluous anyway.
Yes, that is indeed the point. I am sorry if that has not been clear from my postings. The kinds of similarities I have been pointing to are those which have no apparent relation to function and are thus a challenge to the view that they are explainable by reference to design.
There is no design reason why the whale needs seven vertebrae in its neck which is very short and rigid and would be just as good with fewer. There is no known functional reason why the recurrent laryngeal nerve in, for example, mammals goes from the brain down past the larynx to near the heart, around the ductus arteriousus and then back up to the larynx. This is an unecessary detour in all mammals but a remarkably bizarre one in giraffes.
These examples make sense from the point of view of evolution because these structures are seen as the result of modifying structures inherited from ancestors way back in time (where the structures were functional, or were not suboptimal in the way that they have become).
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Rex Monday
None but a blockhead
# 2569
|
Posted
quote:
Originally posted by sharkshooter:
Would it have been necessary for God to have created each species completely different for people to believe that He created them? So, for example, no two species with the same number of vertebrae, no two species with the same number or design of eyes, no two species with the same basic limb design, etc.
Not at all. But whatever created all living things has arranged things so that there is no species with a feature that could not easily - and in almost all cases obviously -- have evolved from an earlier feature, even when an alternate arrangement would be obviously advantageous.
It's not a case of a creator god having to design all land vertebrates (say) with different basic limb designs: it's a case of there being no land vertebrates with different basic limb designs. Newt, bat, snake, whale and man are obviously fellows.
There is no species of eight-legged mouse. Were there to be one, and were it clearly impossible for its skeletal structure to have evolved from the common mammalian ancestor, then that would be a clear indication that evolution as we understand it is wrong.
Yet even where there are gross differences in basic design - say between the invertebrates and the vertebrates - there is genetic evidence pointing back to a common (and very distant) ancestor.
If you believe in a creator god, then all the evidence is that the act of creation is through evolution.
R
--------------------
I am largely against organised religion, which is why I am so fond of the C of E.
Posts: 514 | From: Gin Lane | Registered: Mar 2002
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
quote:
Originally posted by Faithful Sheepdog:
To sum up, I think the direct analogy between evolutionary algorithms and biological evolution breaks down at the point of the programmed selection constraints. Natural selection in biology is probably too complex a phenomenon to be modelled accurately at present.
Natural selection can be very simple or very complex depending on the factors influencing variation on which it operates and the factors influencing whether or not the organism gets to reproduce. Rather than computer modelling it there have been many direct experiments to demonstrate it, from work on showing how bacteria can acquire enzymic activity they did not previously have, through to alteration in fish sizes as a result of predation and many more.
There is an argument put forward by some intelligent design advocates that says that selection operating on random variation cannot produce novelty. The point about the evolutionary algorithms is that they demonstrate that it can. So: Rest-In-Peace one argument against evolution by natural selection.
You may still want to maintain that evolution from common descent by natural selection has NOT given us the biological diversity that we have today, but in that case you will have to target some other element of the theory.
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
sharkshooter
Not your average shark
# 1589
|
Posted
quote:
Originally posted by Glenn Oldham:
alteration in fish sizes as a result of predation
Ah, bigger (or smaller) fish - evidence for evolution. Gotcha.
--------------------
Let the words of my mouth, and the meditation of my heart, be acceptable in thy sight, O LORD, my strength, and my redeemer. [Psalm 19:14]
Posts: 7772 | From: Canada; Washington DC; Phoenix; it's complicated | Registered: Oct 2001
| IP: Logged
|
|
sharkshooter
Not your average shark
# 1589
|
Posted
quote:
Originally posted by Rex Monday:
There is no species of eight-legged mouse. Were there to be one, and were it clearly impossible for its skeletal structure to have evolved from the common mammalian ancestor, then that would be a clear indication that evolution as we understand it is wrong.
...
If you believe in a creator god, then all the evidence is that the act of creation is through evolution.
The lack of an eight-legged mouse is not evidence of anything.
--------------------
Let the words of my mouth, and the meditation of my heart, be acceptable in thy sight, O LORD, my strength, and my redeemer. [Psalm 19:14]
Posts: 7772 | From: Canada; Washington DC; Phoenix; it's complicated | Registered: Oct 2001
| IP: Logged
|
|
HenryT
Canadian Anglican
# 3722
|
Posted
quote:
Originally posted by sharkshooter:
The lack of an eight-legged mouse is not evidence of anything.
The common body pattern of four limbs is not, as sharkshooter says, evidence of anything. It is, however, a phenomenon that needs to be accounted for in any theory of how things came to be as they are. Evolution has a simple answer, conforming to Ockham's famous heuristic.
Any form of creationism, including Intelligent Design, pretty much has to account for why the common plan is used when something else would make more sense. This occurs, IMO, at both micro- and macro- levels. Human sinuses are a small but annoying example. Residual flippers in whales and the snake's pelvis are more macro level
instances.
Let's see the ID arguments other than "irreducible complexity." What makes you gasp and say "that was designed"?
--------------------
"Perhaps an invincible attachment to the dearest rights of man may, in these refined, enlightened days, be deemed old-fashioned" P. Henry, 1788
Posts: 7231 | From: Ottawa, Ontario, Canada | Registered: Dec 2002
| IP: Logged
|
|
ken
Ship's Roundhead
# 2460
|
Posted
The point about the homologies is that they nest within groups.
You can draw trees connecting particular characters of organisms (genetic, anatomical, metabolic, whatever).
And you find that when one character is shared by many or most of one group of organisms, and not by another, then so are many other characters as well.
So mammals are not just identifiable by milk, hair, 7 neck vertebrae and so on, but by hundreds of characters we share with each other, but not with any other organisms.
And within mammals, a group like the primates will share some characters.
Recently we've been finding large numbers of DNA sequences that are also shared.
It it possible to connect these into trees as well. On a large scale These trees almost always show the same patterns of descent as anatomy does.
Where the DNA evidence has changed our view of phylogenetics it nearly always supports hypotheses put forward by at least some taxonomists.
And these are testable hypotheses. When a taxonomist puts two organisms into the same taxon [named group of organisms] and a third in a different one, they are making the hypothesis that those two organisms are more closely related to each other by common descent - that is real kinship, not just similarity.
And that leads to testable hypotheses. When two organisms placed in once group by anatomy, and a third elsewhere, we can predict that they are also more closely related by gene sequence. And that incoludes so-called "junk" DNA. And (this is a good one) their parasites tend to be more closely related as well.
Overwhelmingly often, these predictions are on the whole borne out. For example, gorillas and chimpanzees really do have gene sequences that are more similar to each other than to
Sometimes the predictions are found to be false. For example chimps are clearly more closely related to humans than either are to gorillas - however most anatomists used to put chimps and gorillas in in once group, and ourselves in another.
On the whole such readjustments have been small-scale, involving closely related and very similar organisms. There have been very few cases where an organism has been found to have been placed in the wrong large-scale taxon.
What has happened more often is thatsequence evidence reinforces the hypothesis of one large group that was difficult to resolve using other characters. Relationships between the large groups - families, classes, phyla - are often easier to resolve with sequence evidence.
Posts: 39579 | From: London | Registered: Mar 2002
| IP: Logged
|
|
Rex Monday
None but a blockhead
# 2569
|
Posted
quote:
Originally posted by sharkshooter:
quote:
Originally posted by Rex Monday:
There is no species of eight-legged mouse. Were there to be one, and were it clearly impossible for its skeletal structure to have evolved from the common mammalian ancestor, then that would be a clear indication that evolution as we understand it is wrong.
...
If you believe in a creator god, then all the evidence is that the act of creation is through evolution.
The lack of an eight-legged mouse is not evidence of anything.
I didn't say it was: it was an illustrative point showing, I hope, that you wouldn't need every creature designed differently to point to there being a creator. Just one would do, either alive or in the fossil record.
We're missing that one.
On the other hand, we have a great many creatures with features that look like bodge jobs from earlier versions, from the macro to the micro.
Evolution accounts for both these observations, the idea of direct creation accounts for neither and would seem somewhat to contradict them.
R
--------------------
I am largely against organised religion, which is why I am so fond of the C of E.
Posts: 514 | From: Gin Lane | Registered: Mar 2002
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Originally posted by Glenn Oldham:
There is an argument put forward by some intelligent design advocates that says that selection operating on random variation cannot produce novelty. The point about the evolutionary algorithms is that they demonstrate that it can. So: Rest-In-Peace one argument against evolution by natural selection.
I’ll repeat what I said to Rex Monday earlier. Evolutionary algorithms are an efficient numerical search pattern for optimising a multi-parametric problem, but they are not actually creating anything new at all. They are simply optimising a string of numbers against programmed constraints.
I think the reported death of the argument against natural selection is greatly exaggerated. ![[Smile]](smile.gif)
quote:
Glenn Oldham said:
Why is the designer so fond of the pentadactyl limb even keeping the bones for two separate fingers when they are joined in the same digit? The pentadactyl limb surely cannot be the best design for all these uses. Nelson and Wells comment that “the suboptimality of the pentadactyl limb pattern has never been empirically demonstrated” - a comment that beggars belief. Apparently the pentadactyl limb is the best design for every one of these uses. It clearly is not in the case of the horse, which has but one toe, – but then why does the horse possess the vestiges of the ones it does not use?
Your post on homology contained a lot of interesting information on animal anatomy and physiology which I will take as a given. The facts are not in dispute, rather it is the interpretation to be made of them. Homology was known about long before Darwin, but the similarities were seen as evidence of archetypes and the common hand of the Creator.
The paragraph that I have selected above shows a common line of argument in writing about Darwinian evolution. It can be found in Darwin’s words (as quoted in your post) and in more recent times, Stephen J. Gould has made a similar argument with his comments about “highly inefficient” Pandas’ thumbs.
They and many others make the same fundamental argument, but it should be noted that the argument is a mixture of science and theology. There are two distinct problems here.
The first is a problem that you chose to gloss over: the empirical demonstration that a given biological anatomy is in fact suboptimal from a scientific perspective. It is asserted, but not demonstrated, that apparently homologous structures are less than perfect (or suboptimal). This presumes that we have access to what would be truly optimal, and can define “perfection” in scientific rather than aesthetic terms.
In the light of what mechanical engineers can actually achieve in artificial anatomical forms, I think this is a very bold claim. I refer you to the list of events in the Robot Olympics to gauge our current level of anatomical engineering.
I would argue that we simply don’t have the data on which to do a scientific comparison, except in a few rare circumstances. As my archetypally dour Scottish dentist once said to me, in connection with a tooth crown, “we’re a very poor second to the Almighty”.
The second problem is that the argument is clearly a theological one. God could not have done it this way, because it would be against what we understand God to be. And if by some chance he did work this way, he’d in any case do a far better job of it. Since this can’t be God’s work, then some other naturalistic mechanism must be found.
At this point we have left the scientific realm far behind, and are well into theology and metaphysics, not to mention the logical fallacy of a false dichotomy.
For a further development of this point, I refer to another article by Paul Nelson at the Access Research Network website. He explores in some depth the theological basis for some evolutionary arguments. Darwin used them frequently, ostensibly to protect the dignity of the Creator, but Darwin’s successors have used them to make the Creator redundant.
You have mentioned several times the giraffe’s neck, and the fact that the laryngeal nerve (presumably the nerve that controls the larynx) travels all the way from the brain down to the near the heart and then back up the neck to the larynx. You opine that this nerve actually takes a very inefficient route, travelling many metres more than necessary, evidence not of an intelligence at work, but of descent with modification from a homologous structure in similar but short necked animals.
