Home » Comp. Sci. / Eng., Darwinism, Science » Why Mathematicians, Computer Scientists, and Engineers Tend to be More Skeptical of Darwinian Claims

Why Mathematicians, Computer Scientists, and Engineers Tend to be More Skeptical of Darwinian Claims

Larry Moran’s presentation in a comment in Granville Sewell’s UD post, I found not particularly persuasive, for the following reasons. I’m not interested in definitions of science; I’m interested in how stuff actually works. I’m perfectly amenable to being convinced that the complexity, information content, and machinery of living systems can be explained by stochastic processes filtered by natural selection, and I would not even demand hard evidence, just some rigorous argumentation based on the following:

1) A particular aspect of any living system that displays a machine-like function (such as a ribosome).
2) Some specifics about what random genetic changes (of any type) would be required to engineer intermediate forms.
3) A reasonable estimate about the likelihood of these random changes occurring.
4) Another reasonable estimate about the likelihood of the hypothetical intermediate forms providing a statistically significant survival value.
5) Some kind of evidence or even reasonable conjecture that the number of individuals and reproductive events could provide the requisite probabilistic resources. Appeals to “deep time” are irrelevant.

These are the kinds of challenges that those of us involved in mathematics, computer science, and engineering tend to present, and the kinds of questions we tend to ask, because we must demonstrate that our stuff can actually work in the real world, or at least that it has a reasonable prospect of working in the real world. That’s why many of us tend to be skeptics.

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

86 Responses to Why Mathematicians, Computer Scientists, and Engineers Tend to be More Skeptical of Darwinian Claims

  1. That is it isn’t it Gil. “Why are engineers et al. tending to be skeptics of evolutionary claims”, well they work in the real world where hand waving and appeals to magic, wait sorry, “deep time”, don’t cut it.

  2. Gil -

    I know this is inappropriate here, but I can’t find your email address anywhere. Based on your background, your posts here, and your interest in stuff that “can actually work in the real world”, I’d like to get your opinion on an artificial intelligence matter. Please write me: help at osmosian dot com

    EDIT: Modified the email address so you don’t get picked up by the spambots. – Patrick

  3. Hi Gil:

    I, too would like to see a sound — or at least reasonably plausible — demonstration of the capacity of RV + NS to get to the mechanisms of life at cellular level, and the body-plan innovations as well.

    I invite Dr Moran to show how, maybe starting with the toy example here.

    GEM of TKI

  4. A great line from kairosfocus’ link in post 3: “Only the layman seems to see the problem with this logic.” I guess there is something to that “can’t see the forest for the trees” expression after all!

  5. The Machine

    Two people are looking at a finely-tuned, highly sophistocated machine that produces massive amounts of complex, functioning items.

    1: “It’s remarkable what purposelessness can do.”

    2: “You think this machine came about and operates without purpose?”

    1: “Yup.”

    2: “That’s incredible. What makes you think that?”

    1: “Come on. Everyone knows that everything in life is purposeless.”

    2: “Why do you say that? All machines I know of have been designed and built by intelligent beings.”

    1: “Yeah, but the intelligent beings came about through a process that’s purposeless.”

    2: “How do you know?”

    1: “Come on. Everyone knows that.”

    2: “Let me take another approach. Can you show me a working model of a machine that generates any type of functioning thing?”

    1: “Here. Here they are. Pointing to Avida, etc.”

    2: “But these all have active information and an overarching purpose built-in! How can you claim that these are models for this machine?!”

    1: “Well, I study portions of this machine for a living. All the books and all the important people in the field say it’s all purposeless. So there. Who are you to question that?”

    2: “I design and build machines every day. When we have an idea for a machine, before committing large amounts of money to its developent, we always need to prove that it will work by designing and building a working model. You want me to commit to the idea that this thing operates completely purposelessly, yet you haven’t shown a proof-of-concept working model for your idea of how it works. Sorry, no can do.”

  6. I was a philosophy and computer science double-major in college. As a philosophy student, I am very interested in the definition of science. As a computer science student, I recoil in horor at the though of injecting random code into my working system unless my working system was designed to handle random injections.

  7. It’d be nice if you distinguished between applied and pure mathematicians. Some of us could care less about things “working in the real world.”

  8. That’s not to say applications don’t matter. But frequently, the pure mathematicians invent/discover new math, and later it is found that it “works” in the real world.

    Incidentally, though I support evolution, this is the reason why I could never be an atheist. Math is just too nice and neat.

  9. I have to wonder if there is a correlation between years of education in the life sciences and the ability to apply common sense to biology. Just observing the contributions of the various engineers here at this site is all the answer I need.

  10. GilDogen
    Further to your criteria 1)-5):

    6) Show a reasonable probability that the mutation will become fixed in the population; and

    7) Show why it’s benefits are not overwhelmed by harmful mutations.

    5), 6) and 7) quantitatively distinguish real science from a hand waving “Just-So” story . (with apologies to Kipling)

    Preliminary estimates can be obtained by applying the major population genetics models as summarized in the appendix of John C. Sanford’s book
    “Genetic entropy and the mystery of the genome” 2005 ISBN-13: 978-1599190020.

    We have yet to see any model of “evolution” that can quantitatively overcome such enormous reproductive improbabilities.

    A critical factor is the probability of beneficial to harmful mutations. The results in the literature range from 1:10,000 to 1:1, million.

    Anyone familiar with geometric series can begin to comprehend the consequences of 10,000 to one million times more harmful mutations then beneficial ones.

    (Or might that be why most evolutionists avoid calculus?!)

    Perhaps Larry Moran would care to demonstrate mathematically how “evolution” achieves such wonders of creative generation.

  11. Gil,

    Why not demand a lot?
    If an archaeologist dug up a shirt and floated a hypothesis that it was the result of purely stochastic processes, I’d want every detail.

    Thanks

  12. DLH

    You’re right. Mine is just a bare-bones list. It might be interesting to see what more could be added to it.

    I think this exercise demonstrates something that should be transparently obvious: only evolutionary biology is given a free pass when it comes to addressing these kinds of challenges and questions. It is the only “scientific” discipline that is not held to reasonably rigorous standards of evidence and logic. I think the reason for this is also obvious: blind-watchmaker mechanisms simply must have the creative power attributed to them, because the possibility that they don’t is philosophically unacceptable.

