Every Hand Is Improbable, Yet Casinos Catch Cheats

There are really only two moves in the sceptical playbook against the argument from improbability. The first says the argument commits a statistical howler: every outcome is equally improbable, so calling one outcome significant is something we do, not something we discover. The second says the argument destroys itself: if improbable things need designers, the designer needs one too. Between them these two moves carry nearly every popular takedown of the design inference, from Richard Dawkins's Ultimate 747 to the YouTube atheism that recycles it.

I've spent years testing both moves — against Stephen Woodford of Rationality Rules, whose video on his channel Debunked is the cleanest statement of the sceptical case I know; against Stephen Meyer of the Discovery Institute, who supplies the formal machinery the popular versions ignore; against the palaeontologist Günter Bechly and his sparring partner Erika of Gutsick Gibbon, who pressed the strongest objection anyone has put to me; and against Seth Hart, a Durham philosopher of science who is not an ID proponent at all. What follows is where the argument actually stands once you let every side speak at full strength.

One thing before we start. I'm not defending the argument as a Christian apologist would, though readers of Reasonable Faith will recognise most of the terrain. I'm a Gaudiya Vaishnava. Where William Lane Craig's argument terminates in the God of classical theism, mine terminates in the conclusion that mind is fundamental to reality — and my tradition names that fundamental mind Krishna, sarva-kāraṇa-kāraṇam, "the cause of all causes." Nothing in the inference itself depends on which tradition you ask afterwards. That matters for the second half of this article.

The debunking at full strength

Woodford's argument deserves to be stated properly, because in its own terms it is correct about something.

Take a fair six-sided die. The chance of rolling five sixes in a row is one in 7,776. But the chance of rolling any particular five-roll sequence is also one in 7,776, because each face is equally likely. So when the design proponent points at a low-probability outcome and declares it significant, Woodford says, the trick is happening in the pointing, not in the probability. The significance is assigned by the observer, arbitrarily.[^1]

He backs this with a lovely historical case. At the Monte Carlo casino in 1913, roulette gamblers lost millions of francs betting against black after black had come up 25 times running, reasoning that a 26th black had odds of one in 67 million. What they missed is that 25 blacks followed by a red has exactly the same odds. The gamblers, like the design proponents, had arbitrarily assigned significance to one sequence among equals.

Then he closes the trap with Carl Sagan's line: if the designer is allowed to exist without a designer — if God is allowed simply to have always existed — why not save a step and say the universe always existed? And behind Sagan stands Dawkins: if life without God is a tornado assembling a 747 in a junkyard, then God existing without a greater God is the Ultimate 747, the most improbable thing of all.

That's the case. It has a real insight in it — the equal-probability premise is true, and plenty of bad apologetics has ignored it. The problem is what Woodford builds on it.

Casinos catch cheats

In 2017 a Maryland casino called in the FBI over a baccarat player who had won 18 of 21 hands, including a straight run of 14 — in a game where the odds of winning any single hand sit just under 50 percent. The casino knew there had been cheating — what nobody yet knew was how. The mechanism only came out later. The inference to cheating came first, from the numbers alone.

Notice what the casino did not do. It did not reason: "every sequence of 21 hands is equally improbable, so this one is no more suspicious than any other." It categorised the outcome by its result — winning — and recognised a statistical outlier in a category that matters. If Woodford's objection were sound, that inference would be invalid, and no casino on earth could screen for fraud. As I put it in the video:

"While this is true, it ignores a very simple fact, that we can categorize events as significant based on their result. This is not arbitrary. Everyone knows there's a difference between winning or losing."[^2]

Lives are ruined and fortunes made over the difference between winning and losing. A category doesn't get more objective than that.

The same answer covers Monte Carlo. The gamblers' error wasn't that they noticed the streak — a 25-black run is genuinely a rare event, and a casino auditor would notice it too. Their error was inferring that the wheel now owed them a red, as if past spins reach forward and bend the next one. That's the gambler's fallacy, and it has nothing to do with the design inference, which claims no such influence. The design inference does what the Maryland casino did: it observes an outcome that is both wildly improbable and falls in an independently meaningful category — winning, functioning, meaning something — and asks what explains it.

So the equal-probability premise is true and irrelevant. The question was never whether the outcome is improbable. Every outcome is. The question is whether the outcome lands in a tiny, objectively special region of the possibility space.

Shannon versus specified information

This is exactly the distinction Stephen Meyer's work formalises, and it's why the sophisticated argument was never about bare improbability.

