Intelligent Evolution?

[articulett]

Null response.

That was my thought. More off topic nothingness. Southwind made a valid point in his OP-- a few people thought that his analogy wouldn't work. They didn't realize that it doesn't work for them because of their own lack of understanding, but it actually does work for many. They never conveyed a cohesive difference between natural and "unnatural" selection-- why the evolution of a disease causing virus is so different than the evolution of a computer virus. I think the majority could follow southwind, you, Dawkins, Jones, etc. and in fact, the majority can and do-- I don't think anyone really can pin down what the dissenters are trying to say or if they are on the same page even. And I'm the arrogant one?

The only other people who seem to resort to this (identical) ad hom digression are theists whom I find arrogant because of their claims about faith giving them "knowingness" about what "god" wants.

I think the questions they avoid followed by the ad homs they use show that they are cluing in to how very poor their arguments are. (But I shan't expect an iota of humility from them on this thread.) (Nevertheless... my arrogance finds me tickled at the attack.)

 

[jimbob]

Paulhoff, in answer to your questions, I would contend no, but there have been loads of examples of (human-inducesd) intelligent design.

 

[jimbob]

So evolution can produce solutions that design can not, and there is no fundamental difference?

Well you tell me jimbob: is there a fundamental difference between an intelligence using an evolutionary algorithm and nature? Because if not you cannot complain about what the more abstract form - the computer simulation - can demonstrate about all forms. If there is a fundamental difference you are going to have to tell me what it is.

Because it seems to me that when what you're evolving is algorithms you're going to get algorithms about algorithms. When that happens you can see what appears to be design: and this is because the possibilities the algorithm can afford are far more constrained. My design algorithm will produce design idioms. Those idioms will dictate the sort of construction the design takes. We can recognise them and categorise these design idioms - like bridges being suspension, cantilever, arch and so forth.

The functions of these idioms are completely natural - which is why we see design idioms in nature too. If you don't understand what is going on above you're going to be liable to think these design idioms mean that some intelligence like yourself was responsible.

There is a fundmental difference in the processes of evolution and of engineering development.

Can evolutionary algorithms develop experiments that learn from their failures, or can evolutionary algorithms only build on relative success?

If there is a failure in a design, can evoluionary algorithms analyse what the fault is, and attempt to remedy just that fault?

When you get to Evolution as opposed to "evolutionary algorithms" there is no intelligence.

If you are using an evolutionary algorithm as a tool to design something, then you do not know what form the result will take, but you do know what it will do because you have defined what constitutes fitness. The algorithm is dumb, and that part off the process is, but the whole process of design is not as you decided to use the evolutionary algorithm, set it up, configured it and the specifications, and waited for a product that met your specifications to "evolve" with a lowercase "e". If you set the specifications then the process of design is intelligent.

This is entirely analogous to how ID is claimed to work instead of Evolution. "God directed evolution towards the creation of mankind".

The fundamental difference between use of an evolutionary algorithm and Evolution is based on imperfect self-replication; as otherwise there is an artificial selection criteria needed to select which iterations will act as the seeds for the next one.

There are not an infinite number of trials available.

You don't need them to see the sort of things you say cannot possibly occur. You simply see them in a less exaggerated manner.

Given the same flawed starting design, how many evolutionary trials would be needed to replicate say the XV design iterations of the Spitfire? I will allow any result where just the previous faults are addressed, and there are no other changes.

 

[articulett]

About as many in the nozzle example--and why care if there are more changes--isn't the goal greater efficiency or whatever--not keeping everything exactly identical except the perceived flaw. Boy do you go out of your way not to "get it".

You realize that there are programs that are "designed" with algorithms so they can learn from failures. Failures just don't continue on-- only the successes do. Do you just not understand the nozzle example? Do you not understand that this is an example from a foremost educator on the topic. When the experts on the topic who actually teach it to others tell you that the problem is you-- aren't you at least curious. Or do you have such faith in your viewpoint that no amount of evidence will change it? Your problem with the analogy is only similar to the ID proponents problem in your complete inability (seemingly purposeful) not to allow the analogy to make sense.

