Home » Intelligent Design » A robot in the Cambrian era?

A robot in the Cambrian era?

Proverbially,  it is said that if paleontologists were to discover a rabbit in Cambrian era fossil strata, that would be an empirical refutation of macro-evolutionary theory.  UD contributor, News, has therefore raised a “but what about . . . ? “ in light of finding “complex non-marine multicellular eukaryotes in Precambrian strata . . .  ” and specifically:

large populations of diverse organic-walled microfossils extracted by acid maceration, complemented by studies using thin sections of phosphatic nodules that yield exceptionally detailed three-dimensional preservation. These assemblages contain multicellular structures, complex-walled cysts, asymmetric organic structures, and dorsiventral, compressed organic thalli, some approaching one millimetre in diameter. They offer direct evidence of eukaryotes living in freshwater aquatic and subaerially exposed habitats during the Proterozoic era.

As a further kicker, we must observe a date:”one billion years.”

The very first response, by Dr REC, was dismissive:

A longer, more gradual history of Eukaryotes and of colonization of land renders Darwinism more doubtful?

Where things get very intersting is with the onward suggestion of a gradual unfolding of life from simple to complex forms.

Therein lieth the rub: there ain’t no “simple” life forms.

We can see that if we start from the undeniable fact that we find complex cells in the Cambrian fossils.

Somehow, though, the full force of this point has not hit home hard enough.

So, let’s rephrase: what if we were to discover a ROBOT in the Cambrian era fossils?

You ask me: y’mean like this lovely but unsettling young lady?

Fig. A: Actroid-DER, developed by KOKORO Inc for customer service, appeared in the 2005 Expo Aichi Japan. The robot responds to commands in Japanese, Chinese, Korean, and English.  (Source: Wiki, under GNU.)

Nah, more like this:

Fig. B: A position-arm robot  (Source: Wiki, KUKA Roboter GmbH public domain)

Or, to get more direct, like this:

Fig. C: The Ribosome in action, showing the tRNAs acting as amino acid taxis and position-arm pick and place units that click successive amino acids onto an emerging protein (Source: Wiki, public domain.)

But, that’s different!

Not really.

A pick-place robot arm with a tool-tip is a digitally programmed device that carries out pre-programmed instructions. The tRNA molecules are loaded with amino acid monomers (usually, there are twenty different varieties) and on the opposite ends they have anti-codons that are key-lock coded to the successive three-letter, 64-state codons in the mRNA tape. That mRNA tape in turn derives from the DNA code in the nucleus, and the incrementally assembled protein is folded and dispatched to the place where it can best be put to work, often using the Golgi apparatus as a cellular post office.

In short, the macro-scale position-arm robot and the DNA-Ribosome, mRNA, enzymes and proteins, tRNA system that makes new proteins, are both step by step digital code controlled mechanical assembly processes.

And, so we must ask a few questions, to clarify our views on origins of life and of its diverse forms, and the empirically based grounds for these views:

(1) What is the only empirically observed source of codes, algorithms, and assembly lines?

(2) Is it credible that such entities can spontaneously self-assemble from a soup of monomers in a warm little electrified pond, or a hot undersea vent, or the like?

(3) On what empirically grounded basis do you draw your conclusions?

(4) What does such imply about the origin of life?

(5) Since, embryogenesis transforms a unicellular organism into an integrated assemblage of tissues, organs and the like in a complex, functionally organised body plan, requiring a considerable further body of DNA, what does the above imply about the mechanism for the origin of major body plans?

(6) What is the observed, evidence that small undirected genetic changes in populations, culled out by differential reproductive success, can be and most likely was  responsible for the origin of the complex functional organisation and of digitally coded, algorithmic information and associated implementing machinery found in major body plans and species, including our own?

(7) In short, is the following excerpt from Darwin’s summing up in Origin, Ch 15, suitably updated, an empirically — observationally — well-warranted conclusion, or is it in large part an inference on imposition of the concept that scientific explanation, must only be “naturalistic”?  Darwin:

It is interesting to contemplate a tangled bank, clothed with many plants of many kinds, with birds singing on the bushes, with various insects flitting about, and with worms crawling through the damp earth, and to reflect that these elaborately constructed forms, so different from each other, and dependent upon each other in so complex a manner, have all been produced by laws acting around us. These laws, taken in the largest sense, being Growth with Reproduction; Inheritance which is almost implied by reproduction; Variability from the indirect and direct action of the conditions of life and from use and disuse: a Ratio of Increase so high as to lead to a Struggle for Life, and as a consequence to Natural Selection, entailing Divergence of Character and the Extinction of less-improved forms. Thus, from the war of nature, from famine and death, the most exalted object which we are capable of conceiving, namely, the production of the higher animals, directly follows. There is grandeur in this view of life, with its several powers, having been originally breathed by the Creator into a few forms or into one; and that, whilst this planet has gone cycling on according to the fixed law of gravity, from so simple a beginning endless forms most beautiful and most wonderful have been, and are being evolved. [Origin, Ch 15.]

Most of all: why do you draw your particular conclusions? END

  • Delicious
  • Facebook
  • Reddit
  • StumbleUpon
  • Twitter
  • RSS Feed

41 Responses to A robot in the Cambrian era?

  1. …a Ratio of Increase so high as to lead to a Struggle for Life, and as a consequence to Natural Selection…

    How utterly Malthusian.

    [ED: Darwin is EXPLICITLY Malthusian in his thought, as can be seen in his introductory remarks. Origin owes a lot to the Rev Darwin built on (Malthus) and to the Rev he was trying to overturn [Paley] . . . these are part of why Darwinist thought is inescapably riddled with worldview and theological issues. The issue of inference to best explanation on a remote, unobserved past of origins is also a source of worldviews concerns, epistemological concerns (as in what is the strength of warrant per inference to best explanation and how do we evaluate “best”) and of course the issues in Job 38 apply too. ]

    But if no such increase and struggle ensues, what is the basis for selection?

    (1) What is the only empirically observed source of codes, algorithms, and assembly lines?

    1. Samuel F. B. Morse

    2. Mu?ammad ibn M?s? al-Khw?rizm?

    3. Henry A. Ford

  2. 2
    Elizabeth Liddle

    There is one ENORMOUS difference between your ribosome and the other two robots.