  13. Help me here;

    whenever I mention this to folks,they respond: well, yes, this is an issue for “origin of life”, but that’s a totally different question from “evolution”, and then I get called some label to imply….well you know

    thanks

  14. well, about to show his total ignorance, but has anyone yet even found a way that a nucleotide could arise in nature? I feel I would have heard about this if it had happened. Prof Shapiro had said in his book that that was unknown; I would think it would have been trumpeted at the level of the Miller experiments; and if that’s not something that would be important, why not?

  15. Dear Professor Moran,

    I think the problem that most of us “IDiots” have with the accepted mechanisms of evolution is that the actual data of what HAS occured (i.e. the fossil record) is far different from what Darwinists such as yourself claim has occured. Forgive me for questioning your authority, I’m merely a humble undergraduate. But isn’t a scientific theory supposed to be based on data? The “important mechanisms of evolution” that you were so kind to explain to us – natural selection and random genetic drift – require a gradualist model of evolution. Even as you stated in your post, they work “over many generations” and “over a long period of time.” The problem is that the fossil record doesn’t support such a model. And it’s not only “IDiots” who acknowledge that fact:

    “Phyletic gradualism was an a priori assertion from the start – it was never ‘seen’ in the rocks.”

    - Stephen Jay Gould and Niles Eldredge, “Punctuated Equilibria: the tempo and mode of evolution reconsidered” Paliobiology, 1977. Vol 3 pp 115-151, taken from the abstract.

    “Major transitions in biological evolution show the same pattern of sudden emergence of diverse forms at a new level of complexity…No intermediate ‘grades’ or intermediate forms between different types are detectable.”

    - Eugene Koonin. “The Biological Big Bang model for the major transitions in evolution” Biology Direct, 2007 2:21. Taken from the abstract.

    “Neither classical morphology nor standard Darwinian analysis has provided truly satisfying explanations of such major body plan innovations as the…origin of the vertebrates by body axis inversion…These appear to…have little to do with gradual shifts in population gene frequencies under drive from natural selection.”

    - Aaron Filler (2007) “Homeotic Evolution in the Mammalia: Diversification of Therian Axial Seriation and the Morphogenetic Basis of Human Origins.” PLoS ONE 2(10): e1019 doi:10.1371/journal.pone.0001019

    So it seems that a mechanism different from the one asserted by you and most other Darwinists, is needed. If staunch evolutionists, such as those quoted above, can acknowledge that, then so can we. Perhaps if Darwinists spent less of their intellectual energy speculating how to fit the data to their a priori bias, and more of it making scientific inferences directly FROM the data, their efforts would yield more than just the occasional clever insult of an ID proponent.

    Sincerely,

    A Clumsy Brute

  16. While we’re talking about “rigorous argumentation,” could you offer a rigorous definition of “a machine-like function”?

  17. 7) Show why it’s benefits are not overwhelmed by harmful mutations.

    Done. For asexuals, apply Mike Lynch’s mutational meltdown models – if the population size is large enough, they don’t melt down.

    The situation is quantitatively different for sexuals, as I blogged about in the summer, the punchline is that the advantages of sex in combining “good” alleles becomes greater at smaller population sizes. IIRC, the relevant paper is by Otto and Nick Barton (Barton is a mathematician, BTW. At the previouds ESEB he gave his talk using Mathematica).

    Bob

  18. Bob O’H

    Thanks for the reference Bob to mutational meltown. e.g. Lynch on mutational meltdown in both asexual and sexual populations.

    Michael Lynch, Indiana University

    Michael Lynch home page

    Mutational Meltdown John Conery and Michael Lynch

    Lynch, R. Bürger, D. Butcher, and W. Gabriel. 1993. Mutational meltdowns in asexual populations. J. Heredity 84: 339-344.

    Gabriel, W., M. Lynch, and R. Bürger. 1993. Muller’s ratchet and mutational meltdowns. Evolution 47: 1744-1757. PDF

    Lynch, M., J. Conery, and R. Bürger. 1995. Mutational meltdowns in sexual populations. Evolution 49: 1067-1080 PDF

    Right off, there appears to basis for avoiding mutational meltdown in large asexual populations vs small. There may be a difference in fixation, but that does not lessen the total mutational load. Worth examining Lynch’s arguments in detail to identify the fallicy is.

    One challenge is that greater advantages of sex in smaller populations does not explain anything about how sexual differentiation originated. Smaller population sizes reduce the probabilities even more.

    The larger challenge is how to explain origin and survival of more complex biological systems with slow reproduction rates such as man , elephants and whales. This is particularly challenging in light of the ongoing increase in mutational load much greater than beneficial mutations.

  19. I don’t think that most ‘evolutionists’ avoid calculus. Is that true?

  20. “But frequently, the pure mathematicians invent/discover new math, and later it is found that it “works” in the real world.”

    Is this a source of dissapointment for them ? I did hear comments to that effect from someone :)

  21. [...] Why Mathematicians, Computer Scientists, and Engineers Tend to be More Skeptical of Darwinian Claims I’m not interested in definitions of science; I’m interested in how stuff actually works. I’m perfectly amenable to being convinced that the complexity, information content, and machinery of living systems can be explained by stochastic processes filtered by natural selection, and I would not even demand hard evidence, just some rigorous argumentation based on the following: [...]

  22. there are a lot of problems with sex. Half the population isn’t reproducing and must spend lots of time courting and attracting unwanted attention from predators.

  23. Corey
    From a “gut feel” or ad hoc observation, a foundational understanding of calculus is very important to most commercial designers (mechanical engineers, chemical engineers, software engineers etc.)

    By contrast, the appearance is that most evolutionists (like Richard Dawkins) rarely use population genetics models possibly because from taking biology, zoology etc and avoiding calculus. The work of most population geneticists who do need to use such mathematical models appears to be little known. John Sanford’s summary of population models appears devastating to evolution.

    Google search:
    Calculus engineering 798,000
    Calculus evolution 554,000
    (and most of the latter appear to refer to evolutionary algorithms.)

    Thus the inference that evolutionists avoid calculus.

  24. I like Clumsy Brute’s list above in #15, but in using quotes like that this is what I get (so please help if you can):

    In trying to convince colleagues that ID makes sense because Darwinism has limits I got this reply (from an engineer, no less): “So? God of the Gaps, blah blah, just because science hasn’t figured it out yet doesn’t mean it was designed.”

    So I challenge him to come up with a better explanation. He can’t but says “science will figure it out like they’ve figured out gaps before, you prove design to me.”