 

[cyborg]

Can evolutionary algorithms develop experiments that learn from their failures,

I don't know: have you learnt from your failures to understand yet?

Given the same flawed starting design, how many evolutionary trials would be needed to replicate say the XV design iterations of the Spitfire?

Between 1 and infinity.

 

[cyborg]

There is a fundmental difference in the processes of evolution and of engineering development.

There are fundamental differences between cats and dogs.

jimbob - STOP being dense. Please.

 

[articulett]

In conclusion--it appears that for the majority of people, Southwind's example can be illuminating and helpful in understanding how complexity comes about and what evolution is. For a small minority it is not helpful and though they cannot provide a better way of communicating the concept of how the environment selects information towards increasing complexity and efficiency over time-- how things can look designed, but not be designed from the top down at all. Complex systems like ant colonies occur all the time because of basic algorithms-- not some big plan.

Does the info. work? If so use it, keep it around for further refinement, add to it, tweak it and see if you get something better.

Same basic algorithm running Evolution by natural selection... and the evolution of technology.

And Mijo-- a carbon atom is the exact same thing whether in a person or in a sheet of steel. The atoms are interchangeable. At it's core we are talking about the same thing when we are talking about evolution when we talk about the various systems any given carbon molecule will be a part of through time.

 

[Paulhoff]

Paulhoff, in answer to your questions, I would contend no, but there have been loads of examples of (human-inducesd) intelligent design.

Please, who thinks this is about anything human. We are taking about Intelligent Evolution, intelligent design, a so-called god.

Paul

 

[ImaginalDisc]

In conclusion--it appears that for the majority of people, Southwind's example can be illuminating and helpful in understanding how complexity comes about and what evolution is. For a small minority it is not helpful and though they cannot provide a better way of communicating the concept of how the environment selects information towards increasing complexity and efficiency over time-- how things can look designed, but not be designed from the top down at all. Complex systems like ant colonies occur all the time because of basic algorithms-- not some big plan.

Unable to defend your position through reason, you resort to argumentum ad populum, and you don't even try to substantiate it?

Does the info. work? If so use it, keep it around for further refinement, add to it, tweak it and see if you get something better.


Same basic algorithm running Evolution by natural selection... and the evolution of technology.

An incorrect assessment of Evolution. Does the organism reproduce? If so, pass on all traits, with added random mutation. There is no parsing of the information to find the working bits. That's a feature of design, where intelligent actors are able separate the wheat from the chafe.

And Mijo-- a carbon atom is the exact same thing whether in a person or in a sheet of steel. The atoms are interchangeable. At it's core we are talking about the same thing when we are talking about evolution when we talk about the various systems any given carbon molecule will be a part of through time.

Those systems are fundamentally different in form and function, and they owe those differences to their different types of origin.

 

[cyborg]

Those systems are fundamentally different in form and function, and they owe those differences to their different types of origin.

Null statement.

 

[ImaginalDisc]

Null statement.

False. For reasons I have explained at length. For the benefit of others, I'll summarize two of the most salient points:

Evolution has no goal seeking ability, designers do. Evolved forms frequently reach dead-ends where life forms are unable to made transitions to "better" "designs" across "design space." Scare quotes are intentional, because it is important not to confuse the unordered, unguided, mindless process of Evolution with the intelligent guidance of design. Evolution is unable to implement improvements made in one form into another, designers can.

Example:Humans and guinea pigs have lost the ability to make their own vitamin C, which is a critical failure that often leads to disease and eventually, death. This major flaw is a consequence of (separate) mutations in the otherwise entirely functional enzymatic pathway which produced vitamin C in the ancestors of both species, and which functions quite well in other mammals.

If humans and guinea pigs had been designed, it would be easy enough to lift the functional genetic material from organisms that have them, and to add them to the human and guinea pig genomes. This conspicuous flaw is a testament to the inability of evolutionary processes to transmit information from one species to another.

