Artificial Intelligence Culture Darwinism Intelligent Design

Richard Lenski and Avida: But wait! Can software evolve?

Spread the love

The Avida program was supposed to replicate Darwinian evolution in software. Not only was there hope of proving Darwinism at last but it might lead to self-evolving computer programs.

Richard. W. Stevens is skeptical. He thinks the claims for Avida fail the most basic test. He wrote a simple program, InforMutation,to demonstrate that:

mindmatters.ai/2021/03/why-software-cannot-just-evolve-a-demonstration/(opens in a new tab)

The simulation demonstrated a reality for any kind of computer programming: The program code must follow the syntax and semantic rules of the programming language. If the program code is modified randomly, there is no reason to think the resulting program will still follow those rules.

When a program fails because of a syntax error, as in this example, the program does not run again — ever. A program that fails because of a semantic error usually produces spurious results and dies just as dead.

Neo-Darwinian evolution theory holds that an animal undergoes an undirected mutation and survives to reproduce, with its descendants undergoing other undirected mutations over thousands of reproduction cycles. Given enough such successive mutations, a new feature or function “evolves.” The Avida simulation applies this approach to programming.

As InforMutation shows, however, one small mutation in a software program is very often fatal to the software…

Working with InforMutation exposes the reasons why Avida could not demonstrate neo-Darwinian evolution of “learning” software. Rather, as a product of careful design, the Avida simulation showed the the fundamental elements of intelligent design: Purpose, Plan, Engineering, and Foresight. Only with these elements could randomly-mutated programs survive and run repeatedly so as to accumulate mutations.

Richard W. Stevens, “Why Software Cannot Just Evolve” at Mind Matters News

In other words, to the extent that Avida works, it was designed to work. And the programmer subtly — perhaps unaware — builds in rescues.

We’ve heard a phrase for that: intelligent design

You may also wish to read: Random evolution doesn’t produce algorithmic functions in animals (Richard W. Stevens)

9 Replies to “Richard Lenski and Avida: But wait! Can software evolve?

  1. 1
    Bob O'H says:

    As InforMutation shows, however, one small mutation in a software program is very often fatal to the software…

    And exactly the same thing happens in biology. So if Avida doesn’t evolve because it produces lethal mutants, so what?

    (Also – the link at the end of the quote needs fixing)

  2. 2
    ET says:

    AVIDA doesn’t do anything when realistic parameters are used.

  3. 3
    polistra says:

    Good article. There’s nothing new about automatic adjustment of variables from random permutations. Common technique. I’ve used it.

    Automatically adjusting the code itself is a bit more complex, but still possible if you’re using a high-level language and operating system that won’t block such moves (which look like hacking.)

    If a program WITHOUT any such adjusters decided on its own to build a code-adjuster inside itself, then we might have something interesting. But that’s not what Stevens is describing here.

  4. 4
    News says:

    Bob O’H at 1, thanks, link fixed.

  5. 5
    bornagain77 says:

    Avida, and all the other evolutionary algorithms, are a metaphor for Darwinian ‘science’ writ large.

    Like the supposed ‘science’ presented to the public by Darwinists to try to convince the public that Darwinian evolution is true, the only way these programmers are able to fool themselves, and others, into believing they have evolved anything in their programs is by looking at, and presenting to others, only a partial picture of what is really going on in their evolutionary algorithms.

    In fact, the only way these computer programs are seemingly able to ‘evolve’ anything is by cheating, i.e. by smuggling preexisting information into program in order to find a pre-specified goal.