    So I say “flagellum” and he points me to talkorigins or whatever and says it’s a dead horse and another “God of Gaps”. I looked it over, and the details are too complicated, so he may or may not have a point.

    I say “irreducible complexity” and he says “show me something other than the flagellum”. I say bloodclotting and he starts talking Dover. I talk about Edge of Evolution and he points at a lot of blog entries (although nothing concrete and human readable) about biology details.

    I say CSI and he says it’s not proven and he wants an example and asks “why didn’t Dembski reply to the challenges in such-and-such a paper” (http://www.talkreason.org/arti.....embski.pdf, but the math is too high level for me however the challenges seem easily refuted).

    I say bombardier beetle and many of the other examples that convince me, but he just doesn’t get it. And he’s an engineer, a good “designer” as it were and should be able to spot design a mile away.

    If I mention things like fine-tuning of the universe he starts talking multiverses (!).

    Is there simply something “i can’t tell you what it is, but I know it when I see it” about design? The argument just seems to go absolutely nowhere, in huge circles, like we’re having a big Google and Wikipedia fight.

    I’m not trying to tell the guy his kids have to learn ID or that evolution is completely wrong, just that it shouldn’t be dismissed out of hand. Our little “debates” have an audience and I hate to back down or quit, but I just can’t get anywhere…

    Sorry to rant, just looking for help!

  25. If I mention things like fine-tuning of the universe he starts talking multiverses (!).
    -

    then you won. in 5 billion other universes your friend is a fish that plays golf

  26. So where does that inference come from? The fact that every mention of evolution doesn’t involve calculus? The fact that there are more hits for calculus and engineering than for calculus and evolution?
    Well, the engineering google search i would think should have more hits, but there’s still a lot for the evolution one too, eh?
    I guess my point is do people who believe in evolution have little understanding of calculus?

  27. It’s not about the calculus. The people who come up with such ideas as the multiverse to avoid ID are evolutionists and they certainly know calculus. They’ll take a truck filled with calculus and dump it all over you.

    The real problem is that none of these evolutionists know what works and doesn’t work in the real world.

  28. Re. p.noyola (#24), unfortunately many if not most engineers tend to be hard headed skeptics and materialists if not pathological skeptics like The Amazing Randi. This induces them to reject any notions of ID out of sheer kneejerk reflex. I am an electronics engineer and programmer and have found my point of view to be a distinct minority among my peers. A few engineers somehow break out of the culturally conditioned mindset. They are then able to correlate their unique engineering and software engineering insights with the design of life and ID principles.

  29. This reminds me of something Darwinists (and sometimes the general population) seems to suffer from:

    “Machines-Just-Happen” syndrome

    I was made aware of this when watching the recent movie “Transformers.” In the film, the aliens/robots simply shoot an energy ray at normal houshold items and these items immediately become complex, advanced robotics.

    My engineer mind began thinking “If I were to transform a cellphone into a mobile, autonomous robot, what whould I have to change…?” Then, I thought of at least 10 major components I would have to add hardware-wise, not to mention all the new programming I’d have to intelligently design.

    But no, in the movie all it takes is some energy beam. Of course it is a fictional plot, so I’m not nitpicking; what amazes me, however, is that the audience didn’t recoil in horror at the obvious physical contradiction of what occurred.

    It can be seen in the TV show Heroes (which I love) in that the “ability” to levitate is the result of random mutations. Really, it takes that little special machinery to gain that ability?

    I am an engineer. I actually have the JOB of tranforming one machine into another (I change information processing machinery all day long.) Therefore, I know QUANTATATIVELY, even on a gut level, what kinds and quantity of changes must occur to change machine A into machine B.

    I think this is the root of it.

  30. p.noyola,

    Make the case for ID, and against RV + NS in these simple steps:

    1) Complex, integrated, functional machinery exists.
    2) Two causal classes (A:intelligence and B: non-intelligent forces of nature and chance conditions) are logically possible causes for 1.
    3) Non-intelligent causes have NOT YET been empirically demonstrated to produce complex, integrated, functional machinery. This has only been assumed (see just-so stories.)
    4) Intelligent causes HAVE BEEN (and can be) empirically demonstrated to cause such machines. (Computers, Cars, aircraft, etc).
    5) Therefore, intelligence is currently the best and only explanation for 1. Until another causal class can be empirically demonstrated as a viable cause, ID is the default position.

    It really is that simple.

  31. Atom, unfortunately it really isn’t that simple, in that cogent arguments such as yours make absolutely no headway against closed minds. It gets back to the profound effects of cultural conditioning.

  32. also, atom, your argument works for explaining computers cars and aircraft, but we already knew about them. There’s nothing in your argument that really explains why intelligence is required for the systems we’re trying to explain here.
    And ari, I wasn’t referring to calculus in regards to multiverses, but rather to DLH in #10

  33. Hi Gil, there’s thirty-some-odd comments here, but none yet from the great Dr. Moran. Based upon comments in the previous thread, I wonder if he is being blocked.

    I would love to see him get involved in the discussion, put out his evidence, and see how it flies. I bet that if he engages seriously he will soon begin to sound awfully like an IDer as Dr. MacNeill before him has done.

  34. [A]tom, your argument works for explaining computers cars and aircraft, but we already knew about them.

    Exactly why I used the words “empirical” and “demonstrated”. We can show intelligence in action.

    There’s nothing in your argument that really explains why intelligence is required for the systems we’re trying to explain here.

    I didn’t say it was required (though I think it is); no, my argument is simply that intelligence is capable. No other causes have been shown capable.

    The type of machine is important, not the individual instantiation of that type (car, aircraft, computer, etc.)

  35. “type” being complex, integrated, functional machines, btw.

  36. I don’t think you have shown it to be capable of that either, except that we assume it to be so.
    i.e. How have you shown intelligence to be capable of designing these things that we are looking at?

  37. I don’t think you have shown it to be capable of that either, except that we assume it to be so.
    i.e. How have you shown intelligence to be capable of designing these things that we are looking at?

    Are you serious? So intelligence (human intelligence in particular) is not responsible for the computer you type on?

    You must be joking.

  38. I’ll make it even simpler for you Corey:

    Machine Type A = Complex, integrated, functional machine.