Indeed, the inability to mingle genetic information is the key feature we use to distinguish species. Machines have no such species. Developments in the reduction drag coefficients made in airplanes were applied to automobiles, whereas developments in the reduction of drag coefficients in penguins were not applied to cormorants.

Not only does my statement have semantic content, it is empirically true.

 

[jimbob]

Cyborg and Articulett,

If evolution does not require imperfect self-replication, how are successful iterations propagated?

Is an external agency needed, together with arbitary (i.e. predefined) selection criteria?

 

[cyborg]

ID - if you won't understand anything I say you're going to continually be asserting X is not Y at me.

Yes ID. I know X is not Y. You are fighting an argument I am not making and will continue doing so until you STOP asserting X is not Y and start thinking about under what circumstances one would say 'X is Y' even though this is never, EVER, true if X and Y are physically separate objects.

If evolution does not require imperfect self-replication, how are successful iterations propagated?

Is an external agency needed, together with arbitary (i.e. predefined) selection criteria?

jimbob - I will ask this again: what part of the reproduction of the cell is 'self' reproduction?

If you break it down there is no 'self' being replicated - there are lots of different microbiological devices working in tandem to achieve that higher abstraction but there is no literal 'self' replication.

Once you are willing to accept this it seems quite a fallacy to insist that the mechanism of reproduction must be intimately bound to the mechanism of representation. Is the machinery that copied DNA external to the DNA? It sure doesn't seem to be on the inside!

So I yet again come back to the point that neither of you two seem even able to hear: you HAVE to be able to work at different levels of abstraction. At the cell level there is self-reproduction. At the intra-cell level there is not. What about that is not clear?

 

[ImaginalDisc]

ID - if you won't understand anything I say you're going to continually be asserting X is not Y at me.

Yes ID. I know X is not Y. You are fighting an argument I am not making and will continue doing so until you STOP asserting X is not Y and start thinking about under what circumstances one would say 'X is Y' even though this is never, EVER, true if X and Y are physically separate objects.

Then englighten me. On what basis are you making a comparison between Evolution and design?

On what "level of abstraction" do the results, which are shaped the processes that gave rise to them, similar?

 

[cyborg]

Then englighten me. On what basis are you making a comparison between Evolution and design?

On what "level of abstraction" do the results, which are shaped the processes that gave rise to them, similar?

I've been over that before and progressed nowhere. You simply continually assert that biology is physically different. You do not seem to understand that I am not asserting otherwise. Therefore you will simply bark back at me some aspect of biology an example of a 'cat' or a 'dog' or such and say, "this is NOT X!"

IF, however, you are WILLING to accept that an abstract model of evolution can be related to ALL evolutionary systems BY VIRTUE of its generality then we can progress somewhere. You should also note that SPECIFIC systems, i.e. where the general case is 'fixed', WILL EXHIBIT DIFFERENTIATING BEHAVIOURS. I AM NOT asserting that all consequences of the abstract reasoning about the universal set of evolutionary systems will be the same. I have asserted the opposite repeatedly.

Then you can go back and read my old posts with a different attitude.

 

[jimbob]

Can evolutionary algorithms develop experiments that learn from their failures

I don't know: have you learnt from your failures to understand yet?

Can you?

That is a pretty poor answer, will you actually try answering it?

Given the same flawed starting design, how many evolutionary trials would be needed to replicate say the XV design iterations of the Spitfire?

Between 1 and infinity.

The full quote was: "Given the same flawed starting design, how many evolutionary trials would be needed to replicate say the XV design iterations of the Spitfire? I will allow any result where just the previous faults are addressed, and there are no other changes."

Note that I did not say that evolutionary algorithms couldn't solve the problems, but that they would have left a different signature. Other (pseudo)random features would also have been altered, in a direct analogy to genetic drift.

Note also that I did not ask for the entire design to be replicated, just for the same problems to be solved without altering any unimportant feature.

The processes are fundamentally different.

There is an obvious similarity that both are iteritive.

But that is uninterestingly obvious.

 

[cyborg]

That is a pretty poor answer, will you actually try answering it?

I already did.