    As Robert Marks explained, “Those hoping to establish Darwinian evolution as a hard science with a model have either failed or inadvertently cheated. These models contain guidance mechanisms to land the airplane squarely on the target runway despite stochastic wind gusts. Not only can the guiding assistance be specifically identified in each proposed evolution model, its contribution to the success can be measured, in bits, as active information.,,,
    Models of Darwinian evolution, Avida and EV included, are searches with a fixed goal. For EV, the goal is finding specified nucleotide binding sites. Avida’s goal is to generate an EQU logic function. Other evolution models that we examine in Introduction to Evolutionary Informatics likewise seek a prespecified goal.,,,

    Top Ten Questions and Objections to ‘Introduction to Evolutionary Informatics’ – Robert J. Marks II – June 12, 2017
    Excerpt: There exists no (computer) model successfully describing undirected Darwinian evolution. Period. By “model,” we mean definitive simulations or foundational mathematics required of a hard science.,,,
    We show that no meaningful information can arise from an evolutionary process unless that process is guided. Even when guided, the degree of evolution’s accomplishment is limited by the expertise of the guiding information source — a limit we call Basener’s ceiling. An evolutionary program whose goal is to master chess will never evolve further and offer investment advice.,,,
    There exists no model successfully describing undirected Darwinian evolution. Hard sciences are built on foundations of mathematics or definitive simulations. Examples include electromagnetics, Newtonian mechanics, geophysics, relativity, thermodynamics, quantum mechanics, optics, and many areas in biology. Those hoping to establish Darwinian evolution as a hard science with a model have either failed or inadvertently cheated. These models contain guidance mechanisms to land the airplane squarely on the target runway despite stochastic wind gusts. Not only can the guiding assistance be specifically identified in each proposed evolution model, its contribution to the success can be measured, in bits, as active information.,,,
    Models of Darwinian evolution, Avida and EV included, are searches with a fixed goal. For EV, the goal is finding specified nucleotide binding sites. Avida’s goal is to generate an EQU logic function. Other evolution models that we examine in Introduction to Evolutionary Informatics likewise seek a prespecified goal.,,,

    The most celebrated attempt of an evolution model without a goal of which we’re aware is TIERRA. In an attempt to recreate something like the Cambrian explosion on a computer, the programmer created what was thought to be an information-rich environment where digital organisms would flourish and evolve. According to TIERRA’s ingenious creator, Thomas Ray, the project failed and was abandoned. There has to date been no success in open-ended evolution in the field of artificial life.5,,,
    We show that the probability resources of the universe and even string theory’s hypothetical multiverse are insufficient to explain the specified complexity surrounding us.,,,
    If a successful search requires equaling or exceeding some degree of active information, what is the chance of finding any search with as good or better performance? We call this a search-for-the-search. In Introduction to Evolutionary Informatics, we show that the search-for-the-search is exponentially more difficult than the search itself!,,,
    ,,,we use information theory to measure meaningful information and show there exists no model successfully describing undirected Darwinian evolution.,,,
    ,,, if the fitness continues to change, it is argued, the evolved entity can achieve greater and greater specified complexity,,,
    ,,, We,, dub the overall search structure ‘stair step active information’. Not only is guidance required on each stair, but the next step must be carefully chosen to guide the process to the higher fitness landscape and therefore ever increasing complexity.,,,
    Such fine tuning is the case of any fortuitous shift in fitness landscapes and increases, not decreases, the difficulty of evolution of ever-increasing specified complexity. It supports the case there exists no model successfully describing undirected Darwinian evolution.,,,
    Turing’s landmark work has allowed researchers, most notably Roger Penrose,26 to make the case that certain of man’s attributes including creativity and understanding are beyond the capability of the computer.,,,
    ,,, there exists no model successfully describing undirected Darwinian evolution. According to our current understanding, there never will be.,,,
    https://evolutionnews.org/2017/06/top-ten-questions-and-objections-to-introduction-to-evolutionary-informatics/

    Moreover, the programming, (and engineering) in life is orders of magnitude more sophisticated compared to the best programming in our man-made computers.

    As I heard one commenter on UD mention, the engineering in life “is engineering sci-fi” as far as man is concerned.

