Eric Holloway: The Salem hypothesis as to why engineers doubt Darwin

_{News
June 7, 2022

Evolution, Informatics, Intelligent Design

2}_{Categories
Evolution
Informatics
Intelligent Design}

Share: Facebook; Twitter/X; LinkedIn; Flipboard; Print; Email

Not because they’re terrorists or black-and-white thinkers, as claimed. A simple computer program shows the limits of creating information by chance:

Engineers are more likely to be creationists because they are familiar with what it takes to design complex things for specific tasks. Which is exactly what we see in the biological world. Additionally, engineers who work with computers know about randomized methods, which include evolutionary algorithms. We are aware of their significant limitations…
Let’s set my evolutionary algorithm a simple, fundamental task — to count by ones…
Exponential is bad news. Exponential means it took the evolutionary algorithm twice as long to count to 10 as it did to count to 9.
Let’s put exponential on a cosmological scale. The heat death of the universe is projected to occur in 10¹⁰⁶ years. This is well beyond the lifetime of anyone who’ll even remotely know we existed. Seems like a lot of time, but not for exponential doubling!
If we generously say that a step of P’‘ runs in a nanosecond, which is nine decimal places to the right of the dot, then the universe will undergo heat death before the evolutionary algorithm can evolve a program that counts from 0 to 500. And it takes even longer if the program must start from 1 instead of 0. To go up to 501 doubles even that. Completely impossible.
Eric Holloway, “The Salem Hypothesis: Why engineers view the universe as designed” at Mind Matters News (June 7, 2022)

Takehome: Engineers doubt chance evolution because a computer using an evolution-based program to do simple tasks would be chugging away well past the heat death of our universe, as Eric Holloway demonstrates.

Note: The hypothesis was named in honor of Talk.origins contributor Bill Salem.

You may also wish to read:

Dawkins’ Weasel program vs the information life acquires en route To demonstrate what is wrong with fully naturalist assumptions like those of Richard Dawkins’ Weasel program, I developed Weasel Libs, modeled on Mad Libs. When we apply a Mad Libs “epigenetic” approach to Dawkins’ claims about how life’s information can be created, we quickly see a glaring flaw. (Eric Holloway)

Comments

Relatd:
Evolution: Human beings are here by accident. That’s it.
No, that is evolutionism, ie the claim that life's diversity arose via blind and mindless processes. To say that ID is anti-evolution means it argues against things like anti-biotic resistance. That it argues against descent with modification. It doesn't.ET_{June 10, 2022
June
06
Jun
10
10
2022
04:10 PM
4
04
10
PM
PDT}

Martin_r Applications for genetic algorithms. A company I worked for used them for programming FPGAs. And there is Product-form design model based on genetic algorithmsET_{June 10, 2022
June
06
Jun
10
10
2022
04:07 PM
4
04
07
PM
PDT}

Turing test? No comprende. It's not applicable. Something I have not said yet is this: Some companies want to get rid of workers, especially highly paid workers, so they buy automation/computers that can design/build things for them. Get rid of people, and make more money. But as that happens, who will buy anything from Target?relatd_{June 10, 2022
June
06
Jun
10
10
2022
02:03 PM
2
02
03
PM
PDT}

Relatd @68, You keep saying the same thing. Are you familiar with the Turing test?
“Computer. Tell me how to tie my shoes.” You’ll get nothing.
Have you tried asking your question to Alexa or Siri? Maybe you're simply choking on the term, artificial intelligence. Maybe "simulated intelligence" or "projected intelligence" would be more acceptable. Didn't you look at the link I provided you to Alpha Zero? https://www.chessjournal.com/alphazero/ Artificial intelligence is not the same as human intelligence just as an "iron horse," a term used for a steam locomotive in the early 1800s, is not the same as a biological horse. But an iron horse does the work of a biological horse. We even compare artificial horses using the term "horsepower," and we also use "candelas" to compare artificial light sources with that of wax candles. -QQuerius_{June 10, 2022
June
06
Jun
10
10
2022
01:41 PM
1
01
41
PM
PDT}