    And I submit, it’s the difference that makes the difference :)

  3. I don’t see any enormous difference.

    The robot didn’t make itself, and there’s no evidence to suggest that the ribosome did either. And yet there it is, just like the robots.

  4. Mung,

    I don’t see any enormous difference.

    For one thing, the ribosome’s technology is more advanced.

  5. 5
    Elizabeth Liddle

    Well, the ribosome is part of a completely self-replicating entity.

    The others aren’t.

    The ribosome didn’t “make itself” alone but the organism that it is a component of was “made” by another almost identical organism, which copied itself in order to produce the one containing the ribosome in question.

    It is probably true that the only non-self-replicating machines are those designed by the intelligent designers we call people.

    But self-replication with modification, I would argue, is the alternate explanation for what would otherwise look like it was designed by an intelligent agent.

    I don’t expect you to agree, but it seems to me it’s a point that at least needs to be considered :)

    As I’ve said elsewhere, I actually agree with IDists that there is a signature that human artefacts share with natural artefacts, and for good reason. I get a bit cross with people who argue that specified complexity is a nonsense.

    I just don’t think it’s the signature of what is normally called “intelligent design”. I do think it’s the signature of life, and that it indicates that either the thing is itself a self-replicator or it was made by a self-replicator.

    Discuss.

  6. 6
    Elizabeth Liddle

    Nullasalus: that too.

    Also, the ribosome is less brittle, both literally and metaphorically.

    Human artefacts are typically highly brittle, and are only as robust as their weakest component.

  7. Human artefacts are typically highly brittle, and are only as robust as their weakest component.

    Except ‘human artefacts’ are getting better and better on all fronts. And insofar as it does, the technology gap will keep on shrinking.

    That’s part of why saying… “But self-replication with modification, I would argue, is the alternate explanation for what would otherwise look like it was designed by an intelligent agent.“… doesn’t help much. It’s like positing “made in a factory” in opposition to “made by an intelligent agent”. Agents can employ factories towards ends. They can also employ self-replication with modification towards ends. (Really, we’ve already done this on a certain scale with antennas.)

  8. For one thing, the ribosome’s technology is more advanced.

    I wouldn’t just grant that.

    1. I don’t know how advanced the technology of the ribosome is, or if it can even be classified as a technology.

    2. I don’t know how advanced the technology of the robots are.

    3. I don’t know of any practical way to compare the two in terms of how advanced their technologies are.

    4. Is the technology of the ribosome so advanced that it cannot at all be compared with any known human technological achievement?

    And yet here we are, comparing it ;).

    Perhaps the robots were not the best choice. Perhaps a closer analogy could be found.

  9. 9
    Elizabeth Liddle

    I agree, nullasalus. In fact, I’d suggest that human design is moving towards an evolutionary model :)

    I don’t in fact think they are fundamentally different, except in one respect – because humans are intentional designers we can take shortcuts. But that is proving to be a mixed blessing, because those short-cuts also bypass a lot of search space, which is where GAs come in so useful.

  10. Mung,

    1. I don’t know how advanced the technology of the ribosome is, or if it can even be classified as a technology.

    Well, it’s tough to compare artifacts, period, in a way. Is a pair of wireless headphones more advanced than a digital camera?

    Still, one metric is ‘Can we make something comparable?’ And I don’t think we can create working ribosomes from scratch yet, so in various ways (if we take it as an artifact) it seems fair to call the tech more advanced.

    Of course, I’m of the mind that – if you’re going to start viewing things in the ‘natural’ world as artifacts, there’s no need to stop. Regard it all as artifact. Who needs nature?

  11. I don’t in fact think they are fundamentally different, except in one respect – because humans are intentional designers we can take shortcuts.

    Who says the designer of evolution isn’t an intentional designer? Much less that no shortcuts were taken? Or maybe you’re employing a design filter to detect or rule out designs and designers, eh? ;)

  12. “It is probably true that the only non-self-replicating machines are those designed by the intelligent designers we call people.”–Elizabeth.

    What about the first self-replicating one? Was that not designed? What about the self-replicating feature itself? Was that evolved or designed?

  13. EL @5:

    But self-replication with modification, I would argue, is the alternate explanation for what would otherwise look like it was designed by an intelligent agent.

    I just don’t see that as an ENORMOUS difference. But can we set that aside, since we seem to be moving on into more productive areas?

    What is the evidence for the proposition that ribosomes came about by such a process (self-replication with modification)?

    What is the evidence for the existence of entities capable of self-replication with modification that do not rely on the presence of ribosomes already being there for their capacity to self-replicate?

    If no such evidence exists, what is the basis for your belief?

    I don’t expect you to agree, but it seems to me it’s a point that at least needs to be considered :)

    I have considered it. You’re right, I don’t agree. I’m not aware of any evidence in favor of what you propose.

    I think you’ll find certain ID types quite amenable in the face of evidence. I hope I’m one of them. But don’t ask me to take leaps of faith. :)

    Now, I think it’s only fair to ask you to consider something.

    Consider the seven questions in the OP. Perhaps start with the very first one, as I did.

    What is your answer?

    What is the only empirically observed source of codes, algorithms, and assembly lines?

    1. There are no empirically observed sources of codes, algorithms, and assembly lines.

    2. There are empirically observed sources of codes, algorithms, and assembly lines.

    If 2, what are those sources?

  14. (2) Is it credible that such entities can spontaneously self-assemble from a soup of monomers in a warm little electrified pond, or a hot undersea vent, or the like?

    No. A soup of monomers is hardly even credible, if at all.

    (3) On what empirically grounded basis do you draw your conclusions?

    Santa stopped bringing me presents and the tooth fairy stopped leaving dimes under my pillow.

    (4) What does such imply about the origin of life?

    That life had no origin.

  15. It is probably true that the only non-self-replicating machines are those designed by the intelligent designers we call people.

    And if the universe itself is a machine? Is it busy going about replicating itself?

  16. p.s. I despise the term “self-replication” on philosophical grounds.

  17. EL:

    I find it interesting that when an entity adds to its function the ADDITIONAL capacity of self-replication (and self-assembly), suddenly this enormous increase in complexity is often seen as a way to avoid the empirically known cause of functionally complex organisation; as though information can be had as a free lunch.

    What is the observed evidence for that?

    And, what do we know from the kinematic von Neumann Self Replicator model, which is implemented in the living cell?

