Perry Marshall: What is Random?

In the post Are Evolutionary Mutations Random? Or Are They Engineered for Success? (which I have posted here several times) I show several experiments such as McClintock’s in which what the organism needs is directly connected to the mutations. I cover this extensively in my book which further references a large number of similar experiments. I encourage you to read the referenced literature.

I will engage further once Joshua provides his definition of the term randomness. His definition and mine appear to be far apart.

Yes, and you have also refused to quote relevant portions and refused to engage in discussion here about it. Why not? I still don’t understand.

So you will engage further here if I repeat my definition (again) of random? I hope so, as an inordinate amount of time is being spent talking about talking.

Perry Marshall’s “Random”

Let’s start by noting your definition. Your definition is not consistent with the field of probability and statistics. You write,

“Random” means absence of pattern and purpose.
(Random vs. Stochastic Evolution)

This is an idiosyncratic definition. I’m not sure it is even possible to construct a random variable without a pattern. Every random feature I know about follows a pattern of some sort or another. I’d honestly like to know where you came up with this definition. You offer no citations. I can’t mathematically make sense of your definition.

“Random” In Statistics and Probability

As for my definition, there is no surprises. I’ve already given my definition several times @Perry_Marshall. On both this thread, and on those that have spin off from it. I’ll give it to you again, quoting form wikipedia to demonstrate that my definition is not idiosyncratic.

This, of course starts to raise questions about your distinction between stochastic and random. They are the same things.

I also emphasize that mutations (like all random variables) have patterns and can be the consequence of purposeful intention. Mutations, also, are not independent of fitness, even though they are still random. I’ve given several examples that demonstrate just these facts.

1 Like

I think this bears repeating …

According to Aquinas:

The effect of divine providence is not only that things should happen somehow, but that they should happen either by necessity or by contingency. Therefore, whatsoever divine providence ordains to happen infallibly and of necessity, happens infallibly and of necessity; and that happens from contingency, which the divine providence conceives to happen from contingency.

This is mirrored in the 2004 document Communion and Stewardship: Human Beings Created in the Image of God issued by the Catholic International Theological Commission.

So according to Catholic doctrine, events that we perceive as random certainly do not necessarily appear so to God.


Why not quote from the wikipedia page on randomness?


Great question. Wikipedia is hit or miss on technical definitions. It is just in error to say randomness has no patterns. It hard to even imagine randomness without any pattern. Even the wikipedia page you link to contradicts itself. It says:

A random sequence of events, symbols or steps has no order and does not follow an intelligible pattern or combination.

This seems to match Perry’s definition. But then it says:

For example, when throwing two dice, the outcome of any particular roll is unpredictable, but a sum of 7 will occur twice as often as 4. In t

Clearly, that is a pattern. It is enough of a pattern for EES proponents to call it “not random.” Moreover, it is never -1, nor is 15, which is also a “pattern.” Note also, that the image is of “noise” that we already explained earlier has a pattern. Randomness is not defined as “without any pattern.”

@EricMH and @Perry_Marshall I’d love to see you produce an example of randomness without any patterns. I’m not even sure this exists.


A post was split to a new topic: Evolutionary Encryption


Thank you. It looks like we are getting somewhere.

My definitions non-idiosyncratic. They are standard definitions from the dictionary, Philosophy of Statistics and Gregory Chaitin’s famous paper in Scientific American. I defined them at the outset and referenced my chapter “All About Randomness” in Evolution 2.0.

You can review my definitions at It’s Time to Tighten the Definition of “Random”

You said:

“I’m not sure it is even possible to construct a random variable without a pattern.”

Let’s put some parameters around this. You can go out and buy a hardware random number generator like this one

and we can use it to generate random numbers between 0 and 1.

Our probability distribution is: no outputs less than zero; no outputs greater than one; and equal distribution for all numbers between 0 and 1. In other words there is as little pattern as possible in the distribution of those numbers.

Chaitin points out that if, within those parameters, it is impossible to describe the output with an algorithm that’s simpler than the data itself, then the pattern is random. That is the most rigorous definition of “no pattern.”

A phenomenon may have other nuances and still be considered random. It could be white noise or pink noise. But it is still noise. If converted to audio or video looks and sounds like a TV between stations. You can the bass and treble controls up and down, you can tweak the color filters, but it’s still random.

Joshua, your definition is not as clear as one might hope. Now, I do believe I understand what you mean when you say a completely deterministic system is a random variable with entropy of zero.

It appears to me that you modeling EVERYTHING as a random variable. But then some of your variables are 0% unpredictable (deterministic) and others are 100% unpredictable (high entropy) and other variables are somewhere in between.

OK, you can do that. I believe I follow your logic. I do understand that while the roll of dice is random, they are only capable of producing numbers between 1 and 6 and yes that is a pattern.