A response from an intelligent design perspective would be to examine the engineering claim that this lengthy route for the nerve is in fact inefficient or sub-optimal. Your comment here presumes that optimality would mean the shortest travel distance, or at least, a significantly shorter one than at present, but this ignores all other design criteria apart from length. The onus is on you to demonstrate the inefficiency of the present arrangement, rather than to assert it.
I would turn the question on its head. What benefits accrue to the giraffe from this apparently unusual nerve configuration? Sadly I don’t have the knowledge of animal anatomy or electrical micro-circuits to take this further. However, it does strike me as an interesting research exercise. It is certainly a further example that an intelligent design approach can stimulate scientific questions.
Neil
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
Justinian
Shipmate
# 5357
|
Posted
quote:
Originally posted by Faithful Sheepdog:
quote:
Originally posted by Glenn Oldham:
There is an argument put forward by some intelligent design advocates that says that selection operating on random variation cannot produce novelty. The point about the evolutionary algorithms is that they demonstrate that it can. So: Rest-In-Peace one argument against evolution by natural selection.
I’ll repeat what I said to Rex Monday earlier. Evolutionary algorithms are an efficient numerical search pattern for optimising a multi-parametric problem, but they are not actually creating anything new at all. They are simply optimising a string of numbers against programmed constraints.
I think the reported death of the argument against natural selection is greatly exaggerated.
And animals are not different. They are simply strings of DNA made flesh in a constrained environment.
Constrained systems can come up with new solutions- one case comes to mind where there was an attempt to see how to make a tone detector using the lowest possible number of parts in an electric circuit- and it came up with a solution using fewer than the theoretical minimum number of parts- and a solution in which some of the necessary parts were not even part of the circuit- meaning that in some cases it was the physics of the individual cells that was affecting the circuit, not their properties as part of the circuit.
How's that for creation of something new via a genetic algorithm?
The paper in question:
http://www.cogs.susx.ac.uk/users/adrianth/ices96/paper.html
The relevant page:
http://www.cogs.susx.ac.uk/users/adrianth/ices96/node5.html
quote:
Your post on homology contained a lot of interesting information on animal anatomy and physiology which I will take as a given. The facts are not in dispute, rather it is the interpretation to be made of them. Homology was known about long before Darwin, but the similarities were seen as evidence of archetypes and the common hand of the Creator.
It's not just the homologies that are the question- a creator would probably use them too. It's the stupid homologies that are the problem. If God is responsible for e.g. the giraffe's nerve going five metres out of its way, then he's either lazy or stupid. See also: tonsils (and any other vestigial organ).
quote:
They and many others make the same fundamental argument, but it should be noted that the argument is a mixture of science and theology. There are two distinct problems here.
Science needs to be used for evolution and theology to argue with creationists. Your point?
quote:
The first is a problem that you chose to gloss over: the empirical demonstration that a given biological anatomy is in fact suboptimal from a scientific perspective. It is asserted, but not demonstrated, that apparently homologous structures are less than perfect (or suboptimal). This presumes that we have access to what would be truly optimal, and can define “perfection” in scientific rather than aesthetic terms.
Then please explain in what sense the giraffe's nerves going 5m out of their way is anything like optimal under any scale. Please explain how tonsils or the appendix (or any other vestigial organ) are optimal.
quote:
I would argue that we simply don’t have the data on which to do a scientific comparison, except in a few rare circumstances. As my archetypally dour Scottish dentist once said to me, in connection with a tooth crown, “we’re a very poor second to the Almighty”.
We don't need many cases. One would be enough.
quote:
The second problem is that the argument is clearly a theological one. God could not have done it this way, because it would be against what we understand God to be. And if by some chance he did work this way, he’d in any case do a far better job of it. Since this can’t be God’s work, then some other naturalistic mechanism must be found.
Wrong. The argument for evolution is a scientific one. The one against creationism is a theological one as the creationists have brought God into the picture, and once God is there he needs to be dealt with. If the creationists would just give up and go away, the theological arguments would be lost- but because people keep bringing the subject up, it is considered necessary to protect against them.
Either that or you can have a stupid God who does random things. I can't falsify that- but Occam's Razor tends to indicate against it. I also suspect you don't believe in a stupid God.
At this point we have left the scientific realm far behind, and are well into theology and metaphysics, not to mention the logical fallacy of a false dichotomy.
quote:
You have mentioned several times the giraffe’s neck, and the fact that the laryngeal nerve (presumably the nerve that controls the larynx) travels all the way from the brain down to the near the heart and then back up the neck to the larynx. You opine that this nerve actually takes a very inefficient route, travelling many metres more than necessary, evidence not of an intelligence at work, but of descent with modification from a homologous structure in similar but short necked animals.
A response from an intelligent design perspective would be to examine the engineering claim that this lengthy route for the nerve is in fact inefficient or sub-optimal. Your comment here presumes that optimality would mean the shortest travel distance, or at least, a significantly shorter one than at present, but this ignores all other design criteria apart from length. The onus is on you to demonstrate the inefficiency of the present arrangement, rather than to assert it.
I would turn the question on its head. What benefits accrue to the giraffe from this apparently unusual nerve configuration? Sadly I don’t have the knowledge of animal anatomy or electrical micro-circuits to take this further. However, it does strike me as an interesting research exercise. It is certainly a further example that an intelligent design approach can stimulate scientific questions.
Anything can stimulate scientific questions- but if it's electrical impulses you want, there's the entire spinal cord round there. You could easily move another nerve that did have to go that way rather than waste 5m of nerve.
--------------------
My real name consists of just four letters, but in billions of combinations.
Eudaimonaic Laughter - my blog.
Posts: 3926 | From: The Sea Coast of Bohemia | Registered: Dec 2003
| IP: Logged
|
|
sharkshooter
Not your average shark
# 1589
|
Posted
quote:
Originally posted by Justinian:
You could easily move another nerve that did have to go that way rather than waste 5m of nerve.
Ah - I am now getting the picture. Since God didn't do it the way you (generic) would have, He obviously didn't do it at all. Perhaps if He was as smart as man is, He might done it your way, and then, perhaps you would believe that He could be the Creator.
By the way, "waste" presumes use of materials of some sort, normally of limited supply. Creation from nothing implies an unlimited supply of materials. So, an extra 5m of nerve is irrelevant in creation.
--------------------
Let the words of my mouth, and the meditation of my heart, be acceptable in thy sight, O LORD, my strength, and my redeemer. [Psalm 19:14]
Posts: 7772 | From: Canada; Washington DC; Phoenix; it's complicated | Registered: Oct 2001
| IP: Logged
|
|
Rex Monday
None but a blockhead
# 2569
|
Posted
quote:
Originally posted by Faithful Sheepdog:
I’ll repeat what I said to Rex Monday earlier. Evolutionary algorithms are an efficient numerical search pattern for optimising a multi-parametric problem, but they are not actually creating anything new at all. They are simply optimising a string of numbers against programmed constraints.
Neil
Perhaps instead of repeating it, you could explain it? I don't understand what you're saying.
I think it boils down to you using a different definition of 'new' to the one I'm used to.
Let me define what I mean by 'new': by 'new', I mean something that's different to what's gone before, something that's here when before it wasn't here, something with novelty, something that's changed from the previous state.
With genetic algorithms this 'new' is created by randomising, and in living organisms this 'new' is created by mutation. In both cases, something appears that wasn't there before: this is a new thing.
Are you saying that this new thing that is created by randomness isn't in fact new?
What, then, do you mean by 'new'?
R
--------------------
I am largely against organised religion, which is why I am so fond of the C of E.
Posts: 514 | From: Gin Lane | Registered: Mar 2002
| IP: Logged
|
|
Rex Monday
None but a blockhead
# 2569
|
Posted
quote:
Originally posted by sharkshooter:
quote:
Originally posted by Justinian:
You could easily move another nerve that did have to go that way rather than waste 5m of nerve.
Ah - I am now getting the picture. Since God didn't do it the way you (generic) would have, He obviously didn't do it at all. Perhaps if He was as smart as man is, He might done it your way, and then, perhaps you would believe that He could be the Creator.
By the way, "waste" presumes use of materials of some sort, normally of limited supply. Creation from nothing implies an unlimited supply of materials. So, an extra 5m of nerve is irrelevant in creation.
Just out of interest, what would you accept as evidence for evolution?
R
--------------------
I am largely against organised religion, which is why I am so fond of the C of E.
Posts: 514 | From: Gin Lane | Registered: Mar 2002
| IP: Logged
|
|
Justinian
Shipmate
# 5357
|
Posted
quote:
Originally posted by sharkshooter:
quote:
Originally posted by Justinian:
You could easily move another nerve that did have to go that way rather than waste 5m of nerve.
Ah - I am now getting the picture. Since God didn't do it the way you (generic) would have, He obviously didn't do it at all. Perhaps if He was as smart as man is, He might done it your way, and then, perhaps you would believe that He could be the Creator.
By the way, "waste" presumes use of materials of some sort, normally of limited supply. Creation from nothing implies an unlimited supply of materials. So, an extra 5m of nerve is irrelevant in creation.
Wrong on both counts. The duckbilled platypus wasn't done the way I'd do it- but that's not evidence that God didn't make that thing. There's doing something a way I wouldn't, which is no problem, and there's doing something a stupid way.
Also the extra 5m of nerve may be irrelevant to God, but it's certainly relevant to the giraffe who has to grow the thing- or don't you believe you have to eat to get the energy? (Actually, I think the extra fingerbones in certain animals paws and wings are a better example than the nerve).
--------------------
My real name consists of just four letters, but in billions of combinations.
Eudaimonaic Laughter - my blog.
Posts: 3926 | From: The Sea Coast of Bohemia | Registered: Dec 2003
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Originally posted by Rex Monday:
quote:
Originally posted by Faithful Sheepdog:
I’ll repeat what I said to Rex Monday earlier. Evolutionary algorithms are an efficient numerical search pattern for optimising a multi-parametric problem, but they are not actually creating anything new at all. They are simply optimising a string of numbers against programmed constraints.
Neil
Perhaps instead of repeating it, you could explain it? I don't understand what you're saying.
I think it boils down to you using a different definition of 'new' to the one I'm used to.
Let me define what I mean by 'new': by 'new', I mean something that's different to what's gone before, something that's here when before it wasn't here, something with novelty, something that's changed from the previous state.
With genetic algorithms this 'new' is created by randomising, and in living organisms this 'new' is created by mutation. In both cases, something appears that wasn't there before: this is a new thing.
Are you saying that this new thing that is created by randomness isn't in fact new?
What, then, do you mean by 'new'?
R
Fair enough, I think the word “new” is being used in various senses here, so I’ll try and clarify.
Firstly, there is a sense in which the optimised design that emerges from the evolutionary algorithm can rightly be called new, since that particular combination of parameters has probably never seen the light of day before. The string of N parameters that defines the engineering problem began life as something less efficient, and the algorithm has mutated them into something more efficient. We therefore have a new set of numbers representing an optimised design. I’m not denying for a moment that this may be very useful information to the engineer.
However – and this is the but – the algorithm cannot break away from the original N parameters and the programmed constraints to give us the answer to a problem with N+M parameters and a different set of constraints. The program begins with N parameters for a specific engineering problem, and finishes with N parameters – for that same problem. So in that sense nothing has changed. We simply have a more desirable set of numbers than we started with.
I’ll try and illustrate further with a simple example.
If you imagine a problem with 3 parameters for simplicity (i.e. N=3), then the parameters can be visualised as the x, y and z axes of a 3-dimensional graph. The constraints applied by the engineer will determine the applicable regions of each axis where an optimised solution may be found. For example, x may be constrained to lie between 10 and 1000, y must be greater than 0 but less than 50, and z greater than 500 but less than 10000. These limits define a search space.