    1) Computers are an example of Machine Type A.
    2) Human intelligence can create computers.
    3) Therefore, human intelligence can demonstrably create Machine Type A.
    4) Human intelligence is a specific type of intelligence.
    5) Therefore, intelligence as a causal class (which includes human intelligence) is capable of creating Machine Type A.
    6) Furthermore, no other causal class has been demonstrated capable of producing Machine Type A. Until there is an alternative, empirically demonstrable causal class, ID is the default rational position concerning the origin of any instances Machine Type A.

    Very easy to follow.

  39. Great thread, Gil. I, like many here, have to design and build things to work in the real world. My experience tells me that unguided natural forces will destroy something easily enough, but they never create, and to suggest that all life is the result of these natural forces is, to me, insane. I was watching the Science Channel recently and they were explaining how birds, lobsters, and bacteria navigate their environment; they all have internal compasses. In fact, bacteria will manufacture particles of magnetite within themselves and then align them to form a compass (incredible enough). The alignment sort of looks like a backbone, so of course, here comes the darwinist, explaining how me “may” owe our backbones to these bacteria that make compasses. And the narrator instantly starts speaking about how our backbones come from these bacteria. It’s stuff like this which makes me want to scream. A few particles in a bacteria is *nothing* like a human skeletal system, but in darwin-land, just come up with a nice story, and bang it’s fact. Here’s a fun project: Go take that bacteria’s string of particles, and then from there, provide a list of all the changes which need to occur to go from that to a human skeletal system. When you’ve given up, realize that this experiment was a purposeful, goal-oriented engineering project, and it still couldn’t be done. Time and chance?? No $!#*& way!

  40. I meant life, not things we’ve built, which I already acknowledged.
    Or are you joking?

  41. shaner74: My experience tells me that unguided natural forces will destroy something easily enough
    -

    including natural selection and genetic drift…the very forces that Larry depends on. Larry will have to look for some other kind of Maxwell’s Demon.

    Atom wrote: It can be seen in the TV show Heroes (which I love) in that the “ability” to levitate is the result of random mutations. Really, it takes that little special machinery to gain that ability?
    -
    Evolution is mentioned in almost every episode. Hello? They have supernatural powers!

    It must be a “new age” thing.

  42. Corey, again I already laid this out.

    Life is of Machine Type A (Complex, integrated, functional machines.) Intelligence is the only demonstrated cause capable of producing Machine Type A.

    And as I wrote in the previous comment:

    The type of machine is important, not the individual instantiation of that type (car, aircraft, computer, etc.)

    I’d add life to that list. Unless you want to argue that life is not 1) Complex, 2) Integrated, or 3) functional?

  43. DLH you stated:

    Right off, there appears to basis for avoiding mutational meltdown in large asexual populations vs small. There may be a difference in fixation, but that does not lessen the total mutational load. Worth examining Lynch’s arguments in detail to identify the fallacy is.

    Dr. Sanford goes over this in his book Genetic Entropy.

    He says they get around this problem by ignoring or greatly underestimating “slightly deleterious mutations”. They do this in most studies by declaring them completely neutral so as to avoid the impossibilities soon faced by accumulating deleterious mutations in the entire population.

    It truly is a severe and real problem that evolutionists just hand wave away.

    If you want I will dig out Sanford’s book and quote the exact the slight of hand they use to get by this crushing problem.

    As well, the long term study (270 million years), of all the orders of trilobites, by Webster, confirms the ID/Genetic Entropy , although the genetic meltdown (loss of variation within species and within order) takes far longer than would be expected from Sanfords’ estimates.

  44. p.noyola: rub his nose in his own claims. He is believing in ‘science of the gaps’. He is believing lots of stuff on FAITH, not EVIDENCE. Is that reasonable given his empirical presuppositions? There is no evidence for other universes; it is even possible to test the theory?

    It is really easy to point you at talk origins, but remember Logic 101: he who makes the claim bears the burden. Don’t let him get away with making a claim without HIM providing the evidence.

  45. Corey said:

    How have you shown intelligence to be capable of designing these things that we are looking at?

    An equivalent question: And how has anybody shown that the intelligence that’s looking at it knows what its looking at?

    Of course, this is a fundamental skepticism which underlies my skepticism of materialism and ID. Not just a special purpose skepticism, which is only used as a bait and switch for a replacement certainty.

    If we cannot grasp it, then we have to accept that at some point, regardless of intent, Science is stopped, which makes all this noise about who’s committing the “Science-stoppers” moot. We’re not going to get our evolution toasters, and evolution cars and evolution computers. Believing that there was an evolutionary pathway or not is not going to bring that new Evolution technology, or people being cured by “evolving out of their diseases” — all this technology, which is the hallmark of “emerging from a dark age” (which historians don’t any longer know existed–so how come biologists are so sure?) would have to be put on hold.

    It is well within the “bounds” of this un-stop-able Science to know how to design life even better than the existing complex systems that happened.

  46. One challenge is that greater advantages of sex in smaller populations does not explain anything about how sexual differentiation originated.

    True. For me, that’s Somebody Else’s Problem – it’s a historical question and I doubt we’ll find many transitional forms. :-)

    The larger challenge is how to explain origin and survival of more complex biological systems with slow reproduction rates such as man , elephants and whales.

    I don’t see that as a problem. The niches these organisms occupy require a crtain level of complexity, and size is presumably advantageous. The reduction in reproductive rates is perhaps an inevitable consequence of this. I could go on about r and K selection but that’s so seventies.

    The efects of population size on viabilitgy are well studied, and the sizes have to befairly small (there’s a lot of work on this in conservation genetics). The problems mainly come about due to inbreeding depression, which is how mutational load is expressed.

    Bob

  47. By contrast, the appearance is that most evolutionists (like Richard Dawkins) rarely use population genetics models possibly because from taking biology, zoology etc and avoiding calculus.

    Or that most don’t do population genetics. Those that do have to learn the maths. It’s also a very vibrant areas mathematically at the moment, thanks to bioinformatics. There are some good mathematicians involved. Well, there always have been, even since before Fisher.

    I would agree that many biologists avoid the maths, but for me one of the appeals of population genetics (and why I work in the field) is that it has such a strong mathematical and statistical foundation.

    Bob

  48. Atom

    Thanks for the “simple steps” case for ID. Excellent. Re:

    Life is of Machine Type A (Complex, integrated, functional machines.) Intelligence is the only demonstrated cause capable of producing Machine Type A.