But this misses the point. The point is NOT that dice are restricted to 1-6 or that random numbers are restricted to a range of 0-1. The point is that within stated parameters, the system is unpredictable.

Yours is an unconventional, idiosyncratic way of using the term random. With your definition we have to go a step forward and also define how much entropy is in your random variable. Because until you do that, your statement that something is “random” doesn’t give us any information. In your usage, everything is random.

That is not helpful. One should always prefer precise definitions over vague ones.

The real problem comes when you criticize people like Shapiro and Noble for saying that “mutations are non-random”. You are judging their statement according to your use of the term, not theirs. It is wrong of you to do that. You need to use the terms the way they define them. It’s only fair.

Denis Noble is abundantly clear in Was the Watchmaker Blind? Or Was She One-Eyed? that many things in physics and biology are random. But he then provides many examples in this paper of organisms 1) actively harnessing stochasticity to obtain favorable outcomes; 2) exhibiting sophisticated systematic responses to threats, damage, etc including repair, re-arrangements of coding sequences etc; He also shows 3) that these responses are contextual and depend on the nature of the threat.

Shapiro has also documented this exhaustively - this paper for example

When they say “random,” they mean the same thing I mean. They mean there is little or no pattern, and within stated parameters, randomness resembles noise.

And when they say transposition or horizontal gene transfer or other natural genetic engineering mechanisms are non-random, they are saying that those changes are systemic responses that follow discernible rules. The organism is orchestrating those changes to maintain homeostasis and adapt. They are saying that transposition, epigenetics and HGT are not accidental and not purposeless.

I debated this issue with PZ Myers: Memo To PZ Myers: Damage is Random. Repair is Not. PZ said mutations that generate evolutionary events are random. McClintock showed that they are not. She won the Nobel Prize for determining what the patterns are.

In physics & engineering, the word “random” invokes a continuum:

Random = little or no pattern < —————— > Non-random = systematic, deterministic, predictable

In evolutionary biology, mutations were long believed to be random with respect to fitness. Today, as you have pointed out, we know this is not true. In keeping with this, Shapiro and Noble are saying that evolution is not random events, it is systematic response to random events. They document this thoroughly.

My time is limited. I do not have the bandwidth to explain their work in detail here in this forum. But fortunately the papers cited above make their case quite clear. My book Evolution 2.0 offers a condensed and simplified account of the Extended Synthesis.

I know some people will object to various claims I have made about biology and mutations. I suggest they read these papers with great care.

To me what is important in all of this is that, contrary to popular books by Dawkins, Coyne etc., the mutations that generate evolution are not random, purposeless or accidental. And they are most emphatically not “copying errors.” Cells have a capability of altering their genetic sequences that is truly impressive. There is much we can learn by studying it.


I’m entirely on solid ground here. Why do you feel the need to defend them?

Also your random number generator follows a distribution and has a range. It has pattern. Look at what Shapiro writes about recombination. That logic would lead us to conclude it is not a random number.


That’s very common in biology.

And one should prefer empiricism over rhetoric. Do you agree?

Does he provide any new experiments in the paper?

Has he DONE anything?

No, she did not. She won it for her discovery of mobile genetic elements. You’re desperately trying to pretend that she won the Prize for the words she wrote instead of the real work that she did.

According to you. But the real test is, are they doing anything?

I think we’ve been pretty careful to note that they are only random with respect to fitness, so your removal of that important qualification is troubling.

Indeed, we can learn much more by studying mutations than by writing about them. Aren’t Noble and Shapiro only doing the latter?


So, @swamidass, I think Perry has a good point. You should look at Shapiro and Noble’s definition of random. According to their own definition, are they wrong?

@Perry_Marshall, I don’t know for sure, but @Mercer is probably right that McClintock won the nobel for discovering transposition/transposable elements, not for showing that transposition is a non-random process. Nevertheless, @Mercer, what she wrote is important.


Why should Perry defend Shapiro and Noble? Ask him. But I can answer for myself: I was involved in the science/faith debate for years, and until coming across Shapiro and Noble, I held that evolution definitely occured but that God probably mutated a ton of gametes every several million years to assist with all the major evolutionary transitions. Nothing I read from anywhere convinced me that random mutations (which I understood to mean “independent” of any fitness to the organism) and natural selection were sufficient to bring about all the biological diversity we see today. That was essentially WLC’s position in his last defenders class on creation/evolution.

Shapiro and Noble, popularized through Perry, showed me a variety of mechanisms I’d never heard of before that weren’t random (again, using Perry’s definition) and that generate considerable biological complexity. Wow!!! Now I see how evolution is both really interesting and points towards God THROUGH the naturalistic mechanisms. I’m not denying God’s intervention, maybe he wanted to “intervene” in his creation for Robin Collins’s “incarnational” purposes, maybe not.

