DNA Information and Computer Code

Meyer specifically cites computer code as an analogy to DNA information. In fact, he insists that it is more than an analogy. In chapter 17, he addresses several critiques of ID including the claim that it is based on analogy. On page 386 he states “Although a computer program may be similar to DNA in many respects and dissimilar in others, it exhibits a precise identity to DNA insofar as both contain specified complexity or specified information.
“Accordingly, the design argument developed here does not rely on a comparison of similar effects, but upon the presence of a single kind of effect—specified information—and an assessment of the ability of competing causes to produce that effect. The argument does not depend upon the similarity of DNA to a computer program or human language, but upon the presence of an identical feature in both DNA and intelligently designed codes, languages, and artifacts.” (emphasis in the original)

There is no doubt that there are many intriguing similarities between DNA information and computer code. Many of the same analytical tools can even be used in both systems. But Meyer says that this is not the basis for the inference to an intelligent designer. Rather, it is the fact that both are identical insofar as being specified information. I do not find his argument compelling because the two systems derive their specificity from two different sources, as Meyer himself admits, though he does not pursue the consequences of that difference. DNA information is specified because it works. The cell is able to carry out a very complex function. Cells that do not function, die. Those that function, live and undergo cell division. Computer code is also specified because it works, but now its function is defined by the meaning that intelligent beings have assigned to the “0”s and “1”s that are generated by the computer. The computer system is rife with abstraction and it functions because its physical complexity conforms to the basic principles of computer design and because of the abstract meaning assigned to those physical states. This difference is crucial and stands in contrast to the claim of identicality.

The physical requirements for a computer are simple but not trivial. Rolf Landauer described the requirements for physical states in an information processing system in a paper in 1961 (R. Landauer, “Irreversibility and Heat Generation in the Computing Process” IBM J. Res. Develop. Vol. 5, No. 3, 1961). Essentially, a stable physical binary state must exist which can be switched from one state to another. These states can be designated as “0” or “1” but that designation is independent of the specific physical system being used. The second requirement is a channel to transmit that information. Claude Shannon was the pioneer who quantified the information that can be transmitted in a physical system (C.E. Shannon, “A Mathematical Theory of Communication”, Bell System Technical Journal, vol. 27, pp. 379-423, 623-656, July, October, 1948). The third requirement is a set of logic gates that carry out Boolean logic, first defined by George Boole in the 19th century. These are the basic principles of computers. In addition there are numerous constraints to make a computer practical, fast, and efficient but those need not concern us here.

The key point for our purpose is that the functional specificity of any computer designed on the above principles depends on abstraction. There are many levels of abstraction. The first and most basic is assigning “0” and “1” to a binary system. This assignment is independent of the physical system selected and the “0” and “1” can be interchangeable, as long as it is done consistently. For example, a positive voltage may be designated a “1” and a zero or negative voltage may be a “0”. But there is nothing about the voltage that demands that it be a “1” or a “0”. The definitions can be reversed. Or a polarized photon could be used as a “0” or a “1”. It could be polarized either linearly or circularly. The “1” could be horizontally or vertically polarized and the other would be “0”. This assignment is arbitrary. Higher levels of abstraction are assigned to interpret binary strings in terms of ASCII codes or instruction sets or data or various other meanings. But the chemistry and physics of the system is irrelevant, provided the basic constraints are met.

In contrast, DNA information is composed of a linear set of units called nucleotides. Each sequential step of DNA has one of four nucleotides, usually designated A, C, T, or G for short. The sequence is critical and is the essence of the information, but its function also depends on a large number of supporting biomolecules, most of which are, in turn, generated from various segments of the DNA. Thus far, the similarity to computer code seems strong. But as we look closer, we see that the functional specificity is derived precisely from its ability to survive and reproduce, not from any abstract meaning. If the chemical and physical structure changes, the functionality is changed and may be lost. The information is not a matter of assigning a meaning to the nucleotides, but rather the chemical function that the DNA carries out. This is a crucial distinction because it is abstraction itself which is the strongest (though not the only) indicator of the action of an intelligent designer.

To summarize, the ID community, and specifically Meyer in this book, often cite the comparison of computer code and DNA information as evidence of an intelligent source for both. But though they both have physical complexity that we call information, there is a critical distinction in that computer code specificity is derived from abstraction of the physical complexity whereas DNA specificity is derived from its chemical ability to survive and reproduce the physical complexity. That fundamental distinction undermines the core argument that Meyer seeks to make. Specificity of complex information does not arise uniquely from an intelligent source but also from physical or chemical functionality.