Querius at 67, "...they mimic the behavior of intelligence when their behaviors are actually an extension of intelligence." Machines have no behaviors. Your microwave has no behaviors. Machines are not an extension of intelligence. They have nothing that was not programmed in. "Computer. Tell me how to tie my shoes." You'll get nothing.relatd_{June 10, 2022
June
06
Jun
10
10
2022
11:37 AM
11
11
37
AM
PDT}

Martin_r, My educational background is scientific, but my vocational background is primarily in industry. I've had the privilege of interacting with many engineers in many different disciplines worldwide. Plus, I read a lot. Yes, the biological world screams design to engineers who recognize the mind-boggling technologies and capabilities of biological systems. They realize that shooting bullets randomly at a Tesla will never result in design innovations. Bullets can only destroy things. But if you're in a Tesla being swept down a river in a flood and can't get the window open to exit, bullets may open a window, destroying it save your life. Engineers understand the importance of knowing the specifications and constraints of any project. They are intimately and sometimes painfully familiar with their experiences dealing with trade-offs and compromises. Specifications include performance, time, cost, reliability, manufacturability (including common parts, and tolerance costs and stacking), testability, manufacturing yield, flexibility under changing conditions (temperature, humidity, pressure, dust, mud, corrosion), robustness, repairability, upgradability, usability, compatibility, and "charm." I'm sure I missed other factors. This is one reason why designs improve by successive approximation after being subjected to random use and abuse. An important point is that hyper-optimization and over-specialization makes an organism, an ecosystem, a business, an entire economy, or an individual career fragile and subject to extinction. Even academics realize this as they learn more and more about less and less until they know everything about nothing. (grin) Examples in biology include ecological and agricultural monocultures, and also the risks associated with a lack of genetic diversity in specific organisms. Introducing cats, rats, wild hogs, and a host of other invasive species can devastate an ecosystem, which leads to my observation that the survival of an ecosystem may be at odds with the success of any component species. AI, machine learning, and generative design are programmed to develop and apply rules to an environment along with instant feedback. Also, parametric technologies embed a specified range of variation into a design. The intelligence is already programmed/designed in to generate different configurations, but they mimic the behavior of intelligence when their behaviors are actually an extension of intelligence. My best example of AI is in well-programmed chatbots (or trollbots) that mimic human communication and pass the Turing test. Here at UD, there are a number of contributors that might indeed actually be trollbots, judging from their pattern of abusive comments, lack of new information, and vacuous responses. -QQuerius_{June 10, 2022
June
06
Jun
10
10
2022
10:14 AM
10
10
14
AM
PDT}

ET, Besides the antenna, I would expect, that you will show me more real life products engineered using EA/GA. Instead of that, you have referred me to some article ... you start sounding like a Darwinist... anyway, . i will have a look at that article ...martin_r_{June 10, 2022
June
06
Jun
10
10
2022
09:11 AM
9
09
11
AM
PDT}

ET at 56, Seriously? Let's go the even more complicated route? I don't think so.relatd_{June 10, 2022
June
06
Jun
10
10
2022
08:50 AM
8
08
50
AM
PDT}

Martin_r at 54, The Global Cabal of Relabelers and Repackagers have one goal: to make people believe they are in living in "the future." Everything being designed and built today is based on hundreds of years of knowledge. And I'll say it again, there is no such thing as Artificial Intelligence. No machine has human-like intelligence just programs created by human beings. I doubt the chess playing AI could tell me how to tie my shoes. Machines, including computers, are stupid. Totally stupid unless a human being creates programs for them.relatd_{June 10, 2022
June
06
Jun
10
10
2022
08:28 AM
8
08
28
AM
PDT}

ET at 58, Inventions don't "evolve." Never have. Things are made to satisfy a usually urgent need.relatd_{June 10, 2022
June
06
Jun
10
10
2022
08:22 AM
8
08
22
AM
PDT}

ET at 60, Evolution: Human beings are here by accident. That's it.relatd_{June 10, 2022
June
06
Jun
10
10
2022
08:20 AM
8
08
20
AM
PDT}

Martin-R wings are result of this built-in adaptation ?
:) Ask for scientific evidences to see if you get an answer. PS: I think "genetic algorithms" are a kind of a hoax. Nobody knows all the details of the cell therefore nobody can explain(or programm) what they don't understand.Lieutenant Commander Data_{June 10, 2022
June
06
Jun
10
10
2022
07:57 AM
7
07
57
AM
PDT}