As long as transposition, horizontal gene transfer, epigenetics, niche construction, etc. are simply subheadings under “random variation,” the general public WILL NOT really understand how evolution works. I feel like that’s why Shapiro, Perry, Noble, Jablonka and others are doing very important things.



Outside of discussions of Third Way/EES, how much is epigenetics, developmental bias, transposition or niche construction brought up on this forum? My guess is, not much.

I’m pretty sure @Agauger brought up the fascinating topic of developmental bias because I asked you about it and you said, “of course there is.” Maybe I’m wrong about that. The EES may not matter for you, because you’re very well informed. But it sure matters for scientific laypeople like myself.

1 Like

@Mark your guess is incorrect. We discuss these things all the time. They were known and studied long long before EES, and they were a fundamental part of my education. There is much more to learn here, but this is not EES’s contribution.

Great. Make that known. No one disagrees with this.

One does not need to combine it with misunderstandings about statistics and probability to make the point. I suggest departing from EES’s language, and just focus on the biology. You will get much farther because most biologists, suddenly, will largely agree with you.

Most biologists I know do not have much positive to say about Shapiro and Nobel. Their work depends on large misconceptions of biology and, essentially, a pseudohistory of the field. There is no advantage to hitching your wagon there. This discussion should only be making this assessment more obviously true. It should not require redefining the meaning of “random”, away from the meaning used in statistics and probability theory, to make ones point.

1 Like

@Mark let me suggest that biology matters to you, not EES. We can help you understand all the things you initially heard about in EES, but in a way that won’t provoke an argument with knowledgeable people. Why not get all the benefits of EES, with none of the downsides? Welcome to mainstream evolutionary science and the fundamentals of biology. We can help you get all the advantages with none of the liabilities.

1 Like

You are the first person I’ve met who classifies even deterministic systems as “random.” Your use of this term is different from everyone else’s. That’s shaky ground as far as I’m concerned.

1 Like

Dice rolls are the outcomes of deterministic systems with imperfect information, and they are the archetype of random variables. Nothing innovative in my view, nor did I classify them as random. Rather I said we model them as random variables.

It does seem you come at this from an engineering perspective, rather than probability and statistics. That would explain much of your confusion and why you’ve never encountered this before. In computational biology (and in statistics) we model both deterministic and non deterministic processes with random variables. In biology, statistics is so powerful because it gives a way to quantitatively reason about systems with high amounts of noise and where our knowledge is imperfect.

As I mentioned already, you will also likely get some clarity from understanding how intelligent agents are modeled as random variables too.

Random just never meant “purposeless and without pattern” in biology. Instead, science is a metaphysical desert, silent about divine purpose and using random variables to describe the patterns in the data. Much of what you have learned from EES is valid biology at its foundation, but their unique jargon is a mess. Keep the biology, lose their rhetoric, and you might really have something in Evolution 2.0.


By the way @Perry_Marshall, your story is important. I’m really glad you shared it. If we can just find a way around this hiccup, I’m sure there are ways I can help you forward. You are doing important work, and I appreciate your efforts. I am sorry I can’t go along with you on the road to EES. It would require ignoring too much of what I know of computational biology.

1 Like


I think this quote from you in sy’s thread on this is a very good definition if what you mean by random:

“All that “random” means is that the mutations are not entirely predictable from the modeler’s point of view.”

When Perry used the word, he is using it how it is commonly understood by most non-scientists, I would assume. Never in my life had I had heard of your definiton before. Can you at least see why so many laypeople believe that evolution is driven by mutations that are all as equally unlikely as the other?

Your way seems to require a lot of scientific evangelization regarding what the word random ACTUALLY means. If this is what random means in your field, I think you need to proclaim it from the rooftops because there will be a lot of Christians who breathe a lot easier with that definition, especially when you explain that some mutations or evolutionary changes are are slightly more predictable than others.

Also, do you agree with Perry that most mutations are not “copying errors”? If you do, again, that’s huge, because “copying errors” is all over Adam and the Genome. Again, this would be a watershed moment for a lot of Christians, like it was for me.

By the way, I do agree with Perry. I think his argument that the genetic code IS an actual code is very strong, as does @sygarte, he’s said so on his blog (not specifically about Perry, but that the genetic code IS a code). If it IS a code, then it follows naturally that evolution can’t proceed by copying errors.

Again, I’m NOT saying God had to make the code, it still could have been generated naturally through emergence, cosmic fine-tuning, and self-organization.

1 Like

That is exactly what I have been doing for nearly five years. So have others.

I can see why the public is confused, but I do not agree that @Perry_Marshall is following the common usage of “random.” Instead, he is using the anti-evolution version of the term, which tries to link evolutionary science with Epicureanism and ontological randomness.