11 comments to DNA Information and Computer Code

  • […] is close to what Randy Isaac is getting at in his review of  Meyer’s Signature in the Cell (most recent post in that series). God as designer is unlike any designer we know. Not only is his handiwork designed, he upholds […]

  • William Powers

    I will reply later when I get the chance.  It seems to me, however, that Meyer does not draw as sharp a distinction between “functional” specification and “meaning” specification.  Indeed, I believe he thinks that all information contains both.  It seems that I had independently come to the same conclusion, as I hope, my last post indicates.

    I would refer you to http://www.discovery.org/a/2177 for something that he has written on the subject.  It seems to me that functionally specified information will tend to be the important concept in any design inference since you don’t have immediate access to the meaningful aspects.


  • […] more from the original source: DNA Information and Computer Code | ASA Book Discussion Share and […]

  • Randy Isaac

    Yes, I do think you are right that Meyer does not draw as sharp a distinction though he does express the difference. I believe he has failed to grasp the significance of the distinction. He wants to lump it all together as having the same source.

    Let me restate again the concern I have. All of the examples that Meyer and the ID community have raised to demonstrate that complex specified information comes solely from intelligent agents are information systems that are heavily infused with abstraction. One cannot extrapolate from those examples to systems that are complex and specified but have no evidence of abstraction.

    You may well ask, what are the the examples of complex and specified information without abstraction or intelligent agent as a source? I gave some rather weak examples in several comments on this blog but in my next post I intend to give a much more elaborate example. Stay tuned.


  • William Powers

    Having read your latest post, I have a few comments.
    1) What we are about here is examining some entity and the properties of that entity.  Some of these properties we choose to call functions.  This choice is somewhat arbitrary.  It is not, I think, that one can say that F is a function, and another say that F is not a function.  I only mean to indicate that what we call a function is the result of a creative and intelligent process, and that what is the set of functions associated with an entity is by no means unique or probably even finite.  What we must (if we choose) be careful about is introducing metaphysical notions where do not intend to do so.  The assignment of function is one such opportunity.  It might even be said that notions of teleology are introduced with the assignment of function.
    In the broadest sense, what it seems we have is an entity that responds such and such a way to various antecedent conditions and what we might call behaviors, attributes or properties of the system.  To say that, for example, the function of DNA or the like is to “survive and reproduce” goes too far.  What one might say is that such and such capabilities enable the cell to reproduce.  But that the cell not be able to reproduce is also a kind of functioning.
    2) There appear to be two kinds of “functional specificity,” and, I suppose, one ought to be careful not to confuse the two.  The first is a property of a property, or, to use your language, a property of a particular function.  Given a function, define “working” as when the “function” or property is exhibited.  Working, then, is critically dependent upon what particular property or set of properties we intend.  For example, we could examine the function of “reproduction.”  How we define reproduction may vary.  What exactly is required of reproduction?  Must be it a perfect copy?  Do we permit errors?  If so, how many?  The list could go on and on.
    Keeping this ambiguity in mind, consider a function F (defined either in some fuzzy way or with certain variations).  Now consider how many possible ways that this function can be exhibited.  This is rather vague, but one can imagine confining it to constituents that are available.  This means we have to consider the resources available to produce the desired function.  They attempt to make such estimates for the functioning of certain anti-bodies.  The fewer the number of possible “working” realizations, the more specified is the “information” associated with that particular function.
    2) The second kind of “functional specification” is that associated with a mapping from one physical system to another (physical or non-physical) system.  The relationship between these two systems is in some sense non-causal and independent.  For example, you speak of a mapping from a two-state physical system to 0s and 1s; or, for that matter, to words in the English language, or colors of a pixel.  Such a mapping need not in its working take place in the presence of minds.  Mappings from binary states to printed words can be accomplished by a machine.
    You want to associate the first kind of functional specificity with DNA or cell processes, and the second with a computer.  I have and have always had two problems with this assignment.
    1) This difference between DNA and computers is only possible for you because you are an “insider.”  You know that computers are designed and you know this mapping.  DNA is a black box.
    2) Were you to land on the planet Xenon and find a computer and a cell, how would you judge that the computer is associated with an “abstraction” and the cell was not?  I assert that you cannot.  The only difference you would be able to tell about the computer is that it was a far simpler entity than the cell, and seemingly far more crude and clumsy.  Will we judge that because the computer cannot reproduce that it must be designed?
    I have suggested previously that what distinguishes the computer and the cell is that computer appears to require more of an explanation.  Why is this?  We think that the cell because it can ‘survive and reproduce’ somehow explains itself.  It’s vastly complex ‘functions’ and processes are explained on the basis of a ‘function’ that appears ‘natural’ to us: surviving and reproducing.
    The computer, on the other hand, can take certain voltages as inputs and generate output voltages.  Knowing arithmetic, we can independently specify things like addition, multiplication, “or” gates, and “nand” gates.  We can specify a logic and we can imagine that it might be used as a logic and computational machine.  (All of this very Dembski-like)  Indeed, we might think that independent of such things as arithmetic and logic that we cannot account for the existence of this computer at all.  We judge that it is, therefore, “not natural” and that since it’s existence and properties can “only” be accounted for in the context of “intelligent” processes, like arithmetic and logic, then it must be the product of “intelligence.”
    The presumption in this analysis is that “survival and reproduction” is a “natural” function, but arithmetic and “logic” is not.  Or perhaps, following the design inference, we say perhaps that “survival and reproduction” is associated with “law and chance,” while arithmetic and logic with intelligence.  But such presumptions appear to assume what is to be proved.  Once we assume that “survival and reproduction” are functions of a “natural” world can we escape concluding that such evidence is not a product of intelligence?  We  perhaps need to wonder whether “intelligence” is “natural” or not?  What is “natural”?  Dembski means by “natural” what the resources of “chance and physical law” can accomplish.  Some would affirm that “intelligence” is one of them.  If that is so, intelligence becomes a subcategory of the “natural.”  We might still be able to wonder whether a cell or computer are the products of intelligence or non-intelligence?  But the distinction would have to be different than that between “law and chance” and “intelligence.”
    One last point.  Just because we can imagine that a computer could be used as computational and logic machine, does not entail that it is one.  We can imagine using a log for a bridge, but that does not make the log a bridge.  Nonetheless, where we see logs strapped together and crossing a river we conclude that there is something peculiar about this arrangement.  Trees grow from roots in the ground.  Water flows in rivers.  All this is “familiar” and somehow judged “natural.”  But dead logs strapped together is not alive.  It is not something a tree would do.  It stands out to us against the “natural” living background as unexpected.
    To make myself clear.  I still think that to distinguish computers from cells cannot be done.  Either they are all “natural” and to be expected of our world, or something is odd is about both of them.
    Not knowing what else to say at this point, I’ll stop.
    Thanks for the conversation.  Too bad it seems that we are the only two participating.