How are you defining "evolution"? Evolution is more than universal common descent. Michael Behe:
Again, as I made abundantly clear at trial, it isn’t “evolution” but Darwinism — random mutation and natural selection — that ID challenges. Darwinism makes the large, crucial claim that random processes and natural selection can account for the functional complexity of life. Thus the “burden of proof” for Darwinism necessarily is to support its special claim — not simply to show that common descent looks to be true. How can a demand for Darwinism to convincingly support its express claim be “unreasonable”?
Also Michael Behe:
Scott refers to me as an intelligent design "creationist," even though I clearly write in my book Darwin's Black Box (which Scott cites) that I am not a creationist and have no reason to doubt common descent. In fact, my own views fit quite comfortably with the 40% of scientists that Scott acknowledges think "evolution occurred, but was guided by God."
ET_{June 10, 2022
June
06
Jun
10
10
2022
07:10 AM
7
07
10
AM
PDT}

ET, One more thing... you post very often that ID folk is not anti-evolution. I personally don’t believe that a humming bird or a honey bee or a blue whale or a cheetah evolved from some common ancestor ... from engineering/design point of view, i am pretty sure these species were designed from scratch. Of course, there seems to be some sort of built-in adaptation, but the question is, to what extent could species adapt ... do you believe that e.g. wings are result of this built-in adaptation ?martin_r_{June 10, 2022
June
06
Jun
10
10
2022
07:04 AM
7
07
04
AM
PDT}

Scientific American did an article about 20 years ago (2003?) titled "Evolving Inventions". Read it.ET_{June 10, 2022
June
06
Jun
10
10
2022
06:52 AM
6
06
52
AM
PDT}

ET, At that time (when i did my survey) i was also looking for real life products engineered using EA/GA ... the antenna you have mentioned was the only thing i was able to find.martin_r_{June 10, 2022
June
06
Jun
10
10
2022
06:44 AM
6
06
44
AM
PDT}

Antenna Design using Genetic AlgorithmsET_{June 10, 2022
June
06
Jun
10
10
2022
05:21 AM
5
05
21
AM
PDT}

EricMH,
There is nothing special about evolutionary algorithms that can explain how biological evolution can overcome the difficulties brought up during the Wistar Symposium. In fact, they tend to be bottom of the pile in terms of tools engineers reach for to solve problems.
Let me add to yours ... in my debates with Darwinists, EA/GA were mentioned very often. Darwinian debaters claimed that "look, even engineers use evo/genetic algorithms. People here at UD know that i am a mechanical engineer. I never heard of EA/GA, i never used EA/GA. I heard of optimization algorithms. You may say, that i never heard of EA/GA or never used them (as an engineer) because i am backwards. Few years ago, i made my own survey. I got in touch with some international robotic companies dealing with humanoid robots design (including Boston Dynamics). I have asked a very simple question "if their engineers use EA/GA when designing robots". I have received some answers ... I won't name the companies, i don't have the permission, it was a private communication. Anyway, here is what they replied: A company from Spain:
Our engineers do not use GA/EA in our company and due to that, I can not help you with much more. Wishing you all the best.
A company from Italy - in this case i talked to 2 engineers, both of them are also academics (university professors)
Engineer #1: "Quick answers: I was engineer for 20+ years, No, we haven’t used genetic/evolutionary algorithms." Engineer #2: "From my personal experience I agree with you in that I don’t see a lot of activity around this topics in the robotics industry."
A company from France:
"No we haven’t used such technique in real cases. It seems a great tool allowing the exploration of different design/approach. Yet, I have only seen such uses in research applications atm."
PS: Boston Dynamics did not reply, and i urged it a few times. No reply. The funny thing is, they even have a robot named "DARWIN" https://www.romela.org/darwin-dynamic-anthropomorphic-robot-with-intelligence-2/martin_r_{June 10, 2022
June
06
Jun
10
10
2022
12:00 AM
12
12
00
AM
PDT}