    Once, in the course of “witnessing” to a microbiologist who claimed to be an atheist, I asked him: With all the complexity of life — which has taken the collective efforts of the best brains science can produce over several generations just to begin to understand — how can you say it all arose by accident?

    Actually, life remains a subcategory of Machine Type A, which to date, requires an intelligence more advanced than our own merely to replicate (forget about originate).

  49. Bob O’H at 46, 47
    Good to have an population genetics practioner responding.

    I was popularly describing Haldane’s Dilemma as extended by Walter Ramine.

    ReMine, W. J., 2005, Cost Theory and the Cost of Substitution — a clarification. TJ 19(1), 2005, pp. 113-125,

    i.e., there are a finite small number of mutations that can become fixed with small populations and long reproductive cycles. e.g., Haldane’s 300 to Ramine’s 1667 changes between man and chimp. These are may orders of magnitude smaller than the base pair differences between the species. Just between man and chimp there are nominally over 150 million changes. e.g. the 5% difference shown by Britten 2002.
    Majority difference between closely related DNA samples is due to indels Roy J. Britten et al. PNAS April 15, 2003 Vol 100 Nr 8, 4661-4665

    Michael Behe’s Edge of Evolution discussion shows the limits of multiple changes with very large numbers in Malaria or HIV.

    This is compounded when you need much larger number of changes with complex organisms, yet have much fewer number of generations to achieve them.

    So I see evolutionists trying to bridge gaps many orders of magnitude wide with hand waving “evolution of the gaps” arguments.

  50. “‘But frequently, the pure mathematicians invent/discover new math, and later it is found that it “works” in the real world.’

    Is this a source of dissapointment for them ? I did hear comments to that effect from someone”

    I do remember hearing a story like this. They used to say that number theory was the purest of the maths because there were no applications of it. But then they found out that it could be used for coding, etc. That upset some people, IIRC.

    My own field is topology and some recent work has been done on practical applications, but it’s never bothered me. But me and other graduate students at my school like to say that people going into applied math have went to the dark side. But we’re just kidding…mostly.

  51. i.e., there are a finite small number of mutations that can become fixed with small populations and long reproductive cycles.

    Right, but only in a worsening environment – see for example Leonard Nunney’s paper in Annales Zooloca Fennici (links from here). He also talked about this at ESEB this summer, and made the link to work on the Red Queen Hypothesis.

    Bob

  52. Aagh! Zooloca -> Zoologici

    Bob

  53. The majority of degreed computer scientists, engineers, and mathematicians have completed no college course work in the life sciences. Virtually all have college physics under their belts. Some studied chemistry in college. Relatively few enrolled in college courses in biology.

    Among “expert” critics of scholarly fields not their own, at most one in a thousand makes a substantive contribution. If UD should happen to be chock-full of engineers, computer scientists, and mathematicians who have all caught life scientists in fundamental error, then it would constitute a singular event in the history of science.

    Most of us who have actually contributed to the bodies of knowledge in our own fields know that the most reliable sources of information in other fields are the people devoted to investigation of those fields. Scientists are sometimes in error, but as consumers of scientific knowledge we have no choice but to play the odds. The odds of getting useful explanations of the diversity of life forms on earth are much greater when one goes to people who have studied the matter intensively and have worked their way to consensus than when one goes to people who have little or no higher education on the matter.

    Maverick geniuses are very rare, and if you ever seem to be surrounded by them, you can count on it that appearances are deceiving.

  54. oh now you’re talking about “the odds.”

  55. As a computer science student, I recoil in horor at the though of injecting random code into my working system unless my working system was designed to handle random injections.

    Hmm. So C-language code inserted after return and break statements is problematic? How often do you suppose a randomly-generated condition for an if statement would be satisfied in practice? There are more ways to introduce junk code into programs than are apparent to the casual observer.

    Look into genetic programming, and you’ll find that much of the “injected” code serves no function. Many researchers in the field consider bloat a problem. Google “genetic programming” AND bloat, and you get about ten thousand hits.

    Of course, IDists don’t care much for the claim that much of the genetic code of biota is nonfunctional. But it is commonly the case that much of the code in genetic programs is nonfunctional.

  56. Semiotic 007,

    Life is a subset of machines in general, specifically information processing machines. Engineers design, build, test, debug, and thoroughly understand machines in general and information processors in particular.

    In the same way, I am a software engineer but I have never looked at Microsoft source code. If someone from Microsoft were to say “We designed code that tests for for every possible error, including infinite loops within the program” I’d know they were wrong. Why? Even though I haven’t studied their particular instantiation of code (I’m not an “expert” in Microsoft Code), I am quite familiar with code in general and the underlying concepts. So my general knowledge of code also applies to their Code.

    In the same way, all machines, whether silicon based or carbon based, are subject to the same laws of physics and information processing. We are talking about the limits and flow of information in general; Carbon-based Life is just a particular example we can apply these principles to.

  57. And BTW, I have done university coursework in the life sciences.

  58. All scientific beliefs are tenuous. Anyone who does not want to play the odds in belief formation should go somewhere other than science. I personally recommend religion and philosophy.

  59. semiotic: bloat has a big cost. It takes up energy and resources to do nothing. Maybe not a problem in the digital world but in the real world I would expect natural selection to eliminate the bloat. Why didn’t it? Either the code really is functional and the organism is more more complex OR natural selection doesn’t work very well.

  60. Semiotic 007, I’m going to reword your post a bit to demonstrate that while the structure of it is well thought out, the assumptions behind it merely beg the question:

    “The majority of degreed biologists have completed no college course work in the engineering sciences. Virtually all have college physics under their belts. Some studied chemistry in college. Relatively few enrolled in college courses in engineering or systems design.

    Among “expert” critics of scholarly fields not their own, at most one in a thousand makes a substantive contribution. If the biological sciences should happen to be chock-full of biologists who have all caught engineers, computer scientists, and mathematicians in fundamental error, then it would constitute a singular event in the history of science.

    Most of us who have actually contributed to the bodies of knowledge in our own fields know that the most reliable sources of information in other fields are the people devoted to investigation of those fields. Scientists are sometimes in error, but as consumers of scientific knowledge we have no choice but to play the odds. The odds of getting useful explanations of the design of life forms on earth are much greater when one goes to people who have studied the matter intensively and have worked their way to consensus than when one goes to people who have little or no higher education on the matter.

    Maverick geniuses are very rare, and if you ever seem to be surrounded by them, you can count on it that appearances are deceiving.”