    Let us notice the requisites for such as summarised in the just linked:

    (i) an underlying storable code to record the required information to create not only (a) the primary functional machine [[here, for a "clanking replicator" as illustrated, a Turing-type “universal computer”; in a cell this would be the metabolic entity that transforms environmental materials into required components etc.] but also (b) the self-replicating facility; and, that (c) can express step by step finite procedures for using the facility;

    (ii) a coded blueprint/tape record of such specifications and (explicit or implicit) instructions, together with

    (iii) a tape reader [[called “the constructor” by von Neumann] that reads and interprets the coded specifications and associated instructions; thus controlling:

    (iv) position-arm implementing machines with “tool tips” controlled by the tape reader and used to carry out the action-steps for the specified replication (including replication of the constructor itself); backed up by

    (v) either:

    (1) a pre-existing reservoir of required parts and energy sources, or

    (2) associated “metabolic” machines carrying out activities that as a part of their function, can provide required specific materials/parts and forms of energy for the replication facility, by using the generic resources in the surrounding environment.

    Also, parts (ii), (iii) and (iv) are each necessary for and together are jointly sufficient to implement a self-replicating machine with an integral von Neumann universal constructor.

    That is, we see here an irreducibly complex set of core components that must all be present in a properly organised fashion for a successful self-replicating machine to exist. [[Take just one core part out, and self-replicating functionality ceases: the self-replicating machine is irreducibly complex (IC).]

    This irreducible complexity is compounded by the requirement (i) for codes, requiring organised symbols and rules to specify both steps to take and formats for storing information, and (v) for appropriate material resources and energy sources.

    Indeed, this points tot he key observation made by Paley in his 2nd Chapter (building on the watch in the field in the first chapter):

    Suppose, in the next place, that the person who found the watch should after some time discover that, in addition to all the properties which he had hitherto observed in it, it possessed the unexpected property of producing in the course of its movement another watch like itself — the thing is conceivable; that it contained within it a mechanism, a system of parts — a mold, for instance, or a complex adjustment of lathes, baffles, and other tools — evidently and separately calculated for this purpose . . . .

    The first effect would be to increase his admiration of the contrivance, and his conviction of the consummate skill of the contriver. Whether he regarded the object of the contrivance, the distinct apparatus, the intricate, yet in many parts intelligible mechanism by which it was carried on, he would perceive in this new observation nothing but an additional reason for doing what he had already done — for referring the construction of the watch to design and to supreme art . . . . He would reflect, that though the watch before him were, in some sense, the maker of the watch, which, was fabricated in the course of its movements, yet it was in a very different sense from that in which a carpenter, for instance, is the maker of a chair — the author of its contrivance, the cause of the relation of its parts to their use.

    [[Emphases added. (Note: It is easy to rhetorically dismiss this argument because of the context: a work of natural theology. But, since (i) valid science can be -- and has been -- done by theologians; since (ii) the greatest of all modern scientific books (Newton's Principia) contains the General Scholium which is an essay in just such natural theology; and since (iii) an argument 's weight depends on its merits, we should not yield to such “label and dismiss” tactics. It is also worth noting Newton's remarks that “thus much concerning God; to discourse of whom from the appearances of things, does certainly belong to Natural Philosophy [[i.e. what we now call “science”].” )]

    BTW, I have seldom if ever seen this passage discuss3ed in the hasty attempts to dismiss Paley’s point.

    So, which is it: does the additionality of a self replicating facility call for a more advanced type of design, or does it eliminate the need to explain the origin of FSCO/I?

    Why?

    And, is it reasonable to infer that one can account for novel complex body plans of a range sufficient to account for the diversity we see around us, one small increment at a time?

    On what empirical basis?

    (HINT: Is it reasonable to claim that one can convert a Hello World into a complex operating system, one minor change at a time, retaining functionality and bringing in advantages all the way? On what empirically observed and analytical grounds?)

    GEM of TKI

  18. PS: GA’s only work within islands of function with nice trendy fitness functions keyed to pseudo-spatial variables — cf the issue of moving from hello World to a complex program step by step again. Unsurprisingly, GA’s, for all the headlining of evolution by hill climbing — hence warmer/colder oracles — are actually examples of intelligent design and derive their power from active information that puts us on islands of function to begin with. At most they model micro-evo [variation within a general and already functional body plan], which is non-controversial.

  19. Is it reasonable to claim that one can convert a Hello World into a complex operating system…

    Now is the time you’ve all been waiting for. Finally we get to the classic “first” program. Every decent programming book has a “Hello, World” program, and now we know enough to make a “Hello, World” operating system.
    Write Your Own Operating System

    LOL. I can hardly believe there is a hello world operating system.

  20. 20
    Elizabeth Liddle

    @Collin, #15

    hat about the first self-replicating one? Was that not designed? What about the self-replicating feature itself? Was that evolved or designed?

    It’s technically not correct to describe the “first” of a self-replicating series as having “evolved” at least in the Darwinian sense, because until you have self-replication, Darwinian processes don’t kick in.

    But we do observe spontaneous self-replicating molecules (certain peptides), as well of course as crystals and other systems (sand dunes, for instance) so while it’s an interesting question, I don’t think the question as to how the first self-replicating structures from which (in my view, obviously not in yours!) we descended is in principle a problematic one. We may never know for sure, but, for example, theories that postulate lipid “bubbles” in circulating convection currents, whose peptide contents may determine whether or not the “bubble” divides into two, have at least some explanatory potential.

    But let’s suppose, for now, that the first self-replicators were placed on earth by an Intelligent Designer – my contention would be that once you have self-replicators that replicate with variance, where the variants differ in the efficiency with which they self-replicate, you have the beginnings of a “self-designing” system.

    Part of me is prepared to call it “intelligent”, because in many senses it does what human intelligent designers do (though without the human capacity to simulate ahead of execution, i.e. to select prototypes most likely to succeed).

    But I suggest that the seed itself (the first simple self-replicator that kick-starts the process) could be extremely simple, so simple that spontaneous assembly is not implausible, sometime, somewhere.

  21. 21
    Elizabeth Liddle

    OK, Mung, I will attempt to address the questions in the OP, although you will find my answers question-begging (that’s because I think they are beggable!):

    (1) What is the only empirically observed source of codes, algorithms, and assembly lines?

    objection m’lud. I don’t think there IS a unique “source”. So I’m going to give two:

    a) artefacts produced by living things
    b) self-replicating systems

    However, at a stretch I could put them under a single heading:

    Living things.

