The Atheist’s Riddle

“Perry Marshall presents himself as an invincible defender of his supposed proof of an Intelligent Designer, standing atop a mountain of vanquished counter-arguments from hordes of atheists.”

Argument taken directly from Cosmic Fingerprints:
1) DNA is not merely a molecule with a pattern; it is a code, a language, and an information storage mechanism.
2) All codes are created by a conscious mind; there is no natural process known to science that creates coded information.
3) Therefore DNA was designed by a mind.

Perry Marshall presents himself as an invincible defender of his supposed proof of an Intelligent Designer, standing atop a mountain of vanquished counter-arguments from hordes of atheists.

The plain logical error in the argument is in the second premise, and it’s the one logical fallacy I come across more than any other: an argument from ignorance. “There is no natural process known to science that creates coded information.” That’s not the same as saying there really is no such natural process (which would be a simple unsupported statement rather than a fallacy), but it expects us to assume as much. Is Mr Marshall, or any human alive, familiar with “all codes” in the universe? What qualifies anyone to make such a sweeping statement? This attempted proof by elimination of the origin of DNA must leave room for unknown alternatives to maintain any honesty, and is therefore not a real proof.

I realise that the fact of the logical error is not such a brilliant counter-argument when you’re actually trying to convince people. There are plenty more objections, and Marshall has posted and replied to many on his site. He hasn’t always done so convincingly, though you can judge that for yourself. I’ll just take one approach as an exercise.

As support for the argument that all codes are designed by a mind, Marshall argues that random processes do not produce information. (I’ve been through this at length, years ago.) His primary demonstration is his own text-based random mutation generator which takes a sentence and, through single-letter changes, turns it to nonsense.

Marshall admits that the mutation utility does not simulate natural selection, the non-random element of evolution. Furthermore, he’s not interested in adding that functionality to test his own argument. (He says instead that the reader is free to do it for him. Several people have, beginning thirty years ago with Richard Dawkins’ “Methinks it is like a weasel” program and continuing with browser-friendly programs like Mutate.)

He argues that natural selection would only create sensible sentences if words only mutated into other meaningful words, but that’s not applying natural selection at the letter level. An ideal extension of his program would present several choices of mutation at each step, and allow those letter mutations which destroy the legibility of a word to be manually or automatically ruled out. (The real world equivalent is a serious birth defect, which would keep a creature from breeding or even living long enough to breed.) In Marshall’s program, detrimental mutations are allowed to compound until all sense is lost. Of course we won’t likely get anything useful out of it.

Forgetting even the mechanism of natural selection, I submit a basic argument for the possibility of chance creating information which I’ve used before: think of a large grid of squares which can be either black or white, but all start as white. If you randomly pick the colour of every square at once, there is a chance, however small, that the newly black squares will form a simple but clear picture of a rectangle, or the letter G, or Elvis. Without adding any extra material, chance can increase the amount of information the grid provides. The prebiotic chemicals only had to manage a feat like this once, given potentially unlimited opportunities, to come up with DNA or its precursors.


9 thoughts on “The Atheist’s Riddle”

  1. (Comment by Aaron)
    If you’re going to assert that premise #2, ‘All codes are created by a conscious mind; there is no natural process known to science that creates coded information.’ is essentially an argument from ignorance, then how could an atheist say that the statement ‘god does not exist; there is nothing in the known universe that cannot be explained through natural processes’ does not contain the same fallacy?

    I haven’t spent much time examining Marshall’s argument but so far haven’t been able to see anything obviously incorrect. I don’t know enough about information theory and the like to be able to assess the soundness of Marshall’s argument. As an atheist, I’d love to have a good reply to the ‘dna comes from a mind’ argument.

  2. If an atheist claimed the certainty which is required to say “God does not exist” or “there is nothing which cannot be explained naturally”, without qualifying either, I would challenge my fellow atheist.

    Most atheists however do not claim certainty. The more common positions are based firmly on things we can be more confident about, like, “There is no known substantive, unambiguous evidence for a god, which is itself evident because there are still non-believers everywhere, and is significant because most gods hypothetically have great influence and also desire belief.”

    Instead of your second mock quote, try this: “No real, established phenomenon has been found to be impossible by natural means so far, therefore the supernatural is not yet required to explain anything which exists, and even if it were it wouldn’t necessarily be a god, let alone a particular god.”

    In short, gods are unlikely based on what we know, though perhaps not impossible.

    That’s irrelevant to the Atheist’s Riddle, though. Even if atheists were all in the habit of making arguments from ignorance equivalent to that in premise #2, it wouldn’t mean that premise #2 doesn’t contain an argument from ignorance. The Riddle is supposed to be a sound proof of God. It isn’t that, even if all atheists are intellectually lazy and dishonest hypocrites, which fortunately they aren’t.

    The whole point of the Riddle is that it’s designed by Marshall so that nothing about it is obviously incorrect. Its real power is in the assumptions one tends to make between the lines, accepting the argument from ignorance: “there is no natural process known to science that creates coded information” …therefore there are none at all, and the only alternative explanation is a mind. That’s two leaps too many, but they’re the leaps people make.