Shannon information measures how much information a string of characters can carry, given the number of possible combinations. By that measure, a passage of Shakespeare and an equal-length string produced by a monkey on a typewriter are identical. Shannon's mathematics cannot tell them apart, and nobody claims it can. To say what separates Shakespeare from gibberish you need what Meyer calls specified information: the vanishingly small subset of combinations that achieve function or meaning. Every combination of letters is improbable. Only some spell words. Fewer make sentences. Fewer still are worth reading.

DNA is where this bites. The sceptic's own framing concedes that sequences coding for functional proteins are an infinitesimal island in an ocean of gibberish sequences. The difference between a protein that folds and functions and a string that does neither is not assigned by an observer. It is about as objective as the difference between winning and losing fourteen hands straight.

Meyer has a homely illustration of how the probabilistic reasoning then works. A thief confronts a bicycle lock with four dials — ten thousand combinations. Trying one every ten seconds, he samples over five thousand combinations in fifteen hours, and success becomes more likely than not. Now give the lock ten dials: ten billion combinations. In a hundred years of continuous trying the thief samples about three percent of them. Failure is now overwhelmingly likely, and any hypothesis that says "he'll crack it by chance" is more likely false than true.

"In science, we really prefer not to affirm hypotheses that we know are more likely to be false than true. We go on and look for a better explanation."[^3]

That is the whole argument, in its modest, scientific form. Not "improbable, therefore God." Rather: when a chance hypothesis can sample only a vanishing fraction of the space, and the outcome sits in an objectively specified region of that space, chance is a hypothesis more likely false than true — and science goes looking for a better one.

If probabilities can't infer design, fire the archaeologists

Here's the question I keep putting to sceptics, and I've yet to hear a good answer:

"If we can't use probabilities to infer an intelligent designer, then why can archaeologists do it?"[^4]

When archaeologists find scratchings on a cave wall, they infer carvers they have never seen and never will. The scratchings could in principle be the work of water and wind — nothing logically rules it out — but the pattern sits in a region of possibility space that unguided erosion essentially never visits and intelligent agents visit all the time. That is a probabilistic design inference, made routinely, published in respectable journals, and nobody calls it pseudoscience.

Seth Hart — who, remember, is not an ID advocate — made the same point on my show from the philosophy-of-science side: design inference is already standard scientific practice in SETI, archaeology and criminal investigation, so the inference form used by ID is scientific in kind.[^5] A radio telescope that picked up a signal saying hello would not be told that every electromagnetic pattern is equally improbable. Hart's own verdict on intelligent design is worth quoting precisely because he doesn't share my conclusion:

"That's why I end up saying, yeah, ID theory is scientific. It's just not the best science. It's just not where I think the evidence leads."[^6]

You can hold that position. What you cannot coherently hold is Woodford's: that probabilistic inference to intelligence is invalid in principle. That position doesn't just block the road to God. It fires the archaeologists, shuts down SETI, and empties the FBI's fraud division.

And there's a telling concession buried in Woodford's own video. He quotes Fred Hoyle's famous line — that life arising by chance is like a tornado sweeping through a junkyard and assembling a Boeing 747 — and never disputes it. He doesn't argue that abiogenesis by natural processes is probable. His entire case is the meta-claim that improbability can't license the inference. Once that meta-claim falls, he's left having tacitly granted the improbability itself.

The strongest objection: a designer nobody has met

The best counter-argument I've encountered didn't come from Woodford. It came from Erika of Gutsick Gibbon, in her marathon debate with Günter Bechly on my channel, and it deserves its own section rather than a parenthesis.

Erika's point is that the design inferences everyone accepts — stone tools, cave carvings, cheating gamblers — rest on designers we independently know to exist and whose methods we can observe and mimic. We know there were humans; we can knap flint ourselves. Take that background knowledge away and the inference loses its licence. Her thought experiment is a rock arch on Mars:

"There's no designers that we know of that live on Mars. So I would propose that it's most parsimonious to move with this idea of a natural arch, unless you can provide support that there is kind of a present designer."[^7]

She adds a historical jab with real force: candidate alien signals have, so far, always dissolved into natural explanations on further study. And her general verdict on ID is as blunt as Hart's is irenic — she told us it "relies almost entirely on attempting to put holes in evolution" and can't yet be taken seriously as a scientific hypothesis.[^14] This is the objection at full strength: not that design inference is invalid, but that it's parasitic on prior acquaintance with the designer.

I think it fails, and the reason is the difference between an arch and an inscription.