At each point within this search space, the engineer defines a fitness function which is related to the values x, y and z. For example, if x, y and z values relate to mass, then total mass m = x + y + z.. Other fitness functions (e.g. size) may also be applicable and likewise some function of x, y and z. For each point in the xyz search space, these individual fitness functions are then combined in a programmed fashion to give an overall fitness value.
So at the start, even before the algorithm has begun “number crunching”, all fitness values in the xyz space have been defined in principle by the programmed fitness function(s) and all the other constraints. There is a sense in which the program knows in advance the fitness values for all possible x, y and z values. The only problem remaining is the exact x, y and z location(s) of maximum fitness within the permitted region.
The xyz search space may be so large that it is unclear where to start. This is where the evolutionary algorithms come in, with randomised sets of start parameters that amount to a random scatter of points in the permitted xyz search space. Through the “breeding” process of recombination and mutation, the algorithm can generate further sets of parameters and search the space much more rapidly, always looking for the optimised point(s).
It is important to note that the process is programmed to find the optimum. What is optimum is defined completely by the engineer. The algorithm is written on the presumption that an optimum exists and so goes looking for it until it finds it. The engineer has defined both the problem (where is the optimum?) and its solution (an efficient search algorithm). The numbers may be new, but the creativity belongs to the engineer. The algorithm is simply a creative tool in his or her hands.
I hope that that’s a little clearer, but let me know if it’s not.
Although it’s quite a complex paper, I can recommend the link provided by Justinian above to a further example of the application of evolutionary algorithms in electronic micro-circuits. This problem was defined by 1800 bits of information. I have printed this out to study it further and may come back with some further comments.
Neil
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
quote:
Originally posted by Faithful Sheepdog:
It is important to note that the process is programmed to find the optimum. What is optimum is defined completely by the engineer. The algorithm is written on the presumption that an optimum exists and so goes looking for it until it finds it. The engineer has defined both the problem (where is the optimum?) and its solution (an efficient search algorithm). The numbers may be new, but the creativity belongs to the engineer. The algorithm is simply a creative tool in his or her hands.
I hope that that’s a little clearer, but let me know if it’s not.
So whats to stop you from regarding evolution by natural selection as exactly parallel to this with God as the engineer? In such a case God could be seen as having set up the parameters: the laws of physics and chemistry, and once self replicating entities get going then random variation plus selection by constraints in the environment results in the new configurations being found and succeeding better and then these in turn evolving. As they increase in complexity they impose new selection constraints on each other (predator prey relationships are just one example) and so it continues.
There is nothing whatsoever in your account of the evolutionary algorithms so far that shows that natural selection is and incoherent idea, or that it has limits that would prevent it from generating the range of organisms we see today from a common ancestor.
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
I'll get back to you on the homology matter, Neil, after I get a chance to read the 10,000 word paper by Nelson that you posted.
I would, however, have liked a response to things like the fact that whales and anteaters as foetuses go through a stage where they have teeth. These are especially clear examples of sub-optimality that demand a coherent explanation from Intelligent Design theory. These kinds of facts seem to me to be simply incoherent without the concept of common descent.
The 'who are we to say what God would do' argument is a tricky one for anti-evloutionists to use, since they clearly believe that they can say that God would not have used evolution as a method of creation.
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Originally posted by Glenn Oldham:
I would, however, have liked a response to things like the fact that whales and anteaters as foetuses go through a stage where they have teeth. These are especially clear examples of sub-optimality that demand a coherent explanation from Intelligent Design theory. These kinds of facts seem to me to be simply incoherent without the concept of common descent.
I think any discussion of apparent sub-opimality must take into account all stages of the animal's life. An anatomical feature which is puzzling in a mature specimen may be more explainable at a younger stage - and vice versa. To use another engineering example, the anatomical feature may be "temporary works", essential at a younger stage, but no longer critical for survival past maturity.
Caterpillars turn into butterflies. Crabs must shed their skin - becoming temporarily very vulnerable - in order to grow bigger. Humans lose the teeth they are born with and grow another set. Our lungs were full of water whilst in the womb, but once born they must become full of air for survival.
The natural world is indeed full of amazing feats and remarkable changes. What is the meaning of nascent teeth in whales and anteater foetuses? I'm not sure at present, but I'll think on. I'm going to be away from my computer for a long weekend, so I'll get back to this thread properly next week.
Neil
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
ken
Ship's Roundhead
# 2460
|
Posted
2 things:
1) sub-optimality isn't really important for the homology argument, its just a strong clue. The big deal is the way different characters nest in phylogenetic trees.
And you can't seriously doubt that there are many sub-optimal characters in living things. Ever had gout? That's because you, and every other human (every other ape in fact) has a broken gene that codes for an enzyme to handle purines.
2) Search space arguments neat to be taken with a pinch of salt. Physically inclined people tend to like them more than biologists do. But the numbers are so big its hard to get anywhere with them.
The search space for an individual protein coded for by DNA is larger than the likely number of particles in the universe.
In fact there are self-modifying genes and proteins in your immune system that have search spaces larger than the likely number of particles in the universe.
Forgen mere astronomical numbers - biological numbers are BIG. That confuses physicists and philosophers who like to keep things simple. Biology isn't simple.
Biological "laws" aren't like physical laws, still less like mathematical laws. They are hypotheses about the probability of outcomes, based on observation of actual phenomena. All is contingent. Everything is stochastic. Biology really is Natural History, not Natural Philosophy.
--------------------
Ken
L’amor che move il sole e l’altre stelle.
Posts: 39579 | From: London | Registered: Mar 2002
| IP: Logged
|
|
Justinian
Shipmate
# 5357
|
Posted
quote:
Originally posted by Faithful Sheepdog:
Although it’s quite a complex paper, I can recommend the link provided by Justinian above to a further example of the application of evolutionary algorithms in electronic micro-circuits. This problem was defined by 1800 bits of information. I have printed this out to study it further and may come back with some further comments.
Part of the point is that the optimal solution wasn't just defined by the 1800 bits of information- the solution also involved the physical properties of the board- making for a more efficient solution, but one that was untransferable and completely breaking the chance of encoding the information in terms of a relatively small number of bits. It was also a solution that the creator would never have come up with.
--------------------
My real name consists of just four letters, but in billions of combinations.
Eudaimonaic Laughter - my blog.
Posts: 3926 | From: The Sea Coast of Bohemia | Registered: Dec 2003
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
![[Hot and Hormonal]](icon_redface.gif) HUMBLE PIE ALERT
A CORRECTION AND AN APOLOGY
I have on three separate occassions on this thread stated that all mammals have seven cervical vertebrae. This is not correct. Further enquiry on my part reveals that although the vast majority of mammal species (about 99.9% of them) have seven vertebrae, there are two orders of mammals that depart from this pattern: the order Edentata (edentates) which includes anteaters armadillos and sloths; and the order Sirenia ("sea cows," dugongs, and manatees).
I apologise to shipmates for this error and mistatement.
In terms of the overall argument, however, the great uniformity of the number of cervical vertebrae in the majority of mammal species still calls for an explanation. Bird and Reptiles vary considerably more in this aspect of their anatomy.
Glenn ![[Eek!]](eek.gif)
[ 03. July 2004, 19:58: Message edited by: Glenn Oldham ]
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
So far Neil has posted links to three articles by Paul Nelson (two written with Jonathan Wells). They all relate to homologies and evolution by descent from a common ancestor. It is about time I said something about them. I’ll start with the last one first. It can all be found on the Access Research Network Jettison the Arguments, or the Rule? The Place of Darwinian Theological Themata in Evolutionary Reasoning;
This 1998 article by Nelson is a revised version of one entitled “The Role of Theology in Current Evolutionary Reasoning” published in 1996 in the journal Biology & Philosophy and which has been included in the book Intelligent Design Creationism and Its Critics - Philosophical, Theological, and Scientific Perspectives Edited by Robert T. Pennock (2001) where it is immediately followed by an excellent critique of it by Kelly C. Smith entitled “Appealing to Ignorance Behind the Cloak of Ambiguity” to which, alas, I have been unable to find a link. ( Table of Contents .)
Nelson’s article has two aims. One is to show that the arguments for evolution from homologies and imperfections in design do not rule out some forms of intelligent design theory. The other is to subvert the argument altogether by attacking what (mistakenly) he understands methodological naturalism to entail. I will just respond to his arguments against the argument from homologies and imperfections and leave the issue of methodological naturalism for the time being (it has been much discussed on this thread already).
So, let us keep in mind that the challenge intelligent designers face is to explain features of the world such as these:
- that in some species of whales and in anteaters the foetus develops teeth which they loose before they are born and never get to use, the adults having no teeth at all during their adult lives;
- that there no known functional reason why the giraffe’s recurrent laryngeal nerve does not go direct to the larynx but travels 5 metres out of its way to near the heart and back up to the larynx; and
- the vast majority of mammals have 7 neck vertebrae despite the fact that the necks of those mammals vary so greatly in their length and in the uses to which they are put.
Nelson says that the argument for evolution from features like these (call each such feature ‘p’) goes something like this:
quote:
1. If p is an instance of organic design, then p was produced either by a wise creator, or by descent with modification (evolution).
2. If p (an instance of organic design) was produced by a wise creator, then p should be perfect (or should exhibit no imperfections).
3. Organic design p is not perfect (or exhibits imperfections).
[therefore]
[4] Organic design p was not produced by a wise creator, but by descent with modification. Some organic designs are evidence of evolution.
Nelson offers a number of challenges to this argument.
1) He says that ID theory allows the idea that the organs of some creatures may lose some of their functionality as a result of degeneration. This would explain blind cave animals still having eyes, and flightless birds and beetles still having wings. Well, broadly speaking, that is fair enough – it is what evolutionary theory predicts anyway.
2) In the earlier version of the paper Nelson also mentioned some creationists who take the view that there has been microevolutionary change in the basic types that were originally created and that this could account for some homologies. (Oddly he excludes this from his later version of the paper). Again, fair enough, that is using part of evolutionary theory again anyway - inheritance from a common ancestor.
He notes that the mechanisms in (1) and (2) can’t account for all the kinds of features raised by evolutionist arguments. He therefore offers more.
3) God may not be the kind of God that the argument takes him to be. God might have limited power. (Nelson does not think so, so I am not sure how this helps him).
4) God may be limited by the larger picture. That is to say, God may not be able to realise the goal of perfect organisms and still be able at the same time to achieve the perfection of a larger part (or the whole) of creation.
quote:
On this view, any judgment of perfection or imperfection must be qualified with a proviso that perfection -- defined as divinely created perfection -- can be judged only on the scale of the whole creation. And there is no reason for a creator to optimize one part of the universe at the expense of the whole.
quote:
the finitude of human scientific observation may lead us to infer mistakenly that an organic design (e.g., the panda's pseudothumb) is imperfect, when its imperfection is only apparent, that is, local.
Now there is no doubt truth in this approach. It is the kind of view that that theistic evolutionists need to appeal to. But I do not think that it covers all the remaining examples. Might Nelson be able to tell us what part of the universe would have suffered if the giraffe’s laryngeal nerve was shorter, or if foetal whales and anteaters did not develop and lose the teeth that they never use as adults? Perhaps he would point to these comments of his:
quote:
The creator could have been limited in some way by unknown "compossibility" constraints. In crudest outline, a compossibility analysis would ask whether all possibilities are mutually consistent. … One may have failed to identify the correct reference situation by which to judge the design (perhaps by looking at too narrow a slice of … life history). The flippers of marine turtles, for example, strike us as rather badly designed for digging holes in beach sand to place eggs. The same flippers, however, perform efficiently in the water, where the turtles spend most of their time. Which reference situation takes precedence in an optimality analysis?