  • Randy Isaac

    Quite honestly, Bill, you lost me. I got buried somewhere in there. But maybe I can shed light on many of your concerns by just focusing on your last point that you ” think that to distinguish computers from cells cannot be done.” Let me try to approach it from another angle.

    Instead of thinking about the design and construction of these two systems, think of the verification process, that is, confirming that the system meets the design specification. For the computer code, the hardware can be built and the software written and then we let the system run. Let’s take a simple example and submit input to the computer as “2+2”. Then we look at the output and we find “5”. The computer is happy, having performed all the hardware and software function as implemented. But we’re not happy because the relationship between the input and output is not what we intended. We go in, find the bug, fix it, and, voila, the output becomes “4”. We’re happy. Why? Because we can assess the abstract relationship between the output and the input and compare it with independently know relationships. That is a comparison that requires intelligence. We might be able to devise mechanical aids to help us, but the bottom line is that only an intelligent agent can make the assessment.

    For a living cell, there are many possible functions but let’s just consider the simplest–the ability to undergo cell division to form a clone. That’s a pretty complex function in all the details. When the process is complete, no abstract relationship is needed–either the cloned cell exists, perhaps with some minor modifications, or it doesn’t. It is a physical relationship.

    If the design verification requires abstraction, it follows that the design specification and the construction involve abstraction as well. I am suggesting that when a system involves abstraction at any level, an intelligent agent must be involved. If abstraction is not evident, one cannot conclude that intelligent agents must have been involved.

    Why is this important? Natural selection cannot operate on abstract relationships. It doesn’t know 2+2 is 4 and not 5. Intelligence is required. But when the specification is a physical result without any requirement of an abstract significance, no intelligence is necessary. Natural selection is a possibility. Such an example is coming in my next post.

    This, I suggest is the fundamental distinction between computers and cells. There may be others, but this distinction bears directly on whether or not an intelligent agent is required for verification or for design or for construction of the system.