An arch is improbable but unspecified. Wind and water carve arches; there's no independently given pattern the Martian arch matches, so the natural explanation wins, and Bechly and I would agree with Erika about the arch. But change the example. Suppose the Mars rover finds "John loves Mary" written in the sand, to borrow the case Bechly put to her. Or suppose a telescope receives a signal ticking out the sequence of prime numbers. We know nothing about the sender. We have never met a Martian, cannot mimic one, and have no theory of its methods or motives. Yet every scientist alive would infer intelligence, because the pattern matches something independently given — arithmetic, language — that unguided processes have no route to. The design inference runs on the specification, not on a prior census of available designers, and it stays agnostic about the designer's identity, exactly as SETI does.[^8]

That's also why Erika's dissolved-signals point doesn't cut the way she wants. Pulsing radio sources were reclassified as natural the moment a natural pattern-generator was found — which shows the filter working, not failing. The inference was withdrawn when the specification turned out to be reachable by unguided processes. In DNA's case, the whole debate exists precisely because no one has exhibited such a process for the specified information in the cell.

And if the known-designer requirement were sound, it would prove far too much. It would invalidate SETI, which scans for designers nobody has met by definition. It would invalidate the simulation hypothesis, which a good number of naturalists entertain over their coffee, and which posits a designer outside our universe entirely. And it would gut the fine-tuning discussion, where the pattern — physical constants sitting in the narrow band that permits complex chemistry — is as independently given as patterns come. The requirement is not a rule anyone applies consistently. It's a rule applied once, at the door where God might come in.

Who designed the designer? Gravity has no explanation either

So we come to the second move: Sagan's save-a-step and Dawkins's Ultimate 747. If God always existed, why not just say the universe did? And isn't a being capable of designing life the most improbable thing of all?

The save-a-step argument assumes the two stopping points are equivalent — that halting at "the universe" and halting at "God" are the same move made at different stations. They aren't. We don't stop at the universe because we have evidence inside the universe pointing to something outside it: the specified information in life, and — as Meyer argues — the beginning of the cosmos itself and the fine-tuning of its constants. A crime-scene investigator who says "something's going on here beyond accident" isn't multiplying entities; she's following evidence. Stopping at the universe isn't saving a step. It's refusing to take a step the evidence is paying for.

Behind Sagan's line sits an assumption almost nobody examines: that every explanation must itself be explained before it can explain anything. Applied consistently, that principle doesn't trouble theism — it abolishes science.

"Gravity is used to explain the movements of celestial bodies, but gravity itself has no explanation. But this doesn't stop gravity from being a valid scientific hypothesis."[^9]

We can measure gravity and map its effects with exquisite precision, but at the fundamental level nobody can tell you why mass attracts mass. If "no explanation for the explainer" were disqualifying, Newton's whole edifice would have been pseudoscience for centuries. Every chain of explanation terminates somewhere. The only live question is what sits at the bottom.

Which brings us to the deepest error in the Ultimate 747. Dawkins's argument works only if God is the kind of thing the tornado-in-a-junkyard argument applies to: a complex object assembled from parts, whose configuration is one improbable arrangement among astronomical alternatives. But that is precisely what no serious theist claims. Craig's version says God is a simple, disembodied mind with no moving parts. My version goes further:

"What's being claimed here is that the fundamental element of reality, the thing on which everything else sits, is a mind. With intentionality, with the ability to conceive of ideas and manifest them in reality."[^10]

A fundamental mind never enters the improbability lottery, because the lottery is only for things assembled out of pre-existing components under pre-existing laws. Mind, on this view, is the thing the components and laws sit on. In my tradition that fundamental mind is a person — Krishna, the cause of all causes — but you don't need Vaishnava theology to see the logical point. The objector who insists the designer must be a complex assembled object is not refuting the hypothesis on the table. He is refusing to read it. If he wants to argue that mind cannot be fundamental — that matter must sit at the bottom — he's welcome to make that argument. Presupposing it, inside an objection, is the actual special pleading in this debate.

The parsimony scoreboard

That leaves the materialist with a primitive of his own, because every worldview has one — some ultimate thing that explains without being explained. The materialist's candidate is eternally self-existent matter-energy. And here the evidence has been unkind, as Meyer put it on my show:

"We actually have evidence that the physical world of matter, space, time, and energy began a finite time ago, and therefore that materialism is a crummy candidate to be the thing from which everything else came."[^11]

A primitive that begins isn't a primitive. So the serious naturalist reaches for the multiverse: an eternal ensemble of universes, varied enough that some lottery winner like ours turns up. Set aside whether that even touches specified biological information, and just run the ledger Occam actually asks for. To get a working multiverse you need universe-generating machinery — inflaton fields, extra dimensions, the apparatus of string cosmology — which Meyer counts at around ten postulated entities, none observed, with the fine-tuning of the generator itself still unexplained behind them.