This is an important point and one that evolutionists take into account is assessing adaptations. Is Nelson implying that foetal whales and anteaters somehow need to did develop and lose the teeth that they never use as adults because of unknown "compossibility" constraints? Why should we believe that? Evolutionary theory offers clear reasons for such constraints: the fact that evolution comes about as variation in existing systems and structures leads, through selection to their being adapted and modified often showing traces of what they were modified from. What does ID theory have to offer in the place of this? Who knows?
5) When Nelson looks at the homology issue he offers two defences of the Intelligent Design option. The first is that he challenges the view that these common features (like the pentadactyl limb) are not the optimal designs.
I have to confess that I still find this somewhat breathtaking. Let’s take the fact that the vast majority of mammals (see my immediately previous post) have seven cervical (neck) vertebrae. This applies to those with short rigid necks, long slender necks, medium sized necks, necks that support heads that rotate from side to side, necks that are attached to heads that don’t; necks of tall grazing animals that have to lower them to graze (horses, cows), necks of the big cats, necks of elephants, necks of those animals that butt each other in courtship contests, necks of bats, necks of moles, of giraffes. All of these have seven vertebrae. Paul Nelson is asking us to believe that in every single one of these cases seven is the optimal number. Not five or six or four, but seven. But despite biophysics being a rich field, as far as I am aware we know of no physical principle that would suggest that seven vertebrae would be optimal in these cases. We also know that birds sometimes occupying much the same ecological niches as mammals (on islands where mammals are absent for example) have necks with 9 to 25 vertebrae. Why are they so different from mammals in this respect?
In short, the idea that in all these cases the number of the vertebrae is optimal seems unsupported. We may be wrong about this, but it is hard to see why. An evolutionary explanation for the pattern has been offered (see Why do Mammals have but seven cervical vertebrae?;
6) The second defence is that he says that we can’t discount the possibility that the creator simply choose to repeat the same design in different species.
Nelson is not clear how far he would press this one, but it is not clear that it would cover foetal teeth in whales and anteaters, or the giraffe’s laryngeal nerve. It is always logically possible, of course, that God’s aesthetic sense is so different from ours that it is for artistic reasons that these oddities are there. But we are moving into the realms of ‘who are we to say what God might do?’ As I have pointed out earlier, this question is hardly a help to those against evolution by natural selection, since it challenges all views equally. Who are we to say that God did NOT use evolution by natural selection to make the world?
Conclusion
In short Nelson’s argument comes to this: that the argument from imperfections and homologies does not rule out that there might be a version of Intelligent Design Theory that explains them. That version is not the one that Darwin was arguing against (the one he argued against was the one based on William Paley’s views).
This is a logical and reasonable conclusion. But how likely is it that there is such a version of the Intelligent Design Theory? And if one exists, how likely is it that it is true, compared to how likely evolution by natural selection is true?
A major problem facing ID at the present time is that it lacks any clear theory. The movement has been largely content to argue against evolution by natural selection, but has not offered any clear alternative. (For criticism of this see Robert Pennock’s article DNA by Design?: Stephen Meyer and the Return of the God Hypothesis a link to which is on his web page .
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
ken
Ship's Roundhead
# 2460
|
Posted
quote:
Originally posted by Glenn Oldham:
quote:
The creator could have been limited in some way by unknown "compossibility" constraints.
True of course, but incapable of distinguishing special creation from evolution, because it applies equally to both.
Did Nelson actually address the nature of homology and nested characters? Or just the idea of sub-optimal characters?
--------------------
Ken
L’amor che move il sole e l’altre stelle.
Posts: 39579 | From: London | Registered: Mar 2002
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
quote:
Originally posted by ken:
quote:
Originally posted by Glenn Oldham:
quote:
The creator could have been limited in some way by unknown "compossibility" constraints.
True of course, but incapable of distinguishing special creation from evolution, because it applies equally to both.
Indeed so. As I said in my post ‘Evolutionary theory offers clear reasons for such constraints: the fact that evolution comes about as variation in existing systems and structures leads, through selection to their being adapted and modified often showing traces of what they were modified from. What does ID theory have to offer in the place of this?’
quote:
Originally posted by ken:
Did Nelson actually address the nature of homology and nested characters? Or just the idea of sub-optimal characters?
No, not really, and that is a great failing. He seems to have three lines of approach on this issue. The two I mentioned
- he challenges the idea that they are sub-optimal;
- he says that God might have just chosen to do it that way (which could be extended to cover the nested nature of the homologies); and
- in his other two articles he basically says: prove that the homologies are a result of common descent, show us the mechanism, be it genetic, developmental, etc. that is supposed to cause these features.
Homology: A Concept in Crisis; and
Is Common Descent an Axiom of Biology.
I have a response to that issue in hand.
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Originally posted by Justinian (with my emphasis):
Part of the point is that the optimal solution wasn't just defined by the 1800 bits of information- the solution also involved the physical properties of the board- making for a more efficient solution, but one that was untransferable and completely breaking the chance of encoding the information in terms of a relatively small number of bits. It was also a solution that the creator would never have come up with.
Note that the bold comment above is clearly a theological one. It is based on a clear understanding of what the creator cannot do. That in turn has to be based on some conception of what the creator is and how he will act. The content of this comment may or may not be justified, but the argument in this comment is not in the least bit scientific.
I’ve now been able to study in more detail Justinian’s example of the use of evolutionary algorithms. The problem concerned evolving a very specific functionality to a small portion of a “Programmable Field Gate Array (PFGA)”, which I presume is some kind of electronic microchip.
The functionality required was to differentiate clearly between a 1KHz and a 10kHz square wave input signal. A 1KHz signal should generate a constant output of +5 volts, whereas a 10kHz signal should generate 0 volts. The chip had no access to any kind of timing clock or reference frequency and had to reply on its internal dynamic properties alone to make the differentiation. In other words, the chip had to be configured to act as a simple tone distinguisher.
The chip was connected directly to a computer, and could be modified and controlled from within the computer by a sequence of numerical parameters, amounting to 1800 bits of information. The computer also controlled the input signals and monitored the output signals of the chip.
The performance of the chip under each configuration was therefore not simply modelled by a set of theoretical equations, but was measured in real time, taking into account all the properties of the chip. This probably simplified enormously the engineers’ task and also enabled the experiment to be more realistic as a result.
Even with this simplification the scale of the engineers’ search task can be demonstrated numerically. A basic equation in information theory is that the information content of a configuration I (in bits) is related to its probability of occurrence p (between 0 and 1) as follows:
quote:
I = logarithm to the base 2 of (1/p), or rearranging gives: p = 1/(2 to the power I)
For 1800 bits of information, this equation gives us a probability of 10 to the power of minus 542 – i.e. there is a huge number of potential configurations, with an infinitesimally small probability of achieving a successful configuration by a random guess. Hence the desirability of an efficient targeted search pattern, such as the evolutionary algorithm.
Note the following comments in the paper on the chip configuration:
quote:
No configuration of the cells can cause the device to be damaged -- it is impossible to connect two outputs together, for instance, because all internal connections are uni-directional. So an evolutionary algorithm can be allowed to manipulate the configuration of the real chip without the need for legality constraints or checking.
In other words, there is no such thing as a highly deleterious mutation. Considered as an animal analogue the chip configuration is essentially unkillable, despite any genetic defects, in contrast to the natural world. The “population” size of 50 remains constant regardless of any mutations, good or bad.
The population size was set at 50 “creatures” (i.e. strings of numerical information) and the first 50 were generated randomly. Thereafter the algorithm did its work. The paper provides much information on the fitness function used in the algorithm and the care taken in the definition of this in order to obviate spurious numerical results. Here is another quote from the paper:
quote:
It is important that the evaluation method -- here embodied in the analogue integrator and the fitness function Eqn. 1 -- facilitates an evolutionary pathway of very small incremental improvements. Earlier experiments, where the evaluation method only paid attention to whether the output voltage was above or below the logic threshold, met with failure. It should be recognised that to evolve non-trivial behaviours, the development of an appropriate evaluation technique can also be a non-trivial task.
So the precise form of the fitness function is critical to the experiment’s success. It is essential that the algorithm can detect very small incremental improvements in fitness and then maintain them for the subsequent generations. Is natural selection in biology that fussy?
Here is a comment from the paper on the first (random) generation:
quote:
The individual in the initial random population of 50 that happened to get the highest score produced a constant +5V output at all times, irrespective of the input. It received a fitness of slightly above zero just because of noise. Thus, there was no individual in the initial population that demonstrated any ability whatsoever to perform the task.
So again, considering the numerical strings as an animal analogue, all the initial “creatures” were unable to demonstrate any fitness at all. Nevertheless they all survived to “breed”. In the cyberspace of a numerical experiment nobody dies young and all can get to breed.
This complete lack of fitness lasted for many generations, but definite progress on fitness was reported after 650 generations. Around generation 1100 the output of the best individual in the numerical population was approximately as desired, but the average was well below perfect fitness.
The overall population fitness then continued to improve until generation 4100 delivered perfect behaviour from all members of the population. A further 1000 generations delivered no discernible change at all – “stasis” had been achieved. The algorithm did not allow for the propagation of deleterious mutations or subsequent genetic drift.
Further improvements were achieved by the intelligent intervention of the experimenters. They removed virtually (i.e. by numerical manipulation) parts of the chip not being used in the evolved circuit. This process demonstrated many unexpected subtleties in the behaviour of the chip that the real-time experiment had allowed for, including electromagnetic coupling independent of direct connections.
Earlier on this thread I commented on the differences between “constrained” and “unconstrained” evolutionary algorithms. I consider that this is an example of unconstrained evolution, starting with an completely unfit (i.e. dead or unsuccessfully breeding) life-form and passing through many unfit stages (but still breeding), until the numbers enter a region where the fitness function starts giving desirable answers. Judging by the following comments in the paper, the authors agree that this was an unconstrained evolutionary process:
quote:
Thus, conventional design always requires constraints to be applied to the circuit's spatial structure and/or dynamical behaviour. Evolution, working by judging the effects of variations applied to the real physical hardware, does not. That is why the circuit was evolved without the enforcement of any spatial structure, such as limitations upon recurrent connections, or the imposition of modularity, and without dynamical constraints such as a synchronising clock or handshaking between modules. This sets free all of the detailed properties of the components to be used in developing the required overall behaviour. It is reasonable to claim that the evolved circuit consequently uses significantly less silicon area than would be required by a human designer faced with the same problem, but such assertions are always open to attack from genius designers.
So, although the engineers involved are to be congratulated on their achievement, this process is very far from being a model of Darwinian evolution in biology. It is a good demonstration of what a well-designed numerical process can achieve in a large multi-dimensional search space.
Neil
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
quote:
Originally posted by Faithful Sheepdog:
quote:
Originally posted by Justinian (with my emphasis):
Part of the point is that the optimal solution wasn't just defined by the 1800 bits of information- the solution also involved the physical properties of the board- making for a more efficient solution, but one that was untransferable and completely breaking the chance of encoding the information in terms of a relatively small number of bits. It was also a solution that the creator would never have come up with.
Note that the bold comment above is clearly a theological one. It is based on a clear understanding of what the creator cannot do. That in turn has to be based on some conception of what the creator is and how he will act. The content of this comment may or may not be justified, but the argument in this comment is not in the least bit scientific.