    For example, “the information density in the nucleus is trillions of times higher than on a computer chip”,,,

    3-D Structure Of Human Genome: Fractal Globule Architecture Packs Two Meters Of DNA Into Each Cell – Oct. 2009
    Excerpt: the information density in the nucleus is trillions of times higher than on a computer chip — while avoiding the knots and tangles that might interfere with the cell’s ability to read its own genome. Moreover, the DNA can easily unfold and refold during gene activation, gene repression, and cell replication.
    http://www.sciencedaily.com/re.....142957.htm

    In fact, researchers at Harvard, “have successfully stored 5.5 petabits of data — around 700 terabytes — in a single gram of DNA, smashing the previous DNA data density record by a thousand times.,,, Just think about it for a moment: One gram of DNA can store 700 terabytes of data. That’s 14,000 50-gigabyte Blu-ray discs… in a droplet of DNA that would fit on the tip of your pinky. To store the same kind of data on hard drives — the densest storage medium in use today — you’d need 233 3TB drives, weighing a total of 151 kilos.”

    Harvard cracks DNA storage, crams 700 terabytes of data into a single gram – Sebastian Anthony – August 17, 2012
    Excerpt: A bioengineer and geneticist at Harvard’s Wyss Institute have successfully stored 5.5 petabits of data — around 700 terabytes — in a single gram of DNA, smashing the previous DNA data density record by a thousand times.,,, Just think about it for a moment: One gram of DNA can store 700 terabytes of data. That’s 14,000 50-gigabyte Blu-ray discs… in a droplet of DNA that would fit on the tip of your pinky. To store the same kind of data on hard drives — the densest storage medium in use today — you’d need 233 3TB drives, weighing a total of 151 kilos. In Church and Kosuri’s case, they have successfully stored around 700 kilobytes of data in DNA — Church’s latest book, in fact — and proceeded to make 70 billion copies (which they claim, jokingly, makes it the best-selling book of all time!) totaling 44 petabytes of data stored.
    http://www.extremetech.com/ext.....ingle-gram

    Our best engineers and computer programers can only look at life and drool, hoping someday to be able imitate some of the ‘over the top’ engineering that they see in life.

    The World’s Ideal Storage Medium Is “Beyond Silicon” – January 20, 2017
    Excerpt: it’s easy to see why DNA is “one of the strongest candidates yet” to replace silicon as the storage medium of the future. The read-write speed is about 30 times faster than your computer’s hard drive. The expected data retention is 10 times longer. The power usage is ridiculously low, almost a billion times less than flash memory. And the data density is an astonishing 10^19 bits per cubic centimeter, a thousand times more than flash memory and a million times more than a hard disk. At that density, the entire world’s data could fit in one kilogram of DNA,,,
    https://evolutionnews.org/2017/01/the_worlds_idea/

    Here is another example of the ‘over the top’ engineering in life,,, one repair mechanism that was discovered in life has been compared to “spotting potholes on every street all over the country and getting them fixed before the next rush hour.”

    Quantum Dots Spotlight DNA-Repair Proteins in Motion – March 2010 ?Excerpt: “How this system works is an important unanswered question in this field,” he said. “It has to be able to identify very small mistakes in a 3-dimensional morass of gene strands. It’s akin to spotting potholes on every street all over the country and getting them fixed before the next rush hour.” Dr. Bennett Van Houten – of note: A bacterium has about 40 team members on its pothole crew. That allows its entire genome to be scanned for errors in 20 minutes, the typical doubling time.,, These smart machines can apparently also interact with other damage control teams if they cannot fix the problem on the spot.
    http://www.sciencedaily.com/re.....123522.htm

    Again, “this is engineering sci-fi” as far as man’s best engineering is concerned,,,

    To focus in on the ‘over the top’ programming of life in particular,,, In our computer programs we write computer programs that can be read in only one direction and which have only a single code contained within the programming sequence(s). On the other hand, the programming in life, i.e. the sequences in DNA, can often times be read both forwards and backwards.

    Imagine Designing Codes That Generate Functional Information When Read In Either Direction – November 11, 2014
    Excerpt: DNA is usually read in the “sense” direction — the direction that translates into a protein. The translation machinery can work in the opposite direction sometimes, though, producing an “antisense” RNA. Given our acquaintance with language, this would seem puzzling; what possible meaning could come from reading a paragraph backwards? Two recent papers show that cells can make sense out of antisense, by creating long noncoding RNA transcripts (lncRNAs)
    http://www.evolutionnews.org/2.....91011.html

    ,, and the sequences in DNA are also now known to have multiple overlapping codes stacked on top of one another.