    There’s a huge amount of discussion about the definition of “code” between the two sides of this argument, focused on whether DNA is one and whether other natural phenomena are codes as well. I have no trouble with accepting the idea that DNA, whatever it is, is unique on this planet in its qualities and potential. We’ve always known that life on Earth is unique, because the DNA itself tells us it’s all related and there’s nothing else powering alternative forms of life that we know about.

    The core thing people get wrong is to assume that because something has apparently only happened once, it is impossible by natural means. DNA emerged from something which wasn’t DNA, which is an unlikely thing to happen at any given time and place, but had countless opportunities everywhere on Earth in its first billion years and therefore may not be so unlikely overall. It is after all quite simple in principle, with only four key components or “letters” made of four chemicals.

  3. I think I understand it more clearly now. It comes down to the fact that the god hypothesis is a big leap. The god hypothesis is far less likely than the hypothesis that DNA is a code that arose from natural processes.

    I’ll come back to this later. I’m pretty tired at the moment.

  4. Remember that it doesn’t seem like a leap at all to people who’ve already accepted the god hypothesis.

    You don’t even need to go as far as comparing probabilities anyway. According to the Atheist’s Riddle, the natural origin of DNA is not just unlikely but impossible, necessitating a god. The reply to this is simply that the best evidence so far suggests that it’s possible.

  5. I’m a theist, but Marshall’s argument frustrated me for many of the same reasons. This helped me in my thinking about it, but I’m confused about your last paragraph. I’m only a college student, and not even a student of science so I’m having trouble knowing which information (about DNA, patterns/design, matter/energy/information) on both sides of the argument is reliable.
    Here’s my issue. You said that depending on the placement of color on the grid, the information will be more or less. But information requires an interpreter, someone who places value on the information. An arrangement of blocks that I have never seen before may be familiar and informative to someone from another culture. To me there is no information while to this other person there is at least some.
    Can information exist without energy and matter?
    I hope that was coherent. Thanks for any clarification you can give me.

  6. That was perfectly coherent, and a good question. For full disclosure my science degree was in mathematics and computer science, though I learned most of what’s relevant to this topic after I’d finished.

    The short answer is that information does need matter and/or energy to exist as far as we know, but it does not need an interpreter or even an observer (notwithstanding the philosophical issue of whether matter itself exists without observation).

    I’ll elaborate on these points using the grid example.

    Firstly, the information is contained in and conveyed by the squares of the grid, and would be meaningless at best without them, or an image or conceptual model of them. The information has to be stored somewhere, and apply to something.

    To demonstrate that the information is independent of interpretation, there are examples of information which is unmistakable. Consider two states of the grid: one where every square is white, and one where every square is white except for one 3×3 black square right in the middle. Whichever culture you might be from, you’re going to notice a difference between those two states. You might see it as the difference between “something” and “nothing”, or “blank” and “spot”, or “pure” and “blemished”. Interpretation only serves to contextualise the information which the grid has undeniably provided to you.

    To go farther and demonstrate that information is independent even of observation, there are objective ways to measure an amount of information or complexity, which only require knowledge of what the relevant information is. Kolmogorov complexity is one example of a calculable quantity, normally applied to a string of letters.

    The following is NOT a Kolmogorov calculation, but it will give you a general idea of how the information on the grid can increase or decrease. Imagine writing instructions to replicate a given pattern on a 5×5 grid which was first randomly generated. Here’s one possible case:

    1. Set all 25 squares to white.


    That’s probably the simplest, together with the equivalent for “black”. Now here’s another case:

    1. Set the top row of squares to white.
    2. Set the 2nd and 4th squares from the left in the second row from the top to white and the others in the row to black.
    3. Set the 1st square from the left in the middle row to white and the rest to black.
    4. Set the 1st, 2nd and 5th squares from the left in the second row from the bottom to white, and the rest to black.
    5. Set the squares in the bottom row with white squares directly above them to black, and the rest to white.


    If you’re clever, you might be able to write a simpler set of instructions for the second pattern, but you’ll never achieve the second pattern with instructions as simple as those you can use for the first pattern. There’s just too much more raw information you need to cram into them.

    It’s rare, about one chance in about 33 million, that randomly selected colours will go to all white, but it’s equally rare that they’ll produce any other exact pattern. When all ~33 million possible patterns are considered, it’s also more likely that a random pattern will need complex instructions to replicate than only very simple instructions. So not only is it possible for random action to increase complexity, but we would actually expect it to in this case (remember, the grid starts all white).

    To bring the grid analogy closer to the case of living things, the size of the grid and even the number of possible colours would have to increase as time went on. At first glance you would expect this to produce an unintelligible mess before long, but remember you’ve got natural selection happening all along the way. Any random pattern which has some use or function survives, while the rest are gradually culled through competition and outright failure.

    Anyway, that’s getting beyond the scope of the grid’s original purpose: to demonstrate simply that information can increase as a result of randomisation. It can and does happen.

Comments are closed.