    That works for me, as long as it is understood to include the process by which populations of self-replicators (i.e. living things) differentially self-replicate.

    (2) Is it credible that such entities can spontaneously self-assemble from a soup of monomers in a warm little electrified pond, or a hot undersea vent, or the like?

    No, I think it is extremely unlikely that anything as complex as a ribosome would spontaneously self-assemble from a soup of monomers.

    However I think it is perfectly plausible that a ribosome could form part of a self-replicating entity (we’d call it a living organism) whose ancestors were much simpler, scarcely living blobs of self-replicating lipid bubbles in warm convecting sea of polymers.

    (3) On what empirically grounded basis do you draw your conclusions?

    Organic chemistry.

    (4) What does such imply about the origin of life?

    It implies that life, i.e. complex self-replicating organisms, could have evolved from simpler self-replicators – so simple that we’d be pressed to recognise them as life at all, and might prefer to call them something else – bubbles of peptide-containing membranes, for instance.

    Or rather it doesn’t “imply” it – what “implies it” is evidence that once you have a) a population of self-replicators in which b) the replications vary in their ability to replicate, you have a “search” algorithm that will increasingly find “solutions” to the problem of persistence.

    By definition.

    (5) Since, embryogenesis transforms a unicellular organism into an integrated assemblage of tissues, organs and the like in a complex, functionally organised body plan, requiring a considerable further body of DNA, what does the above imply about the mechanism for the origin of major body plans?

    It implies that signalling between unicellular organisms must have tended to promote self-replication in the past, and probably that it promoted it by altering the physical conformation of the organism.

    (6) What is the observed, evidence that small undirected genetic changes in populations, culled out by differential reproductive success, can be and most likely was responsible for the origin of the complex functional organisation and of digitally coded, algorithmic information and associated implementing machinery found in major body plans and species, including our own?

    Well, this is obviously a huge question, but I’d suggest that the evidence consists of consilience between a number of kinds of sources: the observation of “micro-evolutionary” changes in lab and filed; evidence that complex algorithms can evolve from simple precursors in computational evolutionary programs (yes, I know, but I can respond to that….); genetic studies that increasingly tell us something about the phenotypic effects of even small differences in alleles; and evidence from both morphological and genetic phylogenies.

    (7) In short, is the following excerpt from Darwin’s summing up in Origin, Ch 15, suitably updated, an empirically — observationally — well-warranted conclusion, or is it in large part an inference on imposition of the concept that scientific explanation, must only be “naturalistic”?

    I think it’s neither conclusion nor inference. I think it’s a theory, from which hypotheses can be, and have been, derived, and tested. I find it well supported.

    Howzat?

  22. 22
    Elizabeth Liddle

    @ kairosfocus, #17

    I find it interesting that when an entity adds to its function the ADDITIONAL capacity of self-replication (and self-assembly), suddenly this enormous increase in complexity is often seen as a way to avoid the empirically known cause of functionally complex organisation; as though information can be had as a free lunch.

    What is the observed evidence for that?

    I guess I see self-replication as the bootstrap algorithm that gets the rest of the self-designing process going.

    And as evidence, I would cite a few natural examples (Chesil Beach, on the south coast of England is an interesting examples), as well as (I’m afraid!) evolutionary computer algorithms, which can start extremely simple but become very sophisticated.

    Yes, in GA, obviously we have a human designer who kick-starts the system with the original self-replicator, although that can be extremely simple, and also sets up the fitness function, but I don’t think that invalidates the evidence.

    I see that at #18 you tackle this point, but tbh this is the point at which I think ID makes its greatest error.

    Yes, the fitness landscape “smuggles in” information, but – so what? Obviously it has to, because the whole thing is a human designed virtual universe.

    In nature, the fitness landscape, is, literally, the landscape – the environment, and, crucially, that landscape includes the evolving population (in other words there are powerful feedback loops).

    And my answer to the classic question: how did the information get into the genome? is, simply: from the environment.

    In other words,considered as a problem solving system, evolutionary processes (in the wild) have as their “problem” (scare quotes definitely required there”) the simple, intrinsic “problem” of persistence (intrinsic because what won’t persist won’t, um, persist) within an environment (on which its persistence depends for fuel).

    So what we have to artifically insert into a computer GA is intrinsic to the whole set up in nature: things that tend to persist in an environment will tend to persist in that environment! The fancy part comes in when you consider that that environment itself consists of other moderately successful persisters all competing with each other for fuel, and of course, other modulators of the environment.

    So there will tend to develop “niches”, in which sub-populations, with different fuel-utilising characteristics can thrive.

    Where I am perhaps unusual in non-IDists,is that I recognise this as an intelligent system and I submit that that is why the output from such a system carries a signature that is also characteristic of human-designed artefacts. However,there are a couple of crucial differences, and I think these are reflected in the two kinds of output.

    But maybe more on that later :)

  23. So I’m going to give two:

    a) artefacts produced by living things
    b) self-replicating systems

    One problem here is that a and b aren’t mutually exclusive: b can be one of the ‘artefacts’ given in a.

  24. 24
    Elizabeth Liddle

    @nullasalus, #23

    Yes indeed, but I don’t see that as a problem. That’s why I’d say that the signature that IDists regard as the signature of “Intelligent Design” is generally, the signature of “Living things”, and artefacts generated by Intelligent Living Things are merely a subset of that.

    Although,tbh, I’d want to go further, and say that the common factor is “self-replicators” (sorry Mung!) because things like crystals and sand dunes also exhibit traces of that signature.

  25. Elizabeth Liddle,

    Yes indeed, but I don’t see that as a problem.

    Because the idea of presenting ‘a’ or ‘b’ as an “either-or” case takes on a lot of water when b is subsumed by a, even in principle. It’s like saying ‘Letters are often written by a) humans and b) computer printers.’ Right, but if humans employed or create computer printers, the distinction isn’t as helpful.

    Although,tbh, I’d want to go further, and say that the common factor is “self-replicators”

    And once again, “self-replicators” are also / can be “artefacts” made by humans and intelligent agents.

    What’s your system for determining whether or not a given self-replicator was ultimately created by an intelligent agent, or not?

  26. 26
    Elizabeth Liddle

    OK, let me try to be clearer:

    What I’m saying is that the common denominator in the “robots” itemized in the OP, it seems to me, is that they are products of decision-trees in which successful prototypes are repeated, usually with variation, and less successfuly prototypes are abandoned.