  • David Wallace

    “This, I suggest is the fundamental distinction between computers and cells”
    Randy I have been wondering about an analog computer and where it would be classified in your thinking on abstraction.  In a general purpose analog computer, maybe the plug board could be considered an abstraction, I’m not sure although I have doubts.  However, in a special purpose analog computer the information is in the wiring of such a computer, for example in a flight simulator such as I worked on in the early 1960s.  Information in the configuration ie in the wiring seems very close conceptually to that in dna.
    Sure there was abstraction in the differential equations etc that made up a specific hard wired flight simulator but one could have built a hard wired analog computer to solve some arbitrary differential equation that interested one.  At that point I would have a hard time seeing the abstraction in such a device.
    I find abstraction a very slippery word and hard to pin down and think about.

    • Randy Isaac

        The issue is exactly the same for analog computers as for digital. The test if the computer is working is whether or not it solves the differential equation correctly. The magnitude of the voltage (or whatever parameter is used in the computere design) is correlated with a variable in the differential equation. That is the key abstraction.

  • William Powers

    I’m in a boat similar to David’s.
    Speaking of the computer, you imagine it has a function, e.g., add 2 and 2.  You find, somehow, that it doesn’t fulfill this function.  You fix it and now it function as intended.  All of these are intentional concepts.  When I said that you couldn’t really tell the difference between the cell and the computer, I was speaking as an outsider.  How do you know that this chunk of metal and wires is suppose to add numbers together?  All you have is a physical device.  If in my mind I map what it is doing to arithmetic, that is something I am doing as an intelligent agent.  It is like attempting to find a specification.  But there are many specifications, I could be wrong.  I am presuming, of course, that the object or behavior is correctly patterning a specification.  Could I tell that a broken computer was designed?  Perhaps.
    You say that the cell either divides “correctly” or it doesn’t.  You call this physical, but you don’t think adding 2+2 is physical.  I think you’re correct.  The computer is not adding 2+2.  It is doing something else, something else analogous, but vastly simpler, to cell division.  What the computer does and we interpret as adding 2+2 is just like cell division.  It either does it or it doesn’t.  Why can’t I do the same thing you are doing with a computer with a cell?  The cell when it replicates has to satisfy a large list of criteria that I can establish.  If it satisfies those criteria, I say it is “working.”  If it doesn’t, I say it is broke.  I can go in under the hood and fiddle around.  Perhaps there is a virus in the cell.  I can try to extract it or develop a means where the cell can extract it, and thereby get the cell to “work.”

    It seems to me that the computer requires “abstraction” because you use this machine in an abstract way.  I could probably use a cell in an abstract way.
    Here’s the problem.  You seem to be trying to draw this distinction because you want to be able to say that we can properly infer that the computer is designed, but that we cannot properly do so in the case of the cell.  However, it seems to me that you are always assuming what you are trying to prove.  If you are going to make any headway, it seems to me, you have to start with the same kind of knowledge.  You have to start with a “computer” that you don’t know whether it is designed or not, in the same way that you start with the cell.  You can’t therefore talk of humans in discussing the computer.  In fact, unless you mean something unconventional, I don’t think you can talk of abstractions at all, unless you try to get at it the same way that Dembski does through specification.  I see specification as simply trying to think the thoughts of a designer.  If you can posit a thought process, a logic, or some abstraction that (without being ad hoc) replicates some pattern or behavior, you think you are on the right track.  This can be done with a computer clearly.  I just don’t see why I can’t do it with a cell.  Just by stipulating a function, I am, in a sense, getting, or trying to get, into the head of a potential designer.  Designers think in terms of functions.  The function of cell division it seems to me is as specific, but vastly more complex, as adding 2+2.
    I’m not certain we’re getting any closer.  But I’ll await your next post.

    • Randy Isaac

        whether an outsider can detect the type of design is irrelevant to the actual type of information in the system. The fundamental criterion is whatever determines the specificity of the information in the system. The claim Meyer makes is that a complex system is not necessarily specified. To be specified, it must correlate with some specification. A random cell phone number is not specified. If it correlates with the phone number of someone you wish to call, it meets the criterion for specificity.
      It is the same with computers. What causes one to decide whether it is a system with specified information? Meyer says it can either match an independent pattern or it can carry out some desired function. My point is simply that if that pattern or function involves abstract concepts or relationships, we have an indication of an intelligent designer. If it involves no abstract concepts or relationships, then one cannot conclude that an intelligent designer was involved, though it might have been. The ability of a cell to create a clone of itself requires no abstract relationship. The function is specific, but not abstract.


  • […] that claim by citing a series of examples. One of those examples is computer code. In my previous post, I suggested that this was not an adequate example because of fundamental differences between […]

February 2010
« Jan   Mar »