    Bottom line, the assumptions you make are entirely upside down. It is the biologists who have no special expertise when it comes to evaluating design inferences. They have no special standing to tell those who do have expertise in fields directly impinging on design, probability, and information theory to simply sit down, shut up, and listen to what the non-experts have to say. IMO, the non-expertise of evolutionary biologists in any of these crucially pertinent fields becomes more and more annoyingly manifest with each passing year.

  61. Who was Darwin? All he had was a theology degree. He made evolution into a new religion. Biology is not a science because of this but that will change once Darwin is taken out and engineers are put back in.

  62. @Matteo 60:

    Well put.

  63. Atom,

    “If the only tool you have is a hammer, then every problem’s going to look like a nail.” A senior professor in engineering pointed that out to me many years ago, back when I was a green assistant professor, and I took it to heart.

    Many people take the “genetic program” metaphor literally. Computing professionals seem particularly susceptible to the error. It is revealing that they usually speak in terms of programs for general-purpose computers, which they feel they know and understand, rather than for robots with custom hardware and firmware. Certainly the analogy to a program for a robot is much stronger than the analogy to a program for a general-purpose computer, yet it eludes most who want to speak in terms of programs.

    The “genetic program” does not come close to controlling the state of the “machine” as the program of an electronic digital computer does. And it appears that in morphogenesis of multicellular organisms a great deal of the information comes from the environment, rather than from the genome. Metaphors can be useful for those who remember they are metaphors and who understand their limitations, but can be highly misleading for those who take them literally.

  64. following what I said in 59. Any mutation that increases complexity (I’m not using the term CSI here), even if functional has an energy cost to support the extra complexity. This would make evolution very hard to pull off. Natural selection will only favor mutations that simplify the system.

    Now in a virus, you don’t have this problem so much because it just leaches off all the hard work of a living organism. But a living thing needs to have the ‘capital’ to make the ‘investment.’

  65. Semiotic,

    I used computer code simply for discussion of my microsoft example. From what I understand, gene regulation networks function more like electrical engineering logic circuits. But this is irrelevant.

    Again, we’re not talking about specific instantiations of machines, but rather information processing and information flow principles in general. These apply to Turing Machines, logic networks, carbon-based machinery, wood-based machinery, “silly putty”-based machinery, or any other type of machinery you’d like to create within this universe.

    A simple hammer comes in handy when you are dealing with different types of the general “nail” class.

    (Or perhaps you’d like to argue that carbon based machinery is somehow special and exempt from the general laws?)

  66. Semiotic 007 (#55): “Hmm. So C-language code inserted after return and break statements is problematic? How often do you suppose a randomly-generated condition for an if statement would be satisfied in practice? There are more ways to introduce junk code into programs than are apparent to the casual observer.”

    So how much of the code is functional as opposed to nonfunctional? What if the injected error creates a jump statement into a nonfunctional area after return or break statements (ignoring issues of the particular language)? The fact remains that there is a high likelihood that a random change inserted into a random point in a software program will cause a problem and degrade the software machine (if not cause catastrophic failure and stop it or send it into an endless loop).

    “Of course, IDists don’t care much for the claim that much of the genetic code of biota is nonfunctional. But it is commonly the case that much of the code in genetic programs is nonfunctional.”

    You are ignoring the fact that recent research has found that much or most of the genome is indeed functional and that there is much less “junk DNA” than previously supposed by Darwinists. It is turning out to be an incredibly complex interwoven network with many levels of coding. This has been discussed in numerous other threads on UD, and is directly contradictory to the predictions of Darwinists.

  67. Matteo,

    In my opinion, no one is expert in design inference. There has certainly been some work in the past 10-15 years (a shorter period than Darwin spent coming by his theory of evolution) to develop a theory of design inference, but there is not yet a stable body of knowledge.

    Dr. Dembski’s last paper on design detection through measurement of complex specified information is a far cry from what he presented in The Design Inference. His recent collaborative work with Bob Marks does not mention CSI, but is developed in terms of endogenous, exogenous, and active information. While an IDist would see active information as analogous to CSI, its definition is radically different. It appears that Dr. Dembski has left CSI behind.

    Recall that Prof. Behe indicated in Darwin’s Black Box that his “evolutionary” definition of irreducible complexity was stronger than the original. That is quite significant, because irreducible complexity is a matter of degree in the revised definition, and is all-or-nothing in the original. Dr. Dembski has since written a long paper offering his own revised definition, and it differs substantially from Behe’s.

    ID theory has shifted radically and rapidly in its formulation of design inference. I cannot see that you have a basis for claiming that engineers, computer scientists, and mathematicians have expertise in design inference that other people do not. Again, there are investigators, but there are as yet no experts. Best wishes and good luck to Dembski, Behe, and others in their research.

  68. Ari,

    I hate engaging in seat-of-the-pants “reasoning” about topics I know little about. Do you know how much of the resource requirement for cellular reproduction is accounted for by DNA replication? My guess is that it is little. It’s not obvious to me that there is significant selective pressure for a small genome. Replication errors, on the other hand, are a thermodynamic necessity.

  69. You are ignoring the fact that recent research has found that much or most of the genome is indeed functional and that there is much less “junk DNA”

    I made no claim whatsoever about functionality in the genomes of biota, magnan. And you have not addressed the fact that most genetic programming systems rely heavily on insertion of random code. Casual observations about probability of success have little value when there are empirical data to go on.

  70. well if that’s the case you can end up with bloat but you won’t get evolution.

    In real life, selective pressures are low, the possibility of a mutation that adds functionality is very low, the extra functionality would be very small and it still has to justify the cost of maintaining that extra complexity -it is small but not when compared to the extra benefit.

    On the other hand, a mutation that removes complexity (without hurting it) could be a big energy win and natural selection would favor that. That’s why it is hard for me to see how evolution could work.

  71. Bob O’H et al.
    To correct a post see if you can click the “e” underneath your name and date on the left of the post.

  72. Bob O’H at 51
    Thanks for the reference to:
    Nunney, L. 2003: The cost of natural selection revisited. — Ann. Zool. Fennici 40: 185–194.
    (free screen quality pdf)

    I understand the “cost” in generations per fixed mutation is dominated by the beneficial mutation rate M.