Are you an experiment that learns from its failures? Or perhaps you don't think you're experimental? Hmm... I think certain people might like that idea.

Note that I did not say that evolutionary algorithms couldn't solve the problems, but that they would have left a different signature.

Yes jimbob. This is what I have been saying. Jesus.

This is why the statement:

The processes are fundamentally different.

Is only an axiom of choice - it is NOT fundamental to the system because BOTH process can explain BOTH systems WITH THE RIGHT SIGNATURE.

WE CAN ONLY TELL THE DIFFERENCES BECAUSE OF THE OBSERVED SIGNATURE! We cannot, STRICTLY, rule one or the other out with logic. BUT we posit explanations with FEWER entities. The argument, therefore, DOES NOT help the argument for god UNLESS we allow entities to be endlessly replicated beyond which that we can justify experientially.

 

[jimbob]

ID - if you won't understand anything I say you're going to continually be asserting X is not Y at me.

Yes ID. I know X is not Y. You are fighting an argument I am not making and will continue doing so until you STOP asserting X is not Y and start thinking about under what circumstances one would say 'X is Y' even though this is never, EVER, true if X and Y are physically separate objects.

Both processes are iterative. Later iterations are "more suited" than earlier iterations.
That is the totality of their similarity.

Do you have any more similarities that I have missed?

jimbob - I will ask this again: what part of the reproduction of the cell is 'self' reproduction?

If you break it down there is no 'self' being replicated - there are lots of different microbiological devices working in tandem to achieve that higher abstraction but there is no literal 'self' replication.

Once you are willing to accept this it seems quite a fallacy to insist that the mechanism of reproduction must be intimately bound to the mechanism of representation. Is the machinery that copied DNA external to the DNA? It sure doesn't seem to be on the inside!

So I yet again come back to the point that neither of you two seem even able to hear: you HAVE to be able to work at different levels of abstraction. At the cell level there is self-reproduction. At the intra-cell level there is not. What about that is not clear?

The cell is self-replicating at the level of the cell. (That is tautological) ands agreed.

Do you see that this is fundamentally different from selecting which photocopies are to be themselves photocopied? Or indeed from selecting which iterations of an evolutionary algorithm are used as the basis for the next generation.

Are you arguing for a metasystem made up of engineers and artefacts as some mystical whole?

If the machines could hijack the factory to make imperfect copies of themselves, then they would be subject to Evolution, as then they would be self-replicating.

The important feature is that the copies carry the templates to produce further copies, and are instructed to produce the copies. A robot carrying around its own blueprints, but not being able to act on them would not count. One that was programmed to tread its blueprints and act on them would be self-replication, if the template included instructions for also copying the template. Such a robot design would truly evolve if the template copying process was imperfect.

 

[cyborg]

Do you have any more similarities that I have missed?

You've missed out quite a few things you've assumed as existing.

The cell is self-replicating at the level of the cell. (That is tautological) ands agreed.

Good.

Do you see that this is fundamentally different from selecting which photocopies are to be themselves photocopied?

At which level?

Are you arguing for a metasystem made up of engineers and artefacts as some mystical whole?

Only if I used those words.

If the machines could hijack the factory to make imperfect copies of themselves, then they would be subject to Evolution, as then they would be self-replicating.

And at what level do you think you could apply the concept of 'hijacking'?

The important feature is that the copies carry the templates to produce further copies, and are instructed to produce the copies.

Not quite. Those features are only important at levels of abstraction where there is a distinction to be made about what things are implementing copying.

At a finer granularity self-copying doesn't exist. Therefore it cannot really be said to 'exist' at the coarser level of granularity since their behaviours arose from non-self-copying objects. If you asserting that self-copying is something that 'exists' then you are saying that self-copying cannot be explained by a system where self copying does not exist?

You have unnecessarily replicated entities. This is argument for god.

You are still not able to think on multiple levels.

 

[jimbob]

You still haven't explained why discussing processes with intelligent designers is any use or interest when discussing the theory of evolution. Except to highlight the differences

You've missed out quite a few things you've assumed as existing.

Such as?