Querius, you seem to be familiar with terms like 3D printing, FEM, CAD, CAM, and so on. That is why i have asked you what is your education. You sound like an engineer. However, i have to insist on what i wrote before, this GENERATIVE-DESIGN technology has a very very very limited application. Basically, you confirmed it. It very reminds me of so called evolutionary/genetic algorithms (EA/GA). These also have a very very very limited application. I think, this AI GENERATIVE-DESIGN technology is another fancy [fake] word for EA/GA (like Relatd pointed out) But again, EA/GA is nothing new too. It just another fancy word for optimization algorithm. Of course, i can't entirely judge from one Autodesk PR article, but from what i could understand, i am pretty sure it is still about the same (EA,GA, GENERATIVE-DESIGN, optimization algorithm). Please don't get me wrong, i am not against any additional engineering tool, i am just saying, all these technologies have very very limited use in real life engineering (unlike FEM, CAD, CAM etc.) that is why i called it "another AI hype". Because lay people (e.g. Seversky) may think, that you take a computer with a proper AI software, you put in some variables and it will design a cheetah-robot for you.martin_r_{June 9, 2022
June
06
Jun
9
09
2022
11:28 PM
11
11
28
PM
PDT}

Taken seriously? Is that your argument? There isn't anything about Intelligent Design that is anti-evolution. Intelligent Design is OK with organisms being designed with the information and ability to evolve and adapt:
He [the Designer] indeed seems to have “carefully crafted” information in His species giving them the ability to respond to environmental stimuli to alter their own genome to adapt to new environments. He then evidently let them wander where they will with the ability to adapt.- Dr. Lee Spetner “the Evolution Revolution” p 108
And in "Not By Chance", Spetner responds to "the Blind Watchmaker" thesis. For example, duplications, followed by building it a binding site, and then changing the duplicate for some new function, isn't justifiable as a blind watchmaker mechanism. Transposons carry within their sequence the coding for two of the enzymes required for the gene to move around. . On top of that there is artificial selection. And we see what that can do in a relatively short time. The link between genotype and phenotype is with the traits. Traits such as eye color and shape. Detached or attached earlobes. Cleft chin. Dimples. In his book (English title) “Why is a Fly not a Horse?”, the prominent Italian geneticist Giuseppe Sermonti, tells us the following: Chapter VI “Why is a Fly not a horse?” (same as the book’s title):
”The scientist enjoys a privilege denied the theologian. To any question, even one central to his theories, he may reply “I’m sorry but I do not know.” This is the only honest answer to the question posed by the title of this chapter. We are fully aware of what makes a flower red rather than white, what it is that prevents a dwarf from growing taller, or what goes wrong in a paraplegic or a thalassemic. But the mystery of species eludes us, and we have made no progress beyond what we already have long known, namely, that a kitty is born because its mother was a she-cat that mated with a tom, and that a fly emerges as a fly larva from a fly egg.”
I doubt your algorithm does as you say as no one knows what determines biological form. We know it isn't DNA. DNA controls and influences development but that only means it determines whether or not development doesn't have any issues.ET_{June 9, 2022
June
06
Jun
9
09
2022
07:35 PM
7
07
35
PM
PDT}