    In two of the cases, the process is implemented by intelligent human beings, who do the replicating externally, as it were (usually), set the criteria for success/failure, and only implement variants that have a pre-sifted high probability of success.

    In the third case (the ribosome) I suggest that the replicating is intrinsic to the “robot” itself, in that it is a component within a larger self-replicating “robot”; the criteria for success/failure is simply whether the thing succeeds in replicating, and the variants are not pre-sifted so that even variants with very little chance of success are produced, and replicate if they can.

    But in both scenarios, the result is an increasingly sophisticated, responsive, and effective “robot”. The purpose, of course, of the first two is to serve the purposes of their human designers. The purpose of the last is no more than the purpose of the Lilies of the Field. It toils not, neither does it reap, except to have its pattern replicated in future instantiations.

    So while I think that all three bear a common signature, I don’t think it’s the signature of Intelligent Design. I think it’s the signature of replication-with-modification-and-selection-of-what-works, where what-works depends on the Thing in question. In the case of a “natural” self-replication, what works is, by definition, efficient self-replication.

    So I don’t actually have a “system for determining whether a given self-replicator was ultimate created by an intelligent agent or not”. At least, not a fool-proof one. Faced with a complex apparently functional object, the first thing I’d ask was “does it self-replicate”? If the thing in question was in fact Paley’s watch, the answer of course would be “no, it doesn’t seem to”. So I would infer an external Intelligent Designer (and, just as importantly, manufacturer).

    If the answer was “yes, it does” I would have to say that the jury was out – that it could have evolved from, by means of the iterative process I just described, from something so simple it might have been the product of chemistry, or, alternatively, it might be the fiendishly advanced product of a race of intelligent beings that have managed to design self-replicators that evolve themselves.

    But the common pattern I suggest, is not the signature of Intelligent Design, but of replication-with-modification-and-selection-of-what-works.

  27. Elizabeth Liddle,

    If the answer was “yes, it does” I would have to say that the jury was out – that it could have evolved from, by means of the iterative process I just described, from something so simple it might have been the product of chemistry, or, alternatively, it might be the fiendishly advanced product of a race of intelligent beings that have managed to design self-replicators that evolve themselves.

    But again, a ‘product of chemistry’ can also be a ‘product of an intelligent designer’. They’re not exclusive options. Even the variation and selection process could be artifactual – you seem to recognize as much in terms of evolutionary simulations.

    But the common pattern I suggest, is not the signature of Intelligent Design, but of replication-with-modification-and-selection-of-what-works.

    But how “selection” and “variation” takes place with regards to human artefacts differs distinctly from how “selection” is touted to take place in evolution. I doubt very much that many people would view “artificial selection” as interchangeable with “natural selection”, and Darwin himself threw a fit when Asa Gray even suggested the idea that the variation was itself the product of a mind, to say nothing of the selection.

    Abstracting away from the examples as far as you did would be saying that neither the presence nor lack of an intelligent mind’s involvement was the hallmark. Would you say, then, that as far as science is concerned, whether the origin of life – and evolutionary processes in general – were/are the intentional products of a mind is a question that goes unanswered due to the limitations of the method?

  28. EL:

    A quick note, re your:

    I guess I see self-replication as the bootstrap algorithm that gets the rest of the self-designing process going.

    And as evidence, I would cite a few natural examples (Chesil Beach, on the south coast of England is an interesting examples), as well as (I’m afraid!) evolutionary computer algorithms, which can start extremely simple but become very sophisticated.

    I am afraid Chesil Beach — I hear the surf fishing there is great for Smooth-Hounds — is not a case of a metabolising entity with the ADDITIONAL facility of a kinematic vNSR, based on stored code and algorithmic execution.

    It is a shingle [rock] beach, traditionally seen as a Tombolo. Its production is a wholly dynamic process using essentially undifferentiated parts, and the beach does not carry out a code based metabolic function, such as we can see with the protein assembly process shown above. There is not a code-based algorithm involved.

    For us to get to that kinematic vNSR, as I pointed out above, we need:

    (i) an underlying storable code to record the required information to create not only (a) the primary functional machine [[ . . . in a cell this would be the metabolic entity that transforms environmental materials into required components etc.] but also (b) the self-replicating facility; and, that (c) can express step by step finite procedures for using the facility;

    (ii) a coded blueprint/tape record of such specifications and (explicit or implicit) instructions, together with

    (iii) a tape reader [[called “the constructor” by von Neumann] that reads and interprets the coded specifications and associated instructions; thus controlling:

    (iv) position-arm implementing machines with “tool tips” controlled by the tape reader and used to carry out the action-steps for the specified replication (including replication of the constructor itself); backed up by

    (v) either:

    (1) a pre-existing reservoir of required parts and energy sources, or

    (2) associated “metabolic” machines carrying out activities that as a part of their function, can provide required specific materials/parts and forms of energy for the replication facility, by using the generic resources in the surrounding environment.

    Also, parts (ii), (iii) and (iv) are each necessary for and together are jointly sufficient to implement a self-replicating machine with an integral von Neumann universal constructor.

    That is, we see here an irreducibly complex set of core components that must all be present in a properly organised fashion for a successful self-replicating machine to exist. [[Take just one core part out, and self-replicating functionality ceases: the self-replicating machine is irreducibly complex (IC).]

    This irreducible complexity is compounded by the requirement (i) for codes, requiring organised symbols and rules to specify both steps to take and formats for storing information, and (v) for appropriate material resources and energy sources.

    You will note that Paley pointed to this issue of additionality as far back as 1806. Of course, he would not have known about digital stored code information processing, but in the relevant sense a cam bar — as was used in automata at that time — is a program.

    It is the irreducible complexity involved and the requirement of codes, language, algorithms and execution machinery to be simultaneously co-adapted and to store information for a system that does something that is above and beyond mere self replication that is crucial.

    When it comes to GA’s, the plain evidence is that they are at best models of intelligently designed micro-evolution. I have yet to see a convincing case that the implied vast contiguous continent of function leading to the prospect of finely graded macroevo is a reality.

    Indeed, the evidence highlighted by Gould in his last book points in a very different direction:

    . . . long term stasis following geologically abrupt origin of most fossil morphospecies, has always been recognized by professional paleontologists. [[The Structure of Evolutionary Theory (2002) p. 752.]