Justinian did indeed say that: "It was also a solution that the creator would never have come up with." But his point is surely not intended theologically. It is surely intended to emphasise the point that the solution to the problem that the algorithm came up with was NOT a solution designed by the creator of the program.
It is a frequent, and erroneous, objection to these kinds of analogies for evolution that, just because there is a person involved somewhere in the setting up of the program, the solution produced by the program is therefore designed, because there is a person around to label the designer. That, I am sure, is the kind of view that Justinian is attacking.
The particular program that you describe is clearly not a model of biological evolution (it was modelled on natural selection) but it still remains analagous to it. The key question, therefore, is whether the differences between it and nature that you point to make evolution by natural selection improbable. The answer to that is surely 'no'.
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Originally posted by Glenn Oldham:
quote:
Originally posted by Faithful Sheepdog:
quote:
Originally posted by Justinian (with my emphasis):
Part of the point is that the optimal solution wasn't just defined by the 1800 bits of information- the solution also involved the physical properties of the board- making for a more efficient solution, but one that was untransferable and completely breaking the chance of encoding the information in terms of a relatively small number of bits. It was also a solution that the creator would never have come up with.
Note that the bold comment above is clearly a theological one. It is based on a clear understanding of what the creator cannot do. That in turn has to be based on some conception of what the creator is and how he will act. The content of this comment may or may not be justified, but the argument in this comment is not in the least bit scientific.
Justinian did indeed say that: "It was also a solution that the creator would never have come up with." But his point is surely not intended theologically. It is surely intended to emphasise the point that the solution to the problem that the algorithm came up with was NOT a solution designed by the creator of the program.
On further reflection it's possible that I have misunderstood Justinian at this point. He may have meant "creator" in the sense of "human creator", i.e. the engineers on the project, in which case his comment is psychological rather than theological - it presumes a certain constrained mode of human thinking. I am more comfortable with this.
With probabilities of the order of 10E-542, I think it's fair to say that a human engineer, given the same task, may have come up with a very different solution, indeed if at all. Justinian's reference suggests that before the start of the experiment many were very sceptical that any chip configuration giving the desired functionality was going to be possible with the given constraints.
In my original reply I took "creator" in the sense of "supernatural creator", i.e. God, and hence commented on the implied theology in the comment. This follows on from some of the comments in my earlier posts and the linked paper by Nelson on the use of theological arguments (either intentionally or by accident) in the writings of many Darwinists.
Neil
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
Justinian
Shipmate
# 5357
|
Posted
Clarification: Creator in this case referred to the human creator of the experiment rather than God.
Also it was an experiment done physically rather than in computer modelling - computer modelling would not have come up with that solution either - if it came up with a solution at all, the solution would have had all the cells involved actually connected to each other (unlike this one) unless you could model the exact physical properties into the algorithm (which would take a massive amount of information).
IIRC when New Scientist covered this, they also commented that the solution used fewer than the theoretical minimum number of parts to measure the frequencies.
--------------------
My real name consists of just four letters, but in billions of combinations.
Eudaimonaic Laughter - my blog.
Posts: 3926 | From: The Sea Coast of Bohemia | Registered: Dec 2003
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
Neil made a number of comments in his post of 10 July, 2004 16:13 pointing to what he saw as differences between the evolutionary configuration of the microchip and natural selection in biology. None of them, however, imperils the analogy between that process and natural selection.
quote:
Originally posted by Faithful Sheepdog:
In other words, there is no such thing as a highly deleterious mutation. Considered as an animal analogue the chip configuration is essentially unkillable, despite any genetic defects, in contrast to the natural world. The “population” size of 50 remains constant regardless of any mutations, good or bad.
A population of 50 would be considered to be on the brink of extinction. Natural populations are usually much better endowed with thousands to trillions or more members depending on the type of organism. Populations may also be fairly stable in numbers over generations though this is not critical for natural selection. An organism with a highly deleterious mutation dies and does not breed so provided that the rate of such mutations is not too high it does not imperil the survival of the species.
quote:
So the precise form of the fitness function is critical to the experiment’s success. It is essential that the algorithm can detect very small incremental improvements in fitness and then maintain them for the subsequent generations. Is natural selection in biology that fussy?
Any change in an organism that gives it even a slight advantage over the other members of its species will tend, on average, to improve its chances of reproducing. As it gradually spreads by inheritance through the population then that advantage will be present with more members and thus even more likely to increase. So, yes, it can be fussy.
quote:
So again, considering the numerical strings as an animal analogue, all the initial “creatures” were unable to demonstrate any fitness at all. Nevertheless they all survived to “breed”. In the cyberspace of a numerical experiment nobody dies young and all can get to breed.
But this makes the mistake of measuring fitness in terms of one criterion only. An animal species that has no lens in its eye may be entirely ‘unfit’ in terms of having a lens, but that does not mean that it is not well adapted to its environment and ‘fit’ in many respects. It would not die just because it lacked a lens.
quote:
The algorithm did not allow for the propagation of deleterious mutations or subsequent genetic drift.
Deleterious mutations are generally kept at low levels in populations by the fact that they hinder reproductive success and so don’t get replicated so often. Genetic drift that disturbs the function of beneficial mutations also gets weeded out or kept at low levels by selection. Drift which is neutral has no effect on the evolution of the trait in question.
quote:
I consider that this is an example of unconstrained evolution, starting with an completely unfit (i.e. dead or unsuccessfully breeding) life-form and passing through many unfit stages (but still breeding), until the numbers enter a region where the fitness function starts giving desirable answers. Judging by the following comments in the paper, the authors agree that this was an unconstrained evolutionary process:
See comments on ‘fitness’ above as to why this is not relevant.
Glenn
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
Neil,
Sorry to double post but I just noticed that In your earlier post you said that “This complete lack of fitness lasted for many generations, but definite progress on fitness was reported after 650 generations.” This would be an important point, but I can't find anywhere in the paper that says there was no fitness at all in these earlier generations. Have I missed something?
Glenn
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Originally posted by Glenn Oldham:
Neil,
Sorry to double post but I just noticed that In your earlier post you said that “This complete lack of fitness lasted for many generations, but definite progress on fitness was reported after 650 generations.” This would be an important point, but I can't find anywhere in the paper that says there was no fitness at all in these earlier generations. Have I missed something?
Glenn
The relevant part of the paper is here. Unfortunately, the graphical results in the paper are mostly illegible, so it is necessary to rely on the textual description.
The fitness of the first trial generation (produced randomly) is reported as follows, with my emphasis:
quote:
The individual in the initial random population of 50 that happened to get the highest score produced a constant +5V output at all times, irrespective of the input. It received a fitness of slightly above zero just because of noise. Thus, there was no individual in the initial population that demonstrated any ability whatsoever to perform the task.
So all the initial life-forms were completely unfit, demonstrating none of the desired behaviour (i.e. +5 volts dc output at 1 kHz input; 0 volts output at 10 kHz input). Nevertheless, the algorithm kept doing its work. At 220 generations they report the first microscopic hints of fitness:
quote:
After 220 generations, the best circuit was basically copying the input to the output. However, on what would have been the high part of the square wave, a high frequency component was also present, visible as a blurred thickening of the line in the photograph. This high-frequency component exceeds the maximum rate at which the FPGA can make logic transitions, so the output makes small oscillations about a voltage slightly below the normal logic-high output voltage for the high part of the square wave. After another 100 generations, the behaviour was much the same, with the addition of occasional glitches to 0V when the output would otherwise have been high.
Note the tentativeness of the description here. I suspect that one of the reasons they had to define the fitness function very carefully is so that the algorithm could identify these first microscopic hints of fitness and then progressively refine them.
At generation 650 through to 1100 they report “definite” progress, as follows:
quote:
Once 650 generations had elapsed, definite progress had been made. For the 1kHz input, the output stayed high (with a small component of the input wave still present) only occasionally pulsing to a low voltage. For the 10kHz input, the input was still basically being copied to the output. By generation 1100, this behaviour had been refined, so that the output stayed almost perfectly at +5V only when the 1kHz input was present.
At generation 1400 they report futher progress, as follows:
quote:
By generation 1400, the neat behaviour for the 1kHz input had been abandoned, but now the output was mostly high for the 1kHz input, and mostly low for the 10kHz input...with very strange looking waveforms. This behaviour was then gradually improved.
At generation 2800 they report almost perfect behaviour, and at generation 3500, perfect behaviour.
For a quick summary of the fitness over all the generations, see figure 4 in this section of the paper. This shows the fitness graphically, both the average for the population, and also the maximum within the population. It also shows clearly the very unpromising start with zero fitness.
There was a marked change in fitness behaviour around generation 2660, when the average fitness leapt markedly from a steady value of around 0.3 to a value of 1.0. They discuss this phenomenon in some detail in another paper, but unfortunately I don’t have access to that.
Neil
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
quote:
Originally posted by Faithful Sheepdog:
quote:
Originally posted by Glenn Oldham:
Neil,
Sorry to double post but I just noticed that In your earlier post you said that “This complete lack of fitness lasted for many generations, but definite progress on fitness was reported after 650 generations.” This would be an important point, but I can't find anywhere in the paper that says there was no fitness at all in these earlier generations. Have I missed something?
Glenn
The relevant part of the paper is here. Unfortunately, the graphical results in the paper are mostly illegible, so it is necessary to rely on the textual description.
The fitness of the first trial generation (produced randomly) is reported as follows, with my emphasis:
quote:
The individual in the initial random population of 50 that happened to get the highest score produced a constant +5V output at all times, irrespective of the input. It received a fitness of slightly above zero just because of noise. Thus, there was no individual in the initial population that demonstrated any ability whatsoever to perform the task.
So all the initial life-forms were completely unfit, demonstrating none of the desired behaviour (i.e. +5 volts dc output at 1 kHz input; 0 volts output at 10 kHz input). Nevertheless, the algorithm kept doing its work. At 220 generations they report the first microscopic hints of fitness:
quote:
After 220 generations, the best circuit was basically copying the input to the output. However, on what would have been the high part of the square wave, a high frequency component was also present, visible as a blurred thickening of the line in the photograph. This high-frequency component exceeds the maximum rate at which the FPGA can make logic transitions, so the output makes small oscillations about a voltage slightly below the normal logic-high output voltage for the high part of the square wave. After another 100 generations, the behaviour was much the same, with the addition of occasional glitches to 0V when the output would otherwise have been high.
Note the tentativeness of the description here. I suspect that one of the reasons they had to define the fitness function very carefully is so that the algorithm could identify these first microscopic hints of fitness and then progressively refine them.
At generation 650 through to 1100 they report “definite” progress, as follows:
quote:
Once 650 generations had elapsed, definite progress had been made. For the 1kHz input, the output stayed high (with a small component of the input wave still present) only occasionally pulsing to a low voltage. For the 10kHz input, the input was still basically being copied to the output. By generation 1100, this behaviour had been refined, so that the output stayed almost perfectly at +5V only when the 1kHz input was present.
At generation 1400 they report futher progress, as follows:
quote:
By generation 1400, the neat behaviour for the 1kHz input had been abandoned, but now the output was mostly high for the 1kHz input, and mostly low for the 10kHz input...with very strange looking waveforms. This behaviour was then gradually improved.
At generation 2800 they report almost perfect behaviour, and at generation 3500, perfect behaviour.
For a quick summary of the fitness over all the generations, see figure 4 in this section of the paper. This shows the fitness graphically, both the average for the population, and also the maximum within the population. It also shows clearly the very unpromising start with zero fitness.