ET @ 47 The reason I say "evolution" instead of "Darwinism" or "random processes" is that all evolutionary mechanisms are random processes + natural selection. Nothing outside of that is taken seriously, e.g. saltationism, lamarkianism, and so on. Plus, Dawkins makes a pretty good mathematical argument in The Blind Watchmaker why nothing besides random incrementalism can work for evolution. So there is no real distinction. These are all the genetic mechanisms I've run across: 1. random insertions, deletions, and mutations of nucleotides 2. gene duplication 3. meiosis 4. horizontal gene transfer On top of that is only natural selection, where fit offspring reproduce, and unfit offspring are eliminated. The second part of evolution is the relationship between genotype and phenotype. Dawkins type weasel examples neglect this difference, but my algorithm faithfully reproduces this aspect of evolution. Genetic mechanisms operate on the genotype (the program code in my algorithm), and natural selection operates on the phenotype (the program output in my algorithm). So Fred @ 46, this is the reason my algorithm is representative of the power of evolution. My algorithm includes all of the above. I've not run across any other aspects of evolution that will significantly change the algorithm operation. If you know of any, please mention them. Otherwise, we can consider my algorithm to work or fail insofar as evolution works or fails. In fact, this was the original motivation behind evolutionary algorithms, the mathematical issues brought up during the Wistar Symposium in 1967. John Holland, the inventor of the genetic algorithm, specifically cites the Wistar Symposium as the reason for his interest in genetic algorithms. Where participants like Murray Eden and Schutzenberger saw mathematical difficulties and impossibilities in evolutionary theory, John Holland saw evidence of evolution's spectacular power. Holland then invented the "schema theorem" to explain how evolution could be so effective. Many decades of research into evolutionary algorithms hence, we've seen Holland's anticipation completely squashed (and the schema theorem disproven). There is nothing special about evolutionary algorithms that can explain how biological evolution can overcome the difficulties brought up during the Wistar Symposium. In fact, they tend to be bottom of the pile in terms of tools engineers reach for to solve problems. My algorithm is a good example of this failure. So, at this point, evolutionary theory has been put to the test in real time in the make or break world of commercial software, and has found very wanting. There is no reason to subsequently believe that some magic makes evolutionary theory work in biology. On the other hand, as my example also very clearly indicates, the Wistar Symposium mathematical problems are easily eliminated when a mind is involved in the solution creation process. So, the obvious implication is that a mind is involved in the biological creation process as well. Finally, if any evolution theory die hards think I missed some magical operator that makes evolution work, all you have to do is add it to my code, or tell me here in chat to add it, and we can see if it's the missing piece to this puzzle. No need to have decades of drawn out speculative debates online. We can test evolutionary theory easily on the computer and come to a conclusive answer.EricMH_{June 9, 2022
June
06
Jun
9
09
2022
06:43 PM
6
06
43
PM
PDT}

As Michael Behe said:
Again, as I made abundantly clear at trial, it isn’t “evolution” but Darwinism — random mutation and natural selection — that ID challenges. Darwinism makes the large, crucial claim that random processes and natural selection can account for the functional complexity of life. Thus the “burden of proof” for Darwinism necessarily is to support its special claim — not simply to show that common descent looks to be true. How can a demand for Darwinism to convincingly support its express claim be “unreasonable”?
ET_{June 9, 2022
June
06
Jun
9
09
2022
06:20 PM
6
06
20
PM
PDT}

ET @47, You're right in specifying a qualification. It's Darwinian evolution in tiny increments by random mutations filtered by natural selection that doesn't seem to work. Martin_r @41, The largest users of additive manufacturing are in medical and aeronautical industries. According to an online course I once took in the subject from Deloitte University, there are several technologies commonly employed. One example, used in the class showed how supply chains are changing in some cases. One of the examples they provided involved the U.S. military. To service the rotor of a type of helicopter, a special tool is required. Rather than fill up the supply channel with these tools, they issue 3D printers in the field, where additional tools are printed on demand. AI is indeed used to generate 3D honeycomb patterns in parts for the purpose of reducing weight while maintaining strength. The employment of finite element modeling (aka FEM/FEA) to determine stresses has been used for many decades in mechanical engineering. An architect neighbor of mine fed it my deck cover design to make sure it could support a heavy man walking across it. These data can show exactly where something needs to be beefed up. 3D FEM was used in the 1980s to optimize stratified ignition in the shape of automotive cylinders. I'd agree with you that engineers use technologies as they do any other tool, whether its a 3D printed prototype or CAD/CAM/CAE software. EricMH @43, What the Alpha Zero AI does is not simply a "brute force attack" against a chess position, and it doesn't trim branches of unpromising lines as do typical chess programs, but rather it develops rules and priorities by trial and error. These types of rules enable AI routines to "recognize" general types of shapes in a photograph such as kitten or a boat. I completely agree with you regarding the use of the scientific method for debugging software or finding design flaws or limitations. Out of curiosity, have you ever tried simulating an ecosystem using software? I'd suggest that stable ecosystems are also evidence for design. Optimization for the survival of only one specific organism is likely to "crash" an ecosystem, causing a devastating loss in carrying capacity. It seems that any ecosystem must be harmonized to sustain life (for example, think of what the introduction of rabbits did in Australia or what kudzu did in the southern USA). -QQuerius_{June 9, 2022
June
06
Jun
9
09
2022
05:48 PM
5
05
48
PM
PDT}