    . . . . The great majority of species do not show any appreciable evolutionary change at all. These species appear in the section [[first occurrence] without obvious ancestors in the underlying beds, are stable once established and disappear higher up without leaving any descendants.” [[p. 753.]

    . . . . proclamations for the supposed ‘truth’ of gradualism – asserted against every working paleontologist’s knowledge of its rarity – emerged largely from such a restriction of attention to exceedingly rare cases under the false belief that they alone provided a record of evolution at all! The falsification of most ‘textbook classics’ upon restudy only accentuates the fallacy of the ‘case study’ method and its root in prior expectation rather than objective reading of the fossil record. [[p. 773.]

    I think this serves to lay out where the key differences in our views lie.

    GEM of TKI

  29. 29
    Elizabeth Liddle

    Nullasalus, re #27

    But again, a ‘product of chemistry’ can also be a ‘product of an intelligent designer’. They’re not exclusive options. Even the variation and selection process could be artifactual – you seem to recognize as much in terms of evolutionary simulations.

    Sure they aren’t exclusive options. But that’s why you can’t infer one or the other. Could be either or both.

    That’s why I think there’s a flaw in the whole “Design Inference” thing. I agree that there’s an inference to be made from “Specified Complexity” but I don’t think “Design” is the only inference. I think there are two, and if the thing in question is a self-replicator, there need by no presumption of a Designer. If it isn’t, then, maybe,there is.

    But how “selection” and “variation” takes place with regards to human artefacts differs distinctly from how “selection” is touted to take place in evolution. I doubt very much that many people would view “artificial selection” as interchangeable with “natural selection”, and Darwin himself threw a fit when Asa Gray even suggested the idea that the variation was itself the product of a mind, to say nothing of the selection.

    Well, “artificial selection” (or what Darwin just called “selection) is of course, by definition, selection by an Intelligent Agent (usually a farmer or an animal breeder), but it could also be viewed as a special case of “natural selection” in that the environment in which selection is taking place contains a farmer who obstructs breeding if he doesn’t like the individual!

    However, now of course, variation itself in living things can be the product of an Intellgent human being, in that we can now genetically modify living things – specify the variance – as well as select the phenotypes we like.

    Would you postulate that the Intelligent Designer of ID directed the evolution of terrestrial life-forms similarly both had a hand in selecting appropriate phenotypes and adjusting the genomes to produce the desired phenotypes?

    Abstracting away from the examples as far as you did would be saying that neither the presence nor lack of an intelligent mind’s involvement was the hallmark. Would you say, then, that as far as science is concerned, whether the origin of life – and evolutionary processes in general – were/are the intentional products of a mind is a question that goes unanswered due to the limitations of the method?

    In a sense, yes, although I think Occam’s razor has a role. I don’t,for reasons I’ve given, think that an Intelligent Designer is the only valid inference from the observation of Specified Complexity, and, when the Complex item is itself a self-replicator, the parsimonious conclusion seems to me to be that it evolved by natural means.

    So I don’t rule out the Intelligent Designer on empirical grounds, but I do strongly dissent with the claim that it is the only inference from the data.

    On the other hand, if, for theological reasons, one wants to postulate a role for God that is consistent with the data, I don’t see why you can’t allow him/her to nudge the chemistry where it helps things along, and even organise the odd stroke of luck to help ensure that an ultimately beneficial variant gets a bit of help from what the naive atheist would simply put down to “drift” :)

  30. F/N: I clip the next key block, to insert comments on points:

    _______________

    >> the common denominator in the “robots” itemized in the OP, it seems to me, is that they are products of decision-trees in which successful prototypes are repeated,

    a: as created by intelligent designers, in a context of coded programs used in their operation

    usually with variation, and less successfuly prototypes are abandoned.

    b: Again, by intelligently directed choice

    In two of the cases, the process is implemented by intelligent human beings, who do the replicating externally, as it were (usually), set the criteria for success/failure, and only implement variants that have a pre-sifted high probability of success.

    c: In short, you acknowledge the point, i.e that it is intelligence that is seen empirically as capable of developing a robot (and presumably the arms and legs of Fig A are similar to the position arm device in B).

    In the third case (the ribosome) I suggest that the replicating is intrinsic to the “robot” itself,

    d: The problem here is the irreducible complexity in getting to function, as outlined again just above.

    in that it is a component within a larger self-replicating “robot”;

    e: it is a part of both the self-replicating facility and the metabolic mechanism, and uses a position-arm coded entity the tRNA that key lock fits the mRNA tape that is advanced step by step in protein assembly, and has as well a a starting and a halting process.

    the criteria for success/failure is simply whether the thing succeeds in replicating,

    f: The entity has to succeed at making proteins, which are in turn essential to its operations, i.e this is chicken-egg irreducible complexity. I gather something like up to 75 helper molecules are involved

    and the variants are not pre-sifted so that even variants with very little chance of success are produced, and replicate if they can.

    g: If the ribosome does not work right the first time, there can be no living cell that can either metabolise or self-replicate.

    h: For that to happen, the ribosome has to have functioning examples of the very item it produces — code based, functioning proteins.

    i: In turn, the DNA codes for the proteins have to be right, and have to embrace the full required set, another case of irreducible complexity.

    j: In short, absent the full core self replicating and metabolic system right from get-go, the system will not work

    But in both scenarios, the result is an increasingly sophisticated, responsive, and effective “robot”.

    k: This sort of integrated irreducible complexity embedding massive FSCO/I has only one observed solution and cause: intelligent design.

    l: In addition, the degree of complexity involved goes well beyond the search resources of the observed cosmos to credibly come up with a spontaneous initial working config.

    m: So, we see here a critical issue for the existence of viable cell based life, and it points to the absence of a root for the darwinian style tree of life. >>
    ________________

    I trust the key issue is now apparent.

    GEM of TKI

  31. 31
    Elizabeth Liddle

    I’m afraid I’m getting a bit lost in terminology here, Kairosfocus – what’s a vNSR?

    Yes, Chesil Beach is a dynamic system, but it acts as a pebble-sorter. It’s also, in some respects, a self-replicator, or at least a self-renewer. The surface is constantly renewed, and the size of pebbles on the surface at any given point “codes” for the size of pebbles that will lodge there after the next wave.