    In “Genetic Entropy and the Mystery of the Genome”, John Sanford summarizes reported beneficial to harmful mutation ratios of 1 to 10,000 or 1 to 1 million. It appears Nunney was using very high (optimistic) beneficial mutation rates. I don’t see how his study makes any serious change to the strong limitations of Haldane’s Dilemma. Nunney may relax it by one or two orders of magnitude – IF the high beneficial mutation rate is justified.

    The 150 million changes between human and chimp is still many orders of magnitude greater than the maximum available time.

    PS Walter Remine reports:

    Evolutionary geneticist, Leonard Nunney, reported on his computer simulation that evolves dramatically faster than the Haldane limit. Walter ReMine contacted Nunney and requested a copy of Nunney’s software for detailed study of its results and methods. Nunney declined, saying he would not share his software with “people who do not publish in peer-reviewed journals” — which is evolutionist-code for ‘anti-evolutionists’. Given evolutionist reluctance to have their results verified, Nunney’s simulation must be viewed dimly, especially since it contradicts other simulations.

  73. Semiotic 007 at 53

    Your argument fails to consider the lemming effect in science and paradigm shifts after one brilliant person shows the way. e.g., the Aristotelian academicians were collectively opposed to Galileo and did all they could to stop him. Yet Galileo showed the evidence and Aristotle no longer reigns.

    See Mathematics of Evolution at ResearchIntelligentDesign.org

    Sir Fred Hoyle: spent a major part of his life digging into the mathematics of evolution. He observed:

    If one proceeds directly and straightforwardly in this matter, without being deflected by a fear of incurring the wrath of scientific opinion, one arrives at the conclusion that biomaterials with their amazing measure or order must be the outcome of intelligent design. No other possibility I have been able to think of…[7]

    “Published in his 1982/1984 books Evolution from Space (co-authored with Chandra Wickramasinghe), Hoyle calculated that the chance of obtaining the required set of enzymes for even the simplest living cell was one in 10 ^ 40,000. Since the number of atoms in the known universe is infinitesimally tiny by comparison (10 ^ 80), he argued that even a whole universe full of primordial soup would grant little chance to evolutionary processes.”

    Hoyle further refined his observations in Mathematics of Evolution, building on published population genetics works.

    Yet who today dares to take Hoyle’s mathematics and observations in the face of the wrath of evolutionists in power?

    See especially:
    Fred Hoyle, Mathematics of Evolution, (1987) University College Cardiff Press, (1999) Acorn Enterprises LLC., ISBN 0-9669934-0-3

    Hubert P. Yockey, Information Theory, Evolution, and the Origin of Life 2005 Cambridge University Press ISBN 13 978-0-521-80293-2

    John C. Sanford Genetic Entropy & the Mystery of the Genome. (2005) Ivan Press. ISBN 1599190028.

  74. DLH – thanks for the tip on the e, but I can’t see one. :-( Perhaps it’ll turn up on this message.

    I cited Nunney to clarify the point that Haldane’s dilemma only works in a deteriorating environment. It’s something that isn’t obvious from Haldane’s paper. If most environments are stable (at least with regards to the fitness surface), the dilemma goes away because beneficial mutations increase fitness – in Haldanes model (IIRC) the fitness gets re-set when an allele becomes fixed. That’s an odd feature of the model, but is best interpreted as being equivalent to a decline in the environment.

    I’m well aware of Remine’s complaints about Nunney. Whilst I think Nunney could have been politer, I don’t think Remine has a case. The model is adequately described in the paper, so Remine could code it himself. It should be fairly easy in a high-level maths language.

    Bob

  75. 75

    In addition to asking why mathematicians, computer scientists, and engineers tend to be more skeptical of Darwinian claims, IMO we should also be asking why biologists tend to be less skeptical of Darwinian claims. The lower skepticism of the biologists makes the skepticism of the mathematicians, computer scientists, and engineers seem great by comparison.

    Biologists have an inferiority complex because of the kind of attitude expressed by Lord Rutherford: “All science is either physics or stamp-collecting.” Because of this inferiority complex, biologists have been boasting with a religious-like zeal that biology has something that other branches of science do not have, a grand central overarching supreme unifying “theory of everything,” Darwinism.

  76. 76

    Also, I think that scientists in non-biological sciences — e.g., inorganic chemistry, physics (Lord Rutherford’s “stamp collecting” statement notwithstanding), geology, and astronomy, and even the social sciences, e.g., anthropology, sociology, and psychology — may feel that by defending Darwinism they are defending science in general. Pure scientific research often has no immediate economic benefit and scientists may feel that research funding may be reduced if the attacks on Darwinism cause the public to lose faith in science. On the other hand, the work of engineers and so-called computer “scientists” often has immediate economic benefit and so they may not feel threatened in this way by attacks on Darwinism.

  77. The Machine
    v. 1.1*

    Two people are looking at a finely-tuned, highly sophistocated machine that produces massive amounts of complex, functioning items.

    1: “It’s remarkable what purposelessness can do.”

    2: “You think this machine came about and operates without purpose?”

    1: “Yup.”

    2: “That’s incredible. What makes you think that?”

    1: “Come on. Everyone knows that everything in life is purposeless.”

    2: “Why do you say that? All machines I know of have been designed and built for a purpose, by intelligent beings.”

    1: “Yeah, but the intelligent beings came about through a process that’s purposeless.”

    2: “How do you know?”

    1: “Come on. Everyone knows that.”

    2: “Let me take another approach. Can you show me a working model of a purposeless machine that generates any type of functioning thing?”

    1: “Here. Here they are.” [Pointing to Avida, etc.]

    2: “But these all have active information and an overarching purpose built-in! How can you claim that these are models for this machine?!”

    1: “Well, for a living, I study the output and the few visible portions of this machine. All the books and all the important people in the field say that the machine is entirely purposeless. So there. Who are you to question that?”

    2: “I design and build machines every day. When we have an idea for a machine, before committing large amounts of money to its development for production, we always need to prove that it will work by designing and building a working model. You want me to commit to the idea that this thing operates completely purposelessly, yet you haven’t shown a proof-of-concept working model for your idea of how it works. Sorry, no can do.”

    * With a few minor but clarifying fixes in the wording of my little story at comment 5 above.

  78. Larry

    Good point. Science and engineering that produces practical benefits easily seen by laypersons is not really subject to threat when allocation of finite public funding is prioritized and distributed as compared to science and engineering that does not produce easily discernable practical benefit.

    One program that comes immediately to mind in this regard is the Superconducting Super Collider. Congress critters couldn’t see the cost/benefit of the big atom smasher as superior to that of the International Space Station, couldn’t afford to fund both, so the SSC lost out.