Yes, but we have to also understand that Intelligent Design is NOT anti-evolution. Evolution by means of telic processes, ie intelligent design, is still evolution. Evolution by means of telic processes works, as exemplified by genetic algorithms. Genetic algorithms are goal-oriented programs that utilize a targeted search to solve problems.ET_{June 9, 2022
June
06
Jun
9
09
2022
05:35 PM
5
05
35
PM
PDT}

"The claim is that evolution BY MEANS OF BLIND AND MINDLESS PROCESSES doesn’t work." It should be obvious now.relatd_{June 9, 2022
June
06
Jun
9
09
2022
05:25 PM
5
05
25
PM
PDT}

No one is claiming that "evolution" doesn't work. Eric needs to be more careful with his wording. The claim is that evolution BY MEANS OF BLIND AND MINDLESS PROCESSES doesn't work. As for reality, no one knows how evolution by means of blind and mindless processes "works".ET_{June 9, 2022
June
06
Jun
9
09
2022
03:59 PM
3
03
59
PM
PDT}

Eric upthread
My program shows evolution is not weak, it just doesn’t work.
Don't think so, Eric. Your program is a model. To be able to claim evolution doesn't work, your model needs to bear some resemblance to reality.Fred Hickson_{June 9, 2022
June
06
Jun
9
09
2022
03:41 PM
3
03
41
PM
PDT}

Scientific method and the engineering design process, are so similar it's a distinction without a difference.ET_{June 9, 2022
June
06
Jun
9
09
2022
02:40 PM
2
02
40
PM
PDT}

Regarding the scientific method, software developers exercise the scientific method every single day many times to debug their programs. Bugs often are non obvious, and we have to form hypotheses and falsify or confirm these hypotheses. We also have to infer logical laws and causes to find the root cause for the bug. Additionally, we often use the scientific statistical toolbag in this debugging. Software engineers are lucky, since we can step through our programs to see the behavior in real time. Other engineering disciplines have to be even more rigorous and insightful in their use of the scientific method, since bridges and rockets don't have debuggers you can attach to the finished product :) So, I would say engineers use the scientific method much more frequently than scientists, and are much more aware of how the scientific method works. Finally, our use of the scientific method has to actually work. Scientists just need to get their peers and grant committee to accept their claims, the claims don't actually have to work in the real world.EricMH_{June 9, 2022
June
06
Jun
9
09
2022
02:34 PM
2
02
34
PM
PDT}

@Querius Generative technology definitely has a place. I understand logistics planning makes heavy use of multiobjective optimization techniques, which generate the Pareto front of solutions with different tradeoffs. AIs like Alpha Zero are very powerful, and can beat the top humans (and other chess engines). However, in all these cases, the downside is the inefficiency. Algorithms have to check thousands, millions or billions of solutions to reach the level of humans checking just a few solutions. So, such techniques only work when the problem domain has been constrained enough for brute force search. Information is the reduction in options, and constraining the problem domain to be searchable is a matter of reducing the number of options. That is the role of the human engineers. We create the information that makes these algorithms effective by reducing the search space to a manageable size. There is no way to replace the information produced by engineers using an algorithm. The future of AI technology is going to be this hybrid between humans constraining the problem domain, and the algorithms brute force searching for solutions within the constraints. Humans do smart things slowly, and computers do stupid things quickly. The idea will be to figure out what are the smart slow things to hand off to humans, and the stupid fast things to hand off to algorithms. For example, to get rid of the exponential explosion, the evolutionary algorithm can be constrained to only add the "." and "+" characters to the end of the solutions. This constraint, of course, requires intelligence, since the constrainer has to understand the program pattern of '".+.+.+." and how it generates the count sequence.EricMH_{June 9, 2022
June
06
Jun
9
09
2022
02:14 PM
2
02
14
PM
PDT}

Prev 1 2 3 4 Next

You must be logged in to post a comment.

Leave a Reply