There was a marked change in fitness behaviour around generation 2660, when the average fitness leapt markedly from a steady value of around 0.3 to a value of 1.0. They discuss this phenomenon in some detail in another paper, but unfortunately I don’t have access to that.
Neil
Neil, the figure four on the page you refer to is frustratingly unclear. Is that line from generation 0 to 220 above the zero fitness line or not - it is hard to tell. Presumably if there was no fitness at all in that time then they would have had to explain how they selected the individuals for breeding. They don't, so I imagine there was some but it is less clear than it might be.
I am not entirely sure what relevance your point has to the question of whether this invalidates the experiment's analgousness to evolution by natural selection. As I said earlier "An animal species that has no lens in its eye may be entirely ‘unfit’ in terms of having a lens, but that does not mean that it is not well adapted to its environment and ‘fit’ in many respects. It would not die just because it lacked a lens." The fact that an initial population entirely lacks a trait does not render the individuals completely unfit nor the evolution of that trait impossible. So I am not clear what your point is with reference to the early lack of fitness. Perhaps you have something else in mind.
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Glenn Oldham said:
Neil, the figure four on the page you refer to is frustratingly unclear. Is that line from generation 0 to 220 above the zero fitness line or not - it is hard to tell. Presumably if there was no fitness at all in that time then they would have had to explain how they selected the individuals for breeding. They don't, so I imagine there was some but it is less clear than it might be.
The figures given in the paper are indeed frustratingly unclear, so I relied on the text. Note that fitness is defined numerically by equation (1) given here. The analogue integrator (shown in figure 2) acts as some kind of smoothing and averaging circuit, making more numerical sense of the spikes and quirks in the output.
Despite the effectively zero values of fitness of the first generation, they were able to select the “fittest” one simply on the basis of electronic noise in the signal (either more or less, I’m not sure). Numerical “breeding” continued until the reported fitness was not simply a quirk of electronic noise.
They were able to differentiate marginally between individuals with either zero or extremely low fitness through the skilful design of the fitness function. Reading between the lines of the paper, I suspect that considerable effort was expended in order to determine a suitable fitness function that could differentiate with such low fitness values around, as the following indicates (with my emphasis):
quote:
This fitness function demands the maximising of the difference between the average output voltage when a 1kHz input is present and the average output voltage when the 10kHz input is present. The calibration constants k1 and k2 were empirically determined, such that circuits simply connecting their output directly to the input would receive zero fitness. Otherwise, with k1 = k2 = 1, small frequency-sensitive effects in the integration of the square-waves were found to make these useless circuits an inescapable local optimum.
It is important that the evaluation method -- here embodied in the analogue integrator and the fitness function Eqn. 1 -- facilitates an evolutionary pathway of very small incremental improvements. Earlier experiments, where the evaluation method only paid attention to whether the output voltage was above or below the logic threshold, met with failure. It should be recognised that to evolve non-trivial behaviours, the development of an appropriate evaluation technique can also be a non-trivial task.
So, just to emphasise the point, the fitness function was “intelligently designed” to do its job in the algorithm.
quote:
Glenn Oldham said:
I am not entirely sure what relevance your point has to the question of whether this invalidates the experiment's analogousness to evolution by natural selection. As I said earlier "An animal species that has no lens in its eye may be entirely ‘unfit’ in terms of having a lens, but that does not mean that it is not well adapted to its environment and ‘fit’ in many respects. It would not die just because it lacked a lens." The fact that an initial population entirely lacks a trait does not render the individuals completely unfit nor the evolution of that trait impossible. So I am not clear what your point is with reference to the early lack of fitness. Perhaps you have something else in mind.
One of the assumptions in this experiment is that even the tiniest amount of fitness does convey some advantage, and so is chosen accordingly. In other words, fitness is not a binary on/off type condition, but can be graduated infinitely. Without this assumption it is clear to me, based on the initial lack of fitness, that the experiment would not have delivered any usable results.
The experiment thus has built into it from the start the Darwinian assumption of gradualism, with improvement coming through tiny incremental steps, and a fitness function that can discern this. This assumption has delivered results in electronics, but only because the experiment was skilfully designed to do just that.
The key question, as always with numerical models, is how close the numerical processes mirror the physical reality. Can advantageous biological traits be so infinitely graduated? And can the biological fitness function (i.e. natural selection) be so finely tuned? To my mind these are both very questionable assertions.
Your comment about eyes reminds me of Richard Dawkins’ (in)famous comment that “5% of an eye is better than no eye”. This comment presumes that we have developed some functionality in the nascent eye physiology, i.e. that 5% of a nascent eye delivers an improved function in comparison to no eye.
This is only true if we presume a linear type of relationship. However, biology certainly doesn’t always work that way. Non-linear relationships and binary on/off states are common. This brings us full circle to irreducible complexity, and the biological trait that only works at all when it is fully complete. A string of numbers is not nearly so constrained.
Neil
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
Rex Monday
None but a blockhead
# 2569
|
Posted
quote:
Originally posted by Faithful Sheepdog:
<much rehashing expunged>
The experiment thus has built into it from the start the Darwinian assumption of gradualism, with improvement coming through tiny incremental steps, and a fitness function that can discern this. This assumption has delivered results in electronics, but only because the experiment was skilfully designed to do just that.
The key question, as always with numerical models, is how close the numerical processes mirror the physical reality. Can advantageous biological traits be so infinitely graduated? And can the biological fitness function (i.e. natural selection) be so finely tuned? To my mind these are both very questionable assertions.
They're only questionable if you use the words to mean something other than their common meaning.
The role of fitness in evolutionary biological thinking is as a measure of whether you have more offspring or not. A very minor increase in fitness will thus *by definition* result in more offspring. It can be a very small increase in offspring - maybe only one or two extra per thousand births - but that's fine. In a large population over long time periods, that sort of increment is very important. Tiny changes come to light through statistics all the time: the smaller the change, the more data you need but there's no cut-off point where a change is too small to be registered under any circumstances.
It is not thus possible by definition to have an increase in fitness too small to have any effect. Not 'highly questionable':impossible.
quote:
Your comment about eyes reminds me of Richard Dawkins’ (in)famous comment that “5% of an eye is better than no eye”. This comment presumes that we have developed some functionality in the nascent eye physiology, i.e. that 5% of a nascent eye delivers an improved function in comparison to no eye.
This is only true if we presume a linear type of relationship.
Let's turn that around. In other words, 5% of an eye (let's say a light-sensitive spot) will *not* deliver an improved function in comparison to no eye if there is no 'linear type of relationship'?
I fail to understand what you're saying here. To my mind, an animal that can sense the environment around it more accurately than one that can't, will find it advantageous. I cannot imagine any circumstances where this will not be.
Perhaps you could explain under what circumstances an animal with a light sensitive spot is not at such an advantage, and how that illuminates (excuse me) your 'non-linear relationship' (between what and what?). Or perhaps you could explain what I'm misunderstanding?
quote:
However, biology certainly doesn’t always work that way. Non-linear relationships and binary on/off states are common. This brings us full circle to irreducible complexity, and the biological trait that only works at all when it is fully complete. A string of numbers is not nearly so constrained.
Neil
But - as has been stated so often - there are perfectly valid ways through standard evolutionary biology to get to a biological system that is complex and only works once complete. That's one of the great failures of ID - it hasn't identified any system where preliminary systems and transistions cannot be proposed.
One of the great favourites of the ID apologists, the bacterial flagellum, has been the recipient of much lively debate among biologists, and there are many approaches that may explain its evolution in strictly standard terms. As that paper merrily says, not all of these proposals can be right. It is quite possible that they're all wrong. But there is no point in the flagellum design that is logically unbridgeable by standard evolutionary theory: ID says that there is, but has failed to identify it.
So, the problem you propose that the model may not address doesn't exist, and the lack of correspondence between the model and nature doesn't exist either. (there are plenty of other lacks: it is, like Camelot, only a model.)
R
--------------------
I am largely against organised religion, which is why I am so fond of the C of E.
Posts: 514 | From: Gin Lane | Registered: Mar 2002
| IP: Logged
|
|
Alan Cresswell
Mad Scientist 先生
# 31
|
Posted
quote:
Originally posted by Rex Monday:
I fail to understand what you're saying here. To my mind, an animal that can sense the environment around it more accurately than one that can't, will find it advantageous. I cannot imagine any circumstances where this will not be.
Well, the circumstance where that particular sense is unable to function anyway would be an obvious example. So, in an environment with no light source the value of light detection systems will be very limited (ie: only of use if coupled with some means of light generation), and assuming there is a cost to eyes in terms of the need to produce certain proteins etc then there will be an advantage to having no eyes - which is what we see. Anywhere where the cost of an "improvement" outweighs the advantage gained will behave similarly - so humans don't have the ability to see like hawks as we don't need to be able to do so not hunting small animals from a distance.
Posts: 32413 | From: East Kilbride (Scotland) or 福島 | Registered: May 2001
| IP: Logged
|
|
Justinian
Shipmate
# 5357
|
Posted
Neil: In this case, the system was set up intelligently. This does not mean that it produced results that would have been expected if designed from the top down- it is a case that shows that an evolutionary algorithm can come up with unexpected results. It is also an irreducibly complex object made up of not particularly complex parts (thus blowing that argument out of the water as well).
That experiment had to be designed intelligently because the system had no way of moving on to the next step- this had to be supplied by experimenters. In an atheistic worldview, there is no experimenter therefore the successful objects (at any level) are either formed easily or last a long time and are almost immutable. There are two obvious ways of doing this: either being formed very easily (so there will always be a lot of you around even if you break apart) or by being incredibly stable so that even if there isn't much of you, there will always be some of you. (Most of chemistry is based on these two principles- although they are called bond energy (stability) and entropy (how likely something is to form)). There is also a third and much subtler solution- self replication, and it is this that is responsible for living things.
The fitness criterion changing over time is nothing new- whether through natural disaster, a changing environment or just gradual improvement (look at most athletic records), what is considered fit does change over time.
--------------------
My real name consists of just four letters, but in billions of combinations.
Eudaimonaic Laughter - my blog.
Posts: 3926 | From: The Sea Coast of Bohemia | Registered: Dec 2003
| IP: Logged
|
|
Rex Monday
None but a blockhead
# 2569
|
Posted
quote:
Originally posted by Alan Cresswell:
quote:
Originally posted by Rex Monday:
I fail to understand what you're saying here. To my mind, an animal that can sense the environment around it more accurately than one that can't, will find it advantageous. I cannot imagine any circumstances where this will not be.
Well, the circumstance where that particular sense is unable to function anyway would be an obvious example. So, in an environment with no light source the value of light detection systems will be very limited (ie: only of use if coupled with some means of light generation), and assuming there is a cost to eyes in terms of the need to produce certain proteins etc then there will be an advantage to having no eyes - which is what we see. Anywhere where the cost of an "improvement" outweighs the advantage gained will behave similarly - so humans don't have the ability to see like hawks as we don't need to be able to do so not hunting small animals from a distance.
But in the circumstances where the new sense couldn't function - like the light sensor in a cave - then the organism wouldn't be able to sense its environment any more accurately!
In such cases, and as to why we can't see small voles from two hundred feet, then surely the useless (and costly) modification just won't be selected for - so it won't survive in the population anyway.
I was rather assuming that we were talking about features that had evolved, rather than transient neutral or deleterious mutations.
R
--------------------
I am largely against organised religion, which is why I am so fond of the C of E.