    It’s obviously extremely simple, but it does, interestingly, embody “information” – so much so that legend has it that fisherman, washed up on the beach at night, could extract that information i.e. where they were on the 18 mile stretch by rolling a pebble in their palms, and estimating the size.

    But I think some of our mis-connection arises from non-shared understanding of that dratted word “code”.

    I would argue that in some senses, both the Chesil Beach surface and a stretch of DNA are a “code” in that they both determine that some things will happen (THIS pebble will lodge, THAT pebble will role back; THIS amino acid will be constructed, THAT will not). However, they differ, I would argue, from, say, ASCII code, in that they are not arbitrary. The pebbles at the 4 mile point along Chesil Beach code for similar pebbles because of their physical properties; a DNA codon codes for a particular amino acid because of its chemical properties.

    Whereas we can easily re-assign computer codes to anything we want (and sometimes do).

    So I think we have to be very careful about using the word “code” in both contexts – there is a danger of equivocation that can lead, I suggest, to erroneous inferences.

    However, I think you are absolutely right when you reference a cam. What a cam codes for is not arbitrary at all. And I suggest that DNA is much closer to a cam than a digital code.

    (When I was little, my mother had a lovely Singer sewing machine, and the gadget that wound the thread on the long cylindrical spool was a beautiful heart-shaped cam. I used to love to watch the thread feed move from side to side along the spool, perfectly laying down coils of thread, directed unerringly by the revolving cam).

    Yes, I also agree that GAs are a model of “microevolution”, depending of course by how you define that somewhat contentious term!

    Although we can certainly set up GAs to model speciation.

    And I don’t think Gould’s punctuated equilbrium is particularly difficult to model – for example, I once myself, in an extremely simple model, devised an evolutionary algorithm (essentially a learning algorithm) that would move nicely between optima when I changed a few parameters in the fitness function.

    And this of course happens in nature.

    Anway, thank you for your helpful post – yes it does clarify where some of the differences lie, and I will think more on it.

    Cheers

    Lizzie

  32. Elizabeth Liddle,

    Sure they aren’t exclusive options. But that’s why you can’t infer one or the other. Could be either or both.

    Except it’s more tricky than that. We have undeniable evidence of intelligent agents bringing about everything from self-replicators to all manner of artifacts to products of chemistry. We don’t have comparable evidence that such things are coming about ultimately unguided – we get as far as assumption in such a case, certainly as far as science goes.

    I agree that there’s an inference to be made from “Specified Complexity” but I don’t think “Design” is the only inference.

    The only inference, as in the only logical possibility? Perhaps, but ID doesn’t demand that the inference to a designer be that airtight – just probable given what we know, qualified appropriately.

    However, now of course, variation itself in living things can be the product of an Intellgent human being, in that we can now genetically modify living things – specify the variance – as well as select the phenotypes we like.

    And that’s not static – our own technology continues to advance and get even better. You previously cast
    self-replicators as possibly being a “fiendishly advanced product”. Will it be all that fiendish in 100 more years of technological progress on our part? 500? More?

    Would you postulate that the Intelligent Designer of ID directed the evolution of terrestrial life-forms similarly both had a hand in selecting appropriate phenotypes and adjusting the genomes to produce the desired phenotypes?

    Good question! Seems like a possibility. How would science go about determining this presence or lack, one way or the other?

    I don’t,for reasons I’ve given, think that an Intelligent Designer is the only valid inference from the observation of Specified Complexity, and, when the Complex item is itself a self-replicator, the parsimonious conclusion seems to me to be that it evolved by natural means.

    Natural means? As opposed to, what, the supernatural voodoo-magic Craig Venter uses in his lab?

    I don’t say that to be snarky. I mean that just what ‘natural means’ ultimately are – whether they are a guided thing, or an unguided thing – is an open question. Whether due to direct intervention, or foresight, or a combination of both, or neither. How does science go about determining the answer to a question like this? Or does it not do so?

    On the other hand, if, for theological reasons, one wants to postulate a role for God that is consistent with the data,

    Well, ‘theological reasons’ can also lead one to saying a God did not do this or that. (Ayala, to use a quasi-theistic example. Atheist examples are obvious.) Now, I think philosophical and theological reasons are downright respectable. But I’m asking how science detects the presence or lack of intelligence in these “artefacts” (“possible-artefacts”?), especially given that any “natural” mechanism you could name, would just be a mechanism an intelligent agent in principle could develop and employ?

    Or, again, does science as science just clam up there and does not deal with such? There’s nothing wrong with recognizing the limitations of a field or method after all.

  33. 33
    Elizabeth Liddle

    OK, that’s interesting.

    Well, once we are talking about probabilities, my priors for an object that seemed to exhibit Specified Complexity but did not self-replicate would be high for an Intelligent Designer. However, if the object did self-replicate, then my priors would depend on how much evidence I had for a designer, I guess. I mean, if I found the thing on a computer, I’d have a strong prior that it was Intelligently Designed, or, at least, that the algorithm was Intelligently Designed.

    And I think that’s another place we have to be careful – is the result of a GA “intelligently designed”? I’d say no, even though the GA itself might be “intelligently designed”.

    Anyway, I think we have to keep these things clear.

    Getting a bit garbled as it’s getting late, but in response to your last question:

    I don’t think science “clams up”, but I do think that some questions aren’t necessarily answerable, not because they aren’t scientific questions, but because (to quote Dennett again) they are “historically inert” – the data that would inform us is simply gone, lost to entropy.

    But we can attempt, at least, for example, to determine whether a boulder fell from the cliff top or was pushed. I certainly think (in fact I think it’s abundantly clear) that figuring out whether an Intelligent, and indeed Intentional, agent was responsible for an event or an object is a perfectly valid scientific endeavour.

    And yes, the division between “natural” and “un-natural” is also (heh) somewhat unnatural. At least it is if you count people as natural (and I do :)).

    In some senses Craig Venter is just another stochastic event in the history of the organisms he deals with. Mutations have many causes, and one of them is called Craig Venter.

    But you could probably call him an Intelligent Designer of sorts as well…

    gottagotobed….

    Cheers

    Lizzie

  34. von Neumann Self Replicator.