"You have 10 pure theoretical postulates for which we can have no direct knowledge or empirical support versus the single hypothesis of a transcendent designer."[^12]

Ten unobservable posits against one. The stock reply is Dawkins's: God can't be the simple option, because anything that designs must be complex. But that equivocates on what Occam's razor measures. Parsimony counts the entities and assumptions a hypothesis postulates, not the richness of the entity's inner life. One fundamental mind is one postulate. And unlike the multiverse's machinery, this postulate is of a kind we already know exists and already know produces the effect in question:

"Mind is a reality. It has real powers that we know that exceed the powers of matter and energy. It produces distinctive effects that allow us to detect its activity."[^13]

Each of us has direct acquaintance with a mind conceiving ideas and imposing them on matter — better acquaintance, arguably, than we have with anything else. Specified information is its signature product. We have never observed an inflaton field. We have never observed unguided chemistry writing code. We observe minds doing it every day.

So run the two moves one last time. "Every outcome is equally improbable" is true and beside the point: the design inference runs on specified outcomes, the same way casino fraud detection, archaeology and SETI do, and it no more depends on having met the designer than a prime-number signal from deep space would. "Who designed the designer" refutes only a designer nobody proposed — a complex assembled god — while leaving the actual hypothesis, fundamental mind, standing exactly where the ledger says it should: one postulate, of a kind we know exists, with the known causal power to produce the one thing that needs explaining. The argument from improbability hasn't been debunked. Its critics have been aiming at the wrong argument all along.

[^1]: Has the Argument From Improbability been Debunked @rationality rules (responding to Stephen Woodford), 00:00:54 — https://www.youtube.com/watch?v=hodDd0oKmlw&t=54s [^2]: Has the Argument From Improbability been Debunked @rationality rules, 00:02:53 — https://www.youtube.com/watch?v=hodDd0oKmlw&t=173s [^3]: Has the Argument From Improbability been Debunked @rationality rules (Stephen Meyer clip), 00:05:57 — https://www.youtube.com/watch?v=hodDd0oKmlw&t=357s [^4]: Has the Argument From Improbability been Debunked @rationality rules, 00:09:39 — https://www.youtube.com/watch?v=hodDd0oKmlw&t=579s [^5]: Is Intelligent Design Pseudoscience? How is Science Defined (with Seth Hart), 01:14:20 — https://www.youtube.com/watch?v=YaiVIdzlFs4&t=4460s [^6]: Is Intelligent Design Pseudoscience? How is Science Defined (with Seth Hart), 01:20:01 — https://www.youtube.com/watch?v=YaiVIdzlFs4&t=4801s [^7]: The Fossil Record, Evolution, and Intelligent Design — Erika & Günter Bechly, 01:12:58 — https://www.youtube.com/watch?v=sJsLCSwHsz4&t=4378s [^8]: The Fossil Record, Evolution, and Intelligent Design — Erika & Günter Bechly (Günter Bechly), 01:22:16 — https://www.youtube.com/watch?v=sJsLCSwHsz4&t=4936s [^9]: Has the Argument From Improbability been Debunked @rationality rules, 00:11:48 — https://www.youtube.com/watch?v=hodDd0oKmlw&t=708s [^10]: Has the Argument From Improbability been Debunked @rationality rules, 00:14:33 — https://www.youtube.com/watch?v=hodDd0oKmlw&t=873s [^11]: Answering Claims That ID is Pseudoscience — Stephen Meyer & Seth Hart on the Demarcation Problem (Stephen Meyer), 00:44:21 — https://www.youtube.com/watch?v=DmquKfHUjdk&t=2661s [^12]: Answering Claims That ID is Pseudoscience — Stephen Meyer & Seth Hart on the Demarcation Problem (Stephen Meyer), 00:50:05 — https://www.youtube.com/watch?v=DmquKfHUjdk&t=3005s [^13]: Answering Claims That ID is Pseudoscience — Stephen Meyer & Seth Hart on the Demarcation Problem (Stephen Meyer), 01:15:13 — https://www.youtube.com/watch?v=DmquKfHUjdk&t=4513s [^14]: The Fossil Record, Evolution, and Intelligent Design — Erika & Günter Bechly (Erika), 00:17:31 — https://www.youtube.com/watch?v=sJsLCSwHsz4&t=1051s