Posts: 514 | From: Gin Lane | Registered: Mar 2002
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
The previous responses to Neil's last post cover much of what I was going to say, but I would just like to highlight one rhetorical misrepresentation of the experiment (my emphasis):
quote:
Originally posted by Faithful Sheepdog:
One of the assumptions in this experiment is that even the tiniest amount of fitness does convey some advantage, and so is chosen accordingly. In other words, fitness is not a binary on/off type condition, but can be graduated infinitely . ...
...
Can advantageous biological traits be so infinitely graduated? And can the biological fitness function (i.e. natural selection) be so finely tuned? To my mind these are both very questionable assertions.
The use of the word 'infinitely' misrepresents the experiment. The experiment did not rely on the absurd notion of an infinitely small improvement in fitness and neither does evolutionary theory.
In any case Neil, nothing you have said shows that this experiment is not a good parallel to cases of natural selection operating on a trait that variation can improve in small steps.
The experiment is, therefore, evidence towards the validity of the concept of natural selection and is, emphatically, strong evidence against the Intelligent Design proponent William Dembski's idea that selection cannot create what he calls complex specified information. (If he still holds that view nowadays.)
In any case I thought you accepted micro-evolution, Neil, so I am a little puzzled as to your critique of the experiment. Or do you accept micro-evolution but believe that it does not happen by natural selection?
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Rex Monday said:
Let's turn that around. In other words, 5% of an eye (let's say a light-sensitive spot) will *not* deliver an improved function in comparison to no eye if there is no 'linear type of relationship'?
I fail to understand what you're saying here. To my mind, an animal that can sense the environment around it more accurately than one that can't, will find it advantageous. I cannot imagine any circumstances where this will not be.
Perhaps you could explain under what circumstances an animal with a light sensitive spot is not at such an advantage, and how that illuminates (excuse me) your 'non-linear relationship' (between what and what?). Or perhaps you could explain what I'm misunderstanding?
My comment on “linearity” drew on the fact that humans and chimpanzees share 98% of the same DNA, yet we are dramatically different creatures. To me that indicated a very non-linear relationship – a 2% change has produced amongst other things the human capacity for intellectual and artistic achievements – a massive difference. On this basis it is by no means obvious to me that biology is linear.
Dawkins’ rhetoric about the obvious superiority of 5% of an eye was his attempt to demonstrate the essential Darwinian component of gradualism, and presumed that the function delivered by a nascent evolving eye could be measured in simple percentage terms. His argument assumes that it makes biological (and indeed philosophical) sense to talk about 5% of an eye.
By this I think he meant 5% of the functionality of an eye, but this requires much further definition from a scientific viewpoint. Is that a physiological 5%, or an optometrist’s 5%, or a biochemical 5%, or a brain processing 5%, or something else altogether? 5% of what? Vision is a very complex phenomenon indeed, and to treat it in the simplistic linear fashion that Dawkins does skates over some enormous problems.
With regard to your comment on the “advantage” of eyes, it’s clear that many organisms have no eyes or vision of any kind, as Alan says. I think there are some deep water fish which have eyes but have lost (or never had) all vision. It’s clear that functioning eyes are not always necessary for successful life.
quote:
Rex Monday said:
One of the great favourites of the ID apologists, the bacterial flagellum, has been the recipient of much lively debate among biologists, and there are many approaches that may explain its evolution in strictly standard terms. As that paper merrily says, not all of these proposals can be right. It is quite possible that they're all wrong. But there is no point in the flagellum design that is logically unbridgeable by standard evolutionary theory: ID says that there is, but has failed to identify it.
So, the problem you propose that the model may not address doesn't exist, and the lack of correspondence between the model and nature doesn't exist either. (there are plenty of other lacks: it is, like Camelot, only a model.)
Actually, in the last paragraph I would reverse the order of the auxiliary verbs, and firm up the language. The experimental model definitely does not address the problem of irreducible complexity, and in my engineering judgement such a problem does indeed exist.
The configuration of the chip was governed by 1800 bits held in the computer. According to the authors of the paper, no possible numerical configuration was physically damaging or lethally destructive to the chip. It therefore follows that there is always a safe sequential route from any one configuration to any other configuration changing only one bit at a time.
Consequently this experiment cannot possibly test the notion of irreducible complexity, since the way the experiment was set up bypassed any such possibility. For a design problem in electronics that was a reasonable experimental approach, but it simply begs the question in biology.
I appreciated the link to the Matzke paper, although it is very technical. I have enough maths and physics to comment on electronics with some confidence, but I am less confident in biochemistry, so I will leave it to others to critique this paper. Has it been published in a refereed journal yet, or did it go straight to the web? In any case I think it is fair to say that if Behe had not written his book on irreducible complexity, this paper would not have been written.
quote:
Justinian said:
Neil: In this case, the system was set up intelligently. This does not mean that it produced results that would have been expected if designed from the top down- it is a case that shows that an evolutionary algorithm can come up with unexpected results. It is also an irreducibly complex object made up of not particularly complex parts (thus blowing that argument out of the water as well).
The experiment certainly surprised people, since many predicted that a suitable chip configuration for the given function would never be found. The technicalities of microchips are not my forte, but I noticed that the experiment was quite deliberately set up to use logic gates in a very unorthodox and highly experimental manner. This was equivalent (in terms of the laws of physics) to Rolf Harris making music with a violin bow on a large saw.
On irreducible complexity, see my comment above to Rex Monday, and earlier comments on the graduated nature of fitness function. Note that what was evolved was a numerical configuration for a particular part of a chip, and not any kind of object. The evolved configuration was definitely not irreducibly complex, since we can track it and demonstrate some numerical fitness (as defined in the experiment) at every stage of the way.
That we can do it in this case for an electronic circuit does not mean that we can do the same for every biological system.
quote:
Glenn Oldham said:
The use of the word 'infinitely' misrepresents the experiment. The experiment did not rely on the absurd notion of an infinitely small improvement in fitness and neither does evolutionary theory.
In any case Neil, nothing you have said shows that this experiment is not a good parallel to cases of natural selection operating on a trait that variation can improve in small steps.
I used the word “infinitely” in parallel with “tiniest amount of fitness” and “finely tuned”. I don’t think this is a misrepresentation at all. The numerical fitness function used in the experiment allowed the algorithm to decide numerically to the arithmetical precision limit within the computer itself. In the first generation it was only more or less noise (and some fine tuning of the fitness function) which made any difference to the fitness.
Here is how they “bred” the next generation, with my emphasis:
quote:
The population of size 50 was initialised by generating fifty random strings of 1800 bits each. After evaluation of each individual on the real FPGA, the next generation was formed by first copying over the single fittest individual unchanged (elitism); the remaining 49 members were derived from parents chosen through linear rank-based selection, in which the fittest individual of the current generation had an expectation of twice as many offspring as the median-ranked individual. The probability of single-point crossover was 0.7, and the per-bit mutation probability was set such that the expected number of mutations per genotype was 2.7. This mutation rate was arrived at in accordance with the Species Adaptation Genetic Algorithm (SAGA) theory of Harvey [4], along with a little experimentation.
So the numerical fitness of each “creature” was clearly established at all times. Fitness was a numerical function governed by equation 1 in the paper, which is a mathematically continuous function related to the analogue integrator output. Subject to the arithmetical precision limits of the computer, I consider that an “infinite graduation” of fitness is a fair comment on how the experiment was set up.
As Rex Monday noted above, the experiment could be criticised from a biological perspective in other areas. Note the constant elitism associated with the fittest individual, which is protected from breeding and simply copied unchanged to the next generation. This technique gradually and deliberately ratchets up the fitness of the whole population. This is perfectly reasonable in a numerical search algorithm, but is a teleology forbidden to biological Darwinism.
Note also the absence of highly deleterious mutations. I suspect that a mutation rate of 2.7 per genotype would cause catastophic effects in a real population of 50 animals. Such a population is admittedly low, but not unrealistic for some rare birds of prey in Scotland, for example.
quote:
Glenn Oldham said:
The experiment is, therefore, evidence towards the validity of the concept of natural selection and is, emphatically, strong evidence against the Intelligent Design proponent William Dembski's idea that selection cannot create what he calls complex specified information. (If he still holds that view nowadays.)
In any case I thought you accepted micro-evolution, Neil, so I am a little puzzled as to your critique of the experiment. Or do you accept micro-evolution but believe that it does not happen by natural selection?
Dembski’s ideas on “complex specified information” (CSI) have been developed mathematically with some rigour, but of course he is not without his critics. However, I believe that he still holds them and defends them as necessary.
Your comment on natural selection creating CSI can only be sustained if you first demonstrate that the algorithm is a fair representation of biological natural selection as understood in Darwinism. In my opinion this particular experiment has not achieved that, nor did it set out to do so.
I have no problem accepting microevolution by natural selection. This mechanism works on latent and unexpressed genes (such as moth colouration) that already exist. To my mind this is demonstrable science, and accordingly I have no argument with it.
However, to create the new biological structures required by an evolutionary paradigm requires a mechanism that can create new and coordinated genetic information – and on a continuously grand scale. Artificial selection (i.e. animal and plant breeding) has shown a clear species boundary, and the few speciation events observed scientifically seem frankly trivial in comparison to the claims being made for them.
So, if artificial (i.e. intelligent) selection cannot produce any macroevolution, then I see even less evidence that an undirected natural selection can do so. I consider natural selection a demonstrably failed scientific hypothesis for that kind of creative task.
I do not rule out all macroevolution a priori, and I am intrigued by the non-Darwinian evolutionary models, which see macroevolution as a front-loaded process akin to the growth of an embryo. However, along with IDists, I do consider that without an explicit teleology the scientific research into biological origins is hopelessly handicapped.
Neil
[minor typo]
[ 15. July 2004, 17:05: Message edited by: Faithful Sheepdog ]
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
ken
Ship's Roundhead
# 2460
|
Posted
quote:
Originally posted by Faithful Sheepdog:
Artificial selection (i.e. animal and plant breeding) has shown a clear species boundary,
Quite the opposite in that lots of species hybridise in cultivation that don't in nature.
quote:
and the few speciation events observed scientifically seem frankly trivial in comparison to the claims being made for them.
Not at all. How many natural speciation events do you think we should have observed?
quote:
So, if artificial (i.e. intelligent) selection cannot produce any macroevolution
who says it can't?
--------------------
Ken
L’amor che move il sole e l’altre stelle.
Posts: 39579 | From: London | Registered: Mar 2002
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
![[brick wall]](graemlins/brick_wall.gif)
Glenn
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Alan Cresswell
Mad Scientist 先生
# 31
|
Posted
quote:
Originally posted by Glenn Oldham:
Glenn
Yep, it's a Dead Horse alright.
--------------------
Don't cling to a mistake just because you spent a lot of time making it.
Posts: 32413 | From: East Kilbride (Scotland) or 福島 | Registered: May 2001
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Originally posted by Glenn Oldham:
Glenn
Glenn, I recommend a good stiff whisky. Since I'm currently off booze completely due to ME/CFS, I'm going to have a serious hit on the mineral water. Maybe I'll be thinking more clearly after that.
In the meantime here is a complete diversion from your scheduled entertainment: the Panspermia Boys. Forget both Darwinism and ID, life was seeded from outer space. ![[Big Grin]](biggrin.gif)
Neil
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
Justinian
Shipmate
# 5357
|
Posted
quote:
Originally posted by Faithful Sheepdog:
My comment on “linearity” drew on the fact that humans and chimpanzees share 98% of the same DNA, yet we are dramatically different creatures. To me that indicated a very non-linear relationship – a 2% change has produced amongst other things the human capacity for intellectual and artistic achievements – a massive difference. On this basis it is by no means obvious to me that biology is linear.