  35. code [k??d]
    n
    1. (Electronics & Computer Science / Communications & Information) a system of letters or symbols, and rules for their association by means of which information can be represented or communicated for reasons of secrecy, brevity, etc. binary code Morse code See also genetic code
    2. (Electronics & Computer Science / Communications & Information) a message in code
    3. (Electronics & Computer Science / Communications & Information) a symbol used in a code
    4. a conventionalized set of principles, rules, or expectations a code of behaviour
    5. (Electronics & Computer Science / Communications & Information) a system of letters or digits used for identification or selection purposes
    vb (tr)
    (Electronics & Computer Science / Communications & Information) to translate, transmit, or arrange into a code
    [from French, from Latin c?dex book, codex]

    Collins English Dictionary – Complete and Unabridged © HarperCollins Publishers 1991, 1994, 1998, 2000, 2003

  36. LIz:

    “I certainly think (in fact I think it’s abundantly clear) that figuring out whether an Intelligent, and indeed Intentional, agent was responsible for an event or an object is a perfectly valid scientific endeavour.”

    As happens in archaeology all the time. Is there evidence of an intelligence at that time and place to give strength and plausibility to the inference? If an object was found in a strata which was reliably dated to 20 million years bp which seemed to show design then it falls upon the investigator to find other evidence to support the design hypothesis. Multiple lines of evidence carry weight. A single datum is interesting, worth investigating, but would you stake your life on it??

  37. 37
    Elizabeth Liddle

    von Neumann Self Replicator

    Thanks!

  38. Elizabeth Liddle,

    However, if the object did self-replicate, then my priors would depend on how much evidence I had for a designer, I guess. I mean, if I found the thing on a computer, I’d have a strong prior that it was Intelligently Designed, or, at least, that the algorithm was Intelligently Designed.

    Well, what sort of priors are we limited to here? Strictly tested-in-the-lab scientific? It seems like the only thing we can test for without assumption is design by an intelligent agent – whether nature is a similar product is both the open question, and what we’re working inference towards.

    If we need priors that go beyond science, or work in assumptions that aren’t open to scientific testing, it seems like we’re going beyond science anyway, at least in the relevant sense.

    I don’t think science “clams up”, but I do think that some questions aren’t necessarily answerable, not because they aren’t scientific questions, but because (to quote Dennett again) they are “historically inert” – the data that would inform us is simply gone, lost to entropy.

    Sure, I’d go for that. But they seem to add up to the same thing – either the questions are not scientific (and thus, of course, science can’t answer them), or the data is not available to science (whether ‘lost to entropy’, just plain unavailable due to practical or theoretical limitations, etc), no answer is available from science. I think that’s in essence the same as clamming up.

    But we can attempt, at least, for example, to determine whether a boulder fell from the cliff top or was pushed. I certainly think (in fact I think it’s abundantly clear) that figuring out whether an Intelligent, and indeed Intentional, agent was responsible for an event or an object is a perfectly valid scientific endeavour.

    Sure – about a particular agent, given particular assumptions, given the right data. Trying to figure out whether an intelligent (human) agent or ‘nature’ was responsible for X is one thing. Trying to figure out whether nature itself is the work of an intelligent agent – in whole or in part – is another thing.

    In some senses Craig Venter is just another stochastic event in the history of the organisms he deals with. Mutations have many causes, and one of them is called Craig Venter.

    I think that’s stretching the point not only beyond science’s domain, but to the point of explosion. So intelligent agents can be the cause of variation after all? They can be the cause of natural selection? Evolution by natural selection neither concludes nor rules out intelligent causes as far as science is concerned, even in our own far-flung evolutionary history?

    I mean, think of an alternate way of putting that. “Evolutionary theory only says that variation and selection results in various changes and development in biology and natural history. Said variation and selection can in principle be an intelligent agent.” I can think of a few biologists who would react to that idea with horror.

  39. F/N: I am a bit tired just now, having just come from a meeting.

    But, I just need to point out that the issue with self-replication is not the succession of generations but the source of generation no 1.

    Which requires the creation of a functionally specific, irreducibly complex, information rich, metabolic system that has the additional facility of an algorithmic process of self-replication on coded, stored information.

    A clue for the source of such is this: where do codes and algorithms and assembly line processes come from?

    In short, we are back at the seven questions in the original post.

    GEM of TKI

  40. EL:

    I pause, even before turning in:

    I would argue that in some senses, both the Chesil Beach surface and a stretch of DNA are a “code” in that they both determine that some things will happen (THIS pebble will lodge, THAT pebble will role back; THIS amino acid will be constructed, THAT will not). However, they differ, I would argue, from, say, ASCII code, in that they are not arbitrary. The pebbles at the 4 mile point along Chesil Beach code for similar pebbles because of their physical properties; a DNA codon codes for a particular amino acid because of its chemical properties.

    Actually the main chemistry part is in the backbone of DNA strands, which is more or less independent of which of AGCT is on a given location, the information is in the side-chain.

    As also happens with proteins.

    And in tRNA, the AA is held on the side opposite to the anticodon. In short we see here a translational assignment.

    Here is Crick, shortly after the discovery of DNA, to his son Michael on Mar 19, 1953:

    Now we believe that the DNA is a code. That is, the order of bases (the letters) makes one gene different from another gene (just as one page of print is different from another)

    In short, the information is stored in the SEQUENCE of the side chains.

    It is read based on key-lock fitting, but the info in a magnetic tape is also read in a similar way, based on a modulation pattern. The info is in the code used to control the modulation, not in the physics of detection.

    The pebbles on Chesil beach undergo hydrodynamic sorting based on where they are along the beach but that is not an arbitrary code. In other words, at a given point along the beach, pebbles of a given size will lodge or roll. But, that is not a highly contingent process that can flexibly vary per a code. If a pebble is right size in a graded pattern, it will settle, if not it will roll.

    The sequencing of codons along DNA is more or less free [there is no control that only a particular nucleic acid will fit here on any given strand], so that we can specify any particular sequence of AAs as will do a job that depends on what happens after they are folded and sent to their working site. Similarly, assignments of codons to Amino Acids is more or less not physically determined. It is a code, in short: a certain 3-letter pattern MEANS, add a certain amino acid, and another means add a different amino acid. And finally certain codes mean STOP.

    Where there is a constraint is cross-chain: the double helix has corresponding bases that match from one strand to the other. But that does not determine the sequence within the original strand, only how the two strands will match, building in a redundancy.

    In short, we are really dealing with a digital code. BTW, the original proposal for the kinematic vNSR proposed using prongs of a given length to store information. Maybe, they go the idea from Braille.

    GEM of TKI

  41. F/N: Further response here.

Leave a Reply