    Multiple Overlapping Genetic Codes Profoundly Reduce the Probability of Beneficial Mutation George Montañez 1, Robert J. Marks II 2, Jorge Fernandez 3 and John C. Sanford 4 – published online May 2013
    Excerpt: In the last decade, we have discovered still another aspect of the multi- dimensional genome. We now know that DNA sequences are typically “ poly-functional” [38]. Trifanov previously had described at least 12 genetic codes that any given nucleotide can contribute to [39,40], and showed that a given base-pair can contribute to multiple overlapping codes simultaneously. The first evidence of overlapping protein-coding sequences in viruses caused quite a stir, but since then it has become recognized as typical. According to Kapronov et al., “it is not unusual that a single base-pair can be part of an intricate network of multiple isoforms of overlapping sense and antisense transcripts, the majority of which are unannotated” [41]. The ENCODE project [42] has confirmed that this phenomenon is ubiquitous in higher genomes, wherein a given DNA sequence routinely encodes multiple overlapping messages, meaning that a single nucleotide can contribute to two or more genetic codes. Most recently, Itzkovitz et al. analyzed protein coding regions of 700 species, and showed that virtually all forms of life have extensive overlapping information in their genomes [43].

    38. Sanford J (2008) Genetic Entropy and the Mystery of the Genome. FMS Publications, NY. Pages 131–142.
    39. Trifonov EN (1989) Multiple codes of nucleotide sequences. Bull of Mathematical Biology 51:417–432.
    40. Trifanov EN (1997) Genetic sequences as products of compression by inclusive superposition of many codes. Mol Biol 31:647–654.
    41. Kapranov P, et al (2005) Examples of complex architecture of the human transcriptome revealed by RACE and high density tiling arrays. Genome Res 15:987–997.
    42. Birney E, et al (2007) Encode Project Consortium: Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature 447:799–816.
    43. Itzkovitz S, Hodis E, Sega E (2010) Overlapping codes within protein-coding sequences. Genome Res. 20:1582–1589.
    http://www.worldscientific.com.....08728_0006

    In fact, “there is one human gene that codes for 576 different proteins, and there is one fruit fly gene that codes for 38,016 different proteins!”

    Time to Redefine the Concept of a Gene? – Sept. 10, 2012
    Excerpt: As detailed in my second post on alternative splicing, there is one human gene that codes for 576 different proteins, and there is one fruit fly gene that codes for 38,016 different proteins!
    While the fact that a single gene can code for so many proteins is truly astounding, we didn’t really know how prevalent alternative splicing is. Are there only a few genes that participate in it, or do most genes engage in it? The ENCODE data presented in reference 2 indicates that at least 75% of all genes participate in alternative splicing. They also indicate that the number of different proteins each gene makes varies significantly, with most genes producing somewhere between 2 and 25.
    http://networkedblogs.com/BYdo8

    ,,, such astonishing programming, where codes are stacked on top of one another, to repeat myself, is simply engineering ‘sci-fi’ as far as our best human programmers are concerned.

    And to point out the obvious, and as the title of the paper I cited stated, “Multiple Overlapping Genetic Codes Profoundly Reduce the Probability of Beneficial Mutation”

    Multiple Overlapping Genetic Codes Profoundly Reduce the Probability of Beneficial Mutation
    – George Montañez 1, Robert J. Marks II 2, Jorge Fernandez 3 and John C. Sanford 4 – May 2013
    Conclusions: Our analysis confirms mathematically what would seem intuitively obvious – multiple overlapping codes within the genome must radically change our expectations regarding the rate of beneficial mutations. As the number of overlapping codes increases, the rate of potential beneficial mutation decreases exponentially, quickly approaching zero. Therefore the new evidence for ubiquitous overlapping codes in higher genomes strongly indicates that beneficial mutations should be extremely rare. This evidence combined with increasing evidence that biological systems are highly optimized, and evidence that only relatively high-impact beneficial mutations can be effectively amplified by natural selection, lead us to conclude that mutations which are both selectable and unambiguously beneficial must be vanishingly rare. This conclusion raises serious questions. How might such vanishingly rare beneficial mutations ever be sufficient for genome building? How might genetic degeneration ever be averted, given the continuous accumulation of low impact deleterious mutations?
    – ibid