    Historic biology is of far less practical import than high energy physics. The public would refuse to grant much funding to historic biology if they realized how useless it is.

    I think you’re quite right that this is a major factor in why prehistoric evolutionary biology is so vociferously defended by those with a vested interest in it or similarly useless intellectual pursuits.

  79. DaveScot said,

    I think you’re quite right that this is a major factor in why prehistoric evolutionary biology is so vociferously defended by those with a vested interest in it or similarly useless intellectual pursuits.

    I wouldn’t call paleontology — which to some people is synonymous or nearly synonymous with “prehistoric evolutionary biology” — a “useless intellectual pursuit.” Knowledge is good for its own sake. However, the Darwinists have been claiming that the US is going to lose its international technological competitiveness if Darwinism is not taught dogmatically in schools. Paul R. Gross, co-author — with Barbara Forrest — of “Inside Creationism’s Trojan Horse” and lead author of the Fordham Institute (no connection to Fordham U.) report on state science standards, threatened to drop Ohio’s overall science grade from a B to an F because of Ohio’s evolution lesson plan, even though evolution counts for only 3 points out of 69 in the Fordham grading system — see

    http://im-from-missouri.blogsp.....is-co.html

  80. I cited Nunney to clarify the point that Haldane’s dilemma only works in a deteriorating environment. It’s something that isn’t obvious from Haldane’s paper. If most environments are stable (at least with regards to the fitness surface), the dilemma goes away because beneficial mutations increase fitness – in Haldanes model (IIRC) the fitness gets re-set when an allele becomes fixed. That’s an odd feature of the model, but is best interpreted as being equivalent to a decline in the environment.

    Bob,

    Let me know if I misunderstand your point. Are you saying that in Haldane’s model once an allele reaches fixation its fitness becomes 0 (ceases to confer a fitness benefit) and that Nunney’s model disagrees with this assumption?

    If not, then disregard the rest of this and please clarify.

    If so, I think Haldane’s assumption makes perfect sense, since a trait increases fitness over that of your competitors (other organisms in your population.) If everyone has the same allele (fixation), then there is no more fitness advantage.

    Please correct me if I misunderstood you.

  81. Bob O’H, still there?

  82. Atom – thanks for the heads-up on the other post (this one is too far away! :-)).

    Once an allele becomes fixed, yes there is no fitness benefit so the fitness becomes 1. And yes, this makes perfect sense.

    My somment about the fitness being re-set was wrong. The situation Haldane imagines is where a change in the environment reduced juvenile survival, and this is restored though selection. His argument boils down to saying that if the reduction is large enough, the growth rate goes below 1, so the population will decline, and may go extinct.

    Nunney’s model is almost exactly the same in this regard, except that he has a continuously changing environment (it’s also different in that it has better ecological dynamics, but that’s not critical for the general point).

    Bob

  83. My take on the original post:
    Mathemeticians, engineers, and computer scientists all work in fields dominated by the type of thinking that goes into a formal system. Mathematics is a full-fledged formal system, while the other two are highly rule-driven with an objective to create. They are essentially fields where you use deductive thinking, working from the top down.

    Science is the process of building from the bottom up, using inductive thinking. The prupose is not so much to create but to uncover.

    It’s very natural that the people who were attracted to fields that relied on deductive thinking in the first place are more likely to see the top-down version of reality, with an outside force providing guidance and rules, as a more acceptable vision of the world.

    I’m perfectly amenable to being convinced that the complexity, information content, and machinery of living systems can not be explained by stochastic processes filtered by natural selection, and I would not even demand hard evidence, just some rigorous argumentation based on the following:

    1) An analysis of the unlikelness of evolution that accounts for all 10+ known processes by which the genome accumulates variances, the 3 methods by which that variance is selected, and demonstrates true independence for each facrot before multiplying the factors’ probabilities together.
    2) A reason that scaffolding and similar methods would not be able to create irreducilbe complexity naturally.
    3) A way to measure CSI that is internal to the information contained in the string itself. For example, it should be able to determine with 95% accuracy whether a given four-letter code in a string of 3000 characters either comes from a coding section of DNA or was randomly generated.
    4) A precise limitation on the features that can or can not appear in a living organism over a trillion generations.
    5) A list of the precise interventions that were made in the history and the time frame in which they were made. Appeals to “some time in the past” are irrelevant.

  84. one brow said,

    My take on the original post:
    Mathemeticians, engineers, and computer scientists all work in fields dominated by the type of thinking that goes into a formal system. Mathematics is a full-fledged formal system, while the other two are highly rule-driven with an objective to create. They are essentially fields where you use deductive thinking, working from the top down.

    Yes, I think that one of the reasons why many of us went into those fields (I am a mechanical engineer) was that the laws and fixed rules in those fields gave us feelings of security, order, certainty, and predictability. That is not to say that those fields are not challenging — it is often a challenge to figure out ways to apply the laws and fixed rules.

  85. mathematicians are naturally skeptical of darwinism because mathematicians are inherently risk-averse when it comes to making statements without rigorous proof. the level of rigour in some evolutionary psychology speculations is apalling by mathematical standards. there is much better evidence for numerous ideas in mathematics which mathematicians continue to call conjectures due to absence of formal proof, then there is for some of the evolutionary psychology speculations.

  86. DLH makes an excellent point which has a name: entropy. Engineers have to deal with it in all their designs and even with our best, most carefully contrived efforts we can’t stop it. All we can do is slow it down.

    In the matter of life I totally understand extinctions. That’s fully explained by entropy. What I don’t understand is how some cell lines manage to slow it down enough that they’ve persisted for billions of years.

    In computer design we’ve developed highly protected core software that reset what evolution hath wrought back to a known good state. Typically this is implemented by a read-only memory chip that instructs the hardware to load a protected image of the factory-original software configuration. The factory original is, because of its size, usually supplied as a protected portion of a hard disk drive and/or on removable media such as a DVD disk.

    To insure even greater chance of catastrophe recovery we store redundant copies in different locations such that if a fire destroys one copy others will still be available.

    Since human design more often than not is eventually found to be an analog of something in the design of life my best guess as to how some cell lines manage to slow down entropy enough to survive for millions or even billions of years is by employing protected core code and triggers which serve to restore the organism to a known working state.

    This is an important question. I’ll write a separate article today to address it.

Leave a Reply