In common usage “random” events are unpredictable or unexpected events. Instead of denying that they can be purposeful, we commonly call them “acts of God” or “providential.” We even use “random” to describe clearly purposeful actions (such as the decision to accept or reject an applicant in a competitive situation). There is Scripture that talks about the casts of lots as a place where God makes his purposes known. Even in non-Christian contexts, “luck” or “karma” are used to explain that purposes are being worked out in “random events”. All of this is entirely consistent with (though not part of) the scientific meaning, which is merely saying we can’t entirely predict most events.

Now, there is a long history of treating “random” in evolution as a uniquely insidious challenge to theology because…well, we have no explanation why. Yet, generation after generation of anti-evolutionist has beat that drum, saying that “random is without divine purpose” even though they might call all the “random” events in their own life “providential.” So yes, I do understand why the public is confused, but I disagree with your diagnosis on why they are confused. It is not my fault that anti-evolutionists are self-contradictory and selective in their objections. Somehow, only when mentioned with respect to evolution, “random” must mean ontological randomness, rather than the common meaning of the word which matches the scientific meaning: not fully predictable.

Given my understanding of this situation, I find the EES approach to be increasing the problem rather than fixing it. It doesn’t make sense either in science or in common language. Christians concerned about the word “random” would benefit immensely from learning about the theology and Scripture of providence. With this in mind, there is no valid objection to the term, unless science insists on ontological randomness (which it does not).

“Copying errors” is not neutral enough for me of an explanation. It is a common way to describe mutations, but it is only an analogy. The analogy breads down. I’d rather say that “there many types of mutations and they cause different types of variation to arise.” Perhaps some people call them “copy errors” but it is very easy to find places where that analogy breaks down.

It is a code merely by analogy. We can list out reasons why it is a code. We can also list out reasons why it is not a code. The analogy breaks down. If they haven’t explained the ways the genetic code is not like a code, they have left out half the story. Why would you find this acceptable?


Alright @Perry_Marshall, I want to show you how this random number generator is NOT actually random by the EES definitions that Shapiro uses. For reference, read up on how Shapiro discusses recombination (Take 2: Why Genetic Recombination Is Not Random, and How Cells Take Advantage of Non-randomness | HuffPost). I will go quote by quote through his article, explaining why recombination is not “random.”

Homologous recombination is not accidental.

The “TrueRNG-V3-Hardware-Random-Generator” is not accidental either. It is a USB accessory for goodness sakes, being sold because it has a purpose.

It is a required part of the special cell divisions called “meiosis” that that produce sperm and egg cells with only one copy of each chromosome. Without meiosis, sexual reproduction would not be possible as found today in higher organisms.

The “TrueRNG-V3-Hardware-Random-Generator” is a required part of certain software too.

Molecular studies have revealed that at least a dozen proteins acting in sequence carry out such accurate cutting and splicing (one form of natural genetic engineering).

There are far more parts in the “TrueRNG-V3-Hardware-Random-Generator” than in recombination, and it is a profoundly complex activity to produce the supposedly “random” numbers.

But cells do not leave to chance what happens to their genomes. They use special DNA-cutting enzymes to initiate recombinational exchange. A University of Chicago colleage, Rochelle Esposito, discovered the enzyme used in meiosis. This enzyme prefers certain DNA sequences to cleave and creates so-called “hotspots” where recombination occurs most often. Sequence preference is one source of non-randomness in recombination.

The “TrueRNG-V3-Hardware-Random-Generator” has several parameters that are not set by chance. For example, in the usual configuration it produces 1s and 0s in a ratio of 50%, with high precision. This is a clear pattern.

Recombination between repeats at different locations leads to chromosome rearrangements. Because the locations of the repeats determine where the rearrangements occur, this is another non-random feature of recombinational exchange.

The “TrueRNG-V3-Hardware-Random-Generator” only outputs 0s and 1s, which is another non-random feature.

In this blog, we have seen some functional advantages of non-random natural genetic engineering by homologous recombination.

On this blog we have seen several ways that the “TrueRNG-V3-Hardware-Random-Generator” is not random at all, but the product of a highly engineered process with strict criteria and controls on its behavior.

You heard it from EES first, the “TrueRNG-V3-Hardware-Random-Generator” is not random at all.


I’m very sure. Do you think that Perry will correct or retract his false claim?

Not nearly as important as her work.

They didn’t show them, they merely relabeled them.

Mark, if they really believe their rhetoric, why haven’t any of them devoted their efforts to repeating the Luria-Delbruck experiment (or analogs thereof) in other systems and situations?

Doing??? I’m pretty sure that they are just talking. Wouldn’t it be more significant if they really did something and showed that the Luria-Delbruck experiment cannot be generalized?