Yet we are very simmilar to chimps in many ways. Of course things aren't always linear- picture the difference between water at 99C and at 101C. This doesn't mean that water is more or less linear when it doesn't cross these threshholds.
quote:
His argument assumes that it makes biological (and indeed philosophical) sense to talk about 5% of an eye.
It doesn't. On the other hand, an eye with minimal functionality does- and the 5% figure was shorthand for a weakly functional eye. It was a simplification for whatever argument he was making. By their nature, simplifications leave out useful data, but that doesn't make them pointless.
quote:
With regard to your comment on the “advantage” of eyes, it’s clear that many organisms have no eyes or vision of any kind, as Alan says. I think there are some deep water fish which have eyes but have lost (or never had) all vision. It’s clear that functioning eyes are not always necessary for successful life.
Who ever said they were? I'll tell you if I see a blade of grass with eyes, and that's certainly successful life. On the other hand, there are situations in which eyes are a huge advantage. Different adaptations are useful in different circumstances.
quote:
The experimental model definitely does not address the problem of irreducible complexity, and in my engineering judgement such a problem does indeed exist.
Yet past a certain point, the circuit is irreducably complex- it stops working if you try removing anything.
A human being doesn't work if you remove the brain. This doesn't mean his entire body falls apart.
Possibly it might help if your "judgement as an engineer" was backed up with what you understand by irreducible complexity.
quote:
The configuration of the chip was governed by 1800 bits held in the computer. According to the authors of the paper, no possible numerical configuration was physically damaging or lethally destructive to the chip. It therefore follows that there is always a safe sequential route from any one configuration to any other configuration changing only one bit at a time.
Except that most of them do not work for the intended purpose.
quote:
The experiment certainly surprised people, since many predicted that a suitable chip configuration for the given function would never be found. The technicalities of microchips are not my forte, but I noticed that the experiment was quite deliberately set up to use logic gates in a very unorthodox and highly experimental manner. This was equivalent (in terms of the laws of physics) to Rolf Harris making music with a violin bow on a large saw.
You haven't taken the analogy far enough. It's the equivalent of trying to make music with a violin bow on a large saw, and getting "Air On A G String" sounding as if it was played by a master with a stradivarius.
quote:
The evolved configuration was definitely not irreducibly complex, since we can track it and demonstrate some numerical fitness (as defined in the experiment) at every stage of the way.
What then do you mean by "irreducibly complex"? We can not remove certain cells, we can only backtrack the entire object into something no less complex- it contains the same number of cells, just in a different arrangement.
quote:
Note also the absence of highly deleterious mutations.
Wrong- they were just weeded or backtracked.
quote:
I suspect that a mutation rate of 2.7 per genotype would cause catastophic effects in a real population of 50 animals. Such a population is admittedly low, but not unrealistic for some rare birds of prey in Scotland, for example.
Do you have an actual point here? The mutation rate is much lower in the wild- but the wild isn't controlled conditions. It just shows some behaviour claimed by many influential creationists to be impossible (and I'm sure you'll agree that an argument that something can't be done is trumped by proof it has been).
If you expect a single experiment in a different field done for a different reason to have every property of the world at large, you are either stupid or willfuly blind.
--------------------
My real name consists of just four letters, but in billions of combinations.
Eudaimonaic Laughter - my blog.
Posts: 3926 | From: The Sea Coast of Bohemia | Registered: Dec 2003
| IP: Logged
|
|
Justinian
Shipmate
# 5357
|
Posted
quote:
Originally posted by Faithful Sheepdog:
quote:
Originally posted by Glenn Oldham:
Glenn
Glenn, I recommend a good stiff whisky. Since I'm currently off booze completely due to ME/CFS, I'm going to have a serious hit on the mineral water. Maybe I'll be thinking more clearly after that.
In the meantime here is a complete diversion from your scheduled entertainment: the Panspermia Boys. Forget both Darwinism and ID, life was seeded from outer space.
Neil
That only defers the problem. How did life start in outer space?
--------------------
My real name consists of just four letters, but in billions of combinations.
Eudaimonaic Laughter - my blog.
Posts: 3926 | From: The Sea Coast of Bohemia | Registered: Dec 2003
| IP: Logged
|
|
Faithful Sheepdog
Shipmate
# 2305
|
Posted
quote:
Originally posted by Justinian:
If you expect a single experiment in a different field done for a different reason to have every property of the world at large, you are either stupid or willfuly blind.
Justinian, I am not the one proposing that the success of genetic algorithms represents clear evidence for the correctness of biological Darwinism. This line of argument has come from you and others.
I am quite happy for you to disagree with my views, but I am not prepared to tolerate ad hominem abuse of this kind. I suggest you withdraw your remark and apologise for it.
Neil
--------------------
"Random mutation/natural selection works great in folks’ imaginations, but it’s a bust in the real world." ~ Michael J. Behe
Posts: 1097 | From: Scotland | Registered: Feb 2002
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
quote:
Originally posted by Justinian regarding the link to the Panspermia hypothesis:
That only defers the problem. How did life start in outer space?
Here is that sites 'answer' to your question, (but I wouldn't hold your breath!)What difference does it make?
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Justinian
Shipmate
# 5357
|
Posted
quote:
Originally posted by Faithful Sheepdog:
quote:
Originally posted by Justinian:
If you expect a single experiment in a different field done for a different reason to have every property of the world at large, you are either stupid or willfuly blind.
Justinian, I am not the one proposing that the success of genetic algorithms represents clear evidence for the correctness of biological Darwinism. This line of argument has come from you and others.
I am quite happy for you to disagree with my views, but I am not prepared to tolerate ad hominem abuse of this kind. I suggest you withdraw your remark and apologise for it.
Neil
Neil, I am not claiming that they present clear evidence that evolution has happened - that is simply your misconception. What I am claiming is that they present proof that some of your cherished arguments against evolution are flat out wrong - things that it is claimed are impossible could both happen have been shown to do so. To claim that such models show that biological darwinism clearly happened (rather than simply that many of your objections are false) is your misinterpretation, and although to accidently construct such a strawman is a stupid mistake, I am prepared to apologise for extrapolating the fact that you have both started and maintained such a stupid strawman on this thread to the conclusion that you are stupid. If this is the case, then we all do stupid things and calling you either stupid or wilfuly blind for doing something stupid is one of my stupid mistakes.
Amongst your responses to the experiment, you have come up with a number of objections that the experiment doesn't show certain things shown in the real world. No experiment we could do ever could have the correct initial parameters for the real world, and hence there will always be something missing. This is true for all scientific experiments in any field - and here you bring in such arguments as the mutation rate being much faster than in a real world biological system. Tell me, as an engineer do you ever use models that only simulate a part of the design? Do you ever use models that are not built at 100% scale and in the real world?
--------------------
My real name consists of just four letters, but in billions of combinations.
Eudaimonaic Laughter - my blog.
Posts: 3926 | From: The Sea Coast of Bohemia | Registered: Dec 2003
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
quote:
Originally posted by Faithful Sheepdog:
quote:
Glenn Oldham said:
The experiment is, therefore, evidence towards the validity of the concept of natural selection and is, emphatically, strong evidence against the Intelligent Design proponent William Dembski's idea that selection cannot create what he calls complex specified information. (If he still holds that view nowadays.)
In any case I thought you accepted micro-evolution, Neil, so I am a little puzzled as to your critique of the experiment. Or do you accept micro-evolution but believe that it does not happen by natural selection?
Dembski’s ideas on “complex specified information” (CSI) have been developed mathematically with some rigour, but of course he is not without his critics. However, I believe that he still holds them and defends them as necessary.
Your comment on natural selection creating CSI can only be sustained if you first demonstrate that the algorithm is a fair representation of biological natural selection as understood in Darwinism. In my opinion this particular experiment has not achieved that, nor did it set out to do so.
But Dembski declares selective algorithims on computers to be as incapable as natural selection in creating 'complex specified information'. He draws no distinction between them but damns all trial and error processes as mere combinations of chance and necessity and therefore incapable of generating 'complex specified information'. The experiment we have been looking at refutes that assertion and thus makes his claims against evolution highly doubtful at best.
For example in his article Intelligent Design as a Theory of Information he says: quote:
"If chance and necessity left to themselves cannot generate CSI, is it possible that chance and necessity working together might generate CSI? The answer is No. Whenever chance and necessity work together, the respective contributions of chance and necessity can be arranged sequentially. But by arranging the respective contributions of chance and necessity sequentially, it becomes clear that at no point in the sequence is CSI generated. Consider the case of trial-and-error (trial corresponds to necessity and error to chance). Once considered a crude method of problem solving, trial-and-error has so risen in the estimation of scientists that it is now regarded as the ultimate source of wisdom and creativity in nature. The probabilistic algorithms of computer science (e.g., genetic algorithms-see Forrest, 1993) all depend on trial-and-error. So too, the Darwinian mechanism of mutation and natural selection is a trial-and-error combination in which mutation supplies the error and selection the trial. An error is committed after which a trial is made. But at no point is CSI generated." [my emphasis]
But in where then, in the experiment that we have been looking at, did the complex specified information that was produced come from? We wound up with a complex and functional arrangement. This is an example of complex specified information. Where did it come from? Answer that and you can see that Demski's arguments against natural selection are worthless.
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
Glenn Oldham
Shipmate
# 47
|
Posted
And another couple of things to add to the post of mine just before this one.
quote:
Originally posted by Faithful Sheepdog I am intrigued by the non-Darwinian evolutionary models, which see macroevolution as a front-loaded process akin to the growth of an embryo.
If you know of any such 'front loaded models' I would be glad to hear of them. As far as I can see the ID movement is solely and anti-darwinian-evolution movement with no positive proposals for an alternative explanation of the phenomena.
quote:
Originally posted by Faithful Sheepdog However, along with IDists, I do consider that without an explicit teleology the scientific research into biological origins is hopelessly handicapped.
Biologists have an equivalent to the teleological perspective, it is called 'adaptation', i.e. the concept that structures and systems are produced by evolution that have functions that enable the organism in question to survive and reproduce. It is highly obscure to me how the ID community think that the design hypothesis gives them any advantage over non-ID biologists in the practice of research. There is a paper by Jonathan Wells (adobe acrobat format) called Using Intelligent Design Theory to Guide Scientific Research that looks as if it might answer my bafflement, but utterly fails to do so. He gives an example of how he used design theory to guide his research and this appears to boil down to "In the electron microscope, centrioles look like tiny turbines. Using TOPS [Theory of Organismal Problem-Solving]as my guide, I concluded that if centrioles look like turbines they might actually be turbines."
There is absolutely no resaon why a Darwinian might not think this thought! The darwinian supposes that the centriole has a function and is interested to know what that may be. Where does his non-belief in Intelligent Design prevent him from having a thought like "if centrioles look like turbines they might actually be turbines (albeit not made of sheet metal etc)"?
And Well's suggestion that "Centrosomes … have never been a favorite object of study within the framework of Darwinian theory, because even though they replicate every time a cell divides they contain no DNA (Marshall and Rosenbaum, 2000), and they have no evolutionary intermediates from which to reconstruct phylogenies (Fulton, 1971)." is simply silly. Physiologists and biochemists and cell biologists are interested in function which can be, and usually are, investigated without evolutionary questions in mind.
If you have a better idea as to why Intelligent Design is supposed to be so radically helpful in research please let me know.
Glenn
--------------------
This entire doctrine is worthless except as a subject of dispute. (G. C. Lichtenberg 1742-1799 Aphorism 60 in notebook J of The Waste Books)
Posts: 910 | From: London, England | Registered: May 2001
| IP: Logged
|
|
|
Printer-friendly view