    To put the devastating implications for unguided Darwinian processes bluntly, (and as David Berlinski wryly noted), “applying Darwinian principles to problems of this level of complexity is like putting a Band-Aid on a wound caused by an atomic weapon. It’s just not going to work.”

    “applying Darwinian principles to problems of this level of complexity is like putting a Band-Aid on a wound caused by an atomic weapon. It’s just not going to work.”
    – David Berlinski

  6. 6
    ET says:

    Technical detail but AVIDA isn’t a genetic algorithm. But anyway, the following article is always helpful when discussing AVIDA: Evolution by Intelligent Design: A Response to Lenski et al.:

    When Lenski et al. created a simulation with high irreducible complexity, i.e. there was no selective advantage until the target function arose, EQU never evolved! Consider this quote from the Lenski paper:

    “At the other extreme, 50 populations evolved in an environment where only EQU was rewarded, and no simpler function yielded energy. We expected that EQU would evolve much less often because selection would not preserve the simpler functions that provide foundations to build more complex features. Indeed, none of these populations evolved EQU, a highly significant difference from the fraction that did so in the reward-all environment (P ~= 4.3 x 10-9, Fisher’s exact test).”

    In other words, when there is no selective advantage until you get the final function, the final function doesn’t evolve.

  7. 7

    I was a software developer from 1970 into 2010. Much of that career was spent in debugging code – my own and others.

    Software is design intensive: from initial concept – to design – to prototyping – to coding – to installation and test – to delivery, there is a constant and intentional drive towards, not perfection, but a system “good enough.” All along the way problems show their ugly head and must be stamped out – by intelligent debugging.

    Effective intelligent debugging is a skill attained chiefly by experience. Understanding a complex set of code, often times a distributed set, often takes months and years. Often time “luck” comes into play and a fix is stumbled upon early – this is rare but welcomed.
    Often times a seemingly very small flaw can cause catastrophic failure. A single bit or configuration flag in error can bring failure.

    I recall instances where a sold off and delivered system begins to fail when under the stress of real-world operations. A case in point was when a datalink synchronization bit pattern was inadvertently cut in half – a small matter of 8 bits. The 8 bits of the existing half of the pattern was sufficient for the vast bulk of messages sent and received, but that extremely small percentage that failed caused the entire networked and distributed system to fail resulting in many extremely angry fighter pilots having to interrupt their training missions.

    NO . . . Software does not and cannot evolve. I’ve been there!

  8. 8
    JVL says:

    Ayearningforpublius: A case in point was when a datalink synchronization bit pattern was inadvertently cut in half – a small matter of 8 bits. The 8 bits of the existing half of the pattern was sufficient for the vast bulk of messages sent and received, but that extremely small percentage that failed caused the entire networked and distributed system to fail resulting in many extremely angry fighter pilots having to interrupt their training missions.

    I don’t know when or where you worked for the DoD (directly or as a contractor) but it is amazing how many errors creep through despite extensive hours of Q&A testing. I loved watching experienced naval officers smoke test software simulators; when they gave a simulation their thumbs up that was high praise indeed.

  9. 9

    JVL …
    Our system for years was known as TACT/ACMI. A popular name was the Top Gun System as it was what the Navy Top Gun and the Fighter Weapons School at Nellis AFB used for training. Missions were very expensive, especially as the number of participants increased dramatically over the years. So the users were very intolerant of system failures.
    There were instances where a single bit -on/off – set to the wrong state could bring a system down.
    And yes, our systems were heavily smoke tested in the real world of fast moving and maneuvering fighter aircraft.

Leave a Reply