If you've spent any time debating prophecy fulfillment with Christian apologists, you've encountered the number. It shows up in blog posts, YouTube videos, church sermons, and apologetics courses with the regularity of a liturgical response. "The probability of one person fulfilling just eight messianic prophecies is 1 in 10¹⁷!" They say it with the confidence of someone citing the speed of light. (That is the popular version. The original claim is subtler, and I'll give it in its author's own words below.) And why wouldn't they? It comes wrapped in the language of mathematics, which gives it an aura of precision that theological arguments rarely enjoy.
The number traces back to Peter Stoner, chairman of the departments of mathematics and astronomy at Pasadena City College until 1953, who published his calculation in Science Speaks (first published in 1952 and revised as late as 1976; an earlier form appeared in 1944 as From Science to Souls). Josh McDowell then popularized it in Evidence That Demands a Verdict (1972; revised as The New Evidence That Demands a Verdict, 1999), one of the most widely read works of popular apologetics of the last half-century. McDowell reprints Stoner's figures along with the endorsement of the American Scientific Affiliation, under the claim that coincidence "is ruled out by the science of probability." It is not. It is a set of guesses dressed in scientific notation, and it falls apart the moment you ask what each of its numbers is actually measuring.
I'm not going to just assert that the number is wrong. I'm going to let Stoner make his case first, in his own words and at its strongest, including the modern revision his grandson published online. Then I'll take it apart through three lenses in turn: the probability theory Stoner himself appeals to; a literary and redaction-critical look at the Gospel texts, which are the only witnesses to the "fulfillments"; and a historical and philological reading of the prophecies in their own settings. Reception history gets the last word, because the most striking recent verdict on Science Speaks comes from the grandson who keeps it online. The short version: every factor in the calculation prices an event whose only evidence is a narrative written to show that Scripture was fulfilled, and no probability survives that.
The Argument at Its Strongest, in Stoner's Own Words
Stoner's method is simple to state. For each prophecy he asks, in his words, "One man in how many men has fulfilled this prophecy?" He anticipates the pious objection that no one but Christ could fulfill them, and answers it fairly: individual prophecies can be matched by other people. "It was prophesied that Christ would be crucified; other men have been crucified."
The estimates did not come from nowhere, and a critic who says they did is attacking a weaker book than the one Stoner wrote. They began in a class on Christian evidences sponsored by the Inter-Varsity Christian Fellowship at Pasadena City College, whose students discussed each prophecy at length and agreed on a figure. Stoner then combined these with "estimates given me later by some twelve different classes of college students, representing more than 600 students." He adds, "I have carefully weighed the estimates and have changed some to make them more conservative," and invites the doubter to do the same: "If the reader does not agree with the estimates given, he may make his own estimates and then carry them through to their logical conclusions."
Here are his eight, with the students' figure and the one he used:
| # |
Text |
Stoner's question (paraphrased) |
Students |
Used |
| 1 |
Micah 5:2 |
Born in Bethlehem |
1 in 280,000 |
1 in 2.8 × 10⁵ |
| 2 |
Malachi 3:1 |
Had a forerunner to prepare his way |
1 in 10³ |
1 in 10³ |
| 3 |
Zechariah 9:9 |
Entered Jerusalem as a ruler, riding a colt |
1 in 10⁴ |
1 in 10² |
| 4 |
Zechariah 13:6 |
Betrayed by a friend, the betrayal leading to wounds in his hands |
1 in 10³ |
1 in 10³ |
| 5 |
Zechariah 11:12 |
Of the betrayed, betrayed for exactly thirty pieces of silver |
1 in 10⁴ |
1 in 10³ |
| 6 |
Zechariah 11:13 |
The money returned, refused, thrown down in the house of the Lord, used to buy a field from the potter |
1 in 10⁵ |
1 in 10⁵ |
| 7 |
Isaiah 53:7 |
On trial for his life though innocent, made no defense |
1 in 10⁴ |
1 in 10³ |
| 8 |
Psalm 22:16 |
Crucified |
1 in 10⁴ |
1 in 10⁴ |
Multiplied together, these give 1 in 2.8 × 10²⁸, which Stoner rounds to 10²⁸. Then comes the step that popularizers usually skip. Stoner points out that 10²⁸ answers the question, "One man in how many men has fulfilled these eight prophecies?", whereas the question that matters is the chance "that any man might have lived from the day of these prophecies down to the present time" and fulfilled them all. So he divides by his estimate of everyone who has ever lived, "about 88 billion," rounded up to 10¹¹, and concludes that "the chance that any man might have lived down to the present time and fulfilled all eight prophecies is 1 in 10¹⁷." (An editor's note in the online edition suggests the population figure may be ten times too large, which would make it 10¹⁸.) So 10¹⁷ is not the chance that one named person matches eight prophecies. It is Stoner's estimate of the chance that anyone in history would do so by accident. He has already corrected for the number of people who might have fulfilled them, and McDowell, to his credit, reproduces the "any man" wording correctly.
Stoner then extends the argument. Sixteen prophecies give 1 in 10⁴⁵. Forty-eight, "similar in their human chance of fulfillment to the eight which we originally considered, using a much more conservative number, 1 in 10²¹," give 1 in 10¹⁵⁷, a figure McDowell also reprints. Two small observations, both my own: the forty-eight are hypothetical, since Stoner never evaluates them one by one; and 10¹⁵⁷ is exactly what you get from the original 10¹⁷ times five more sets of eight at the old 10²⁸ each, so the "more conservative" 10²¹ never enters the sum. Then comes the argument meant to close every door: "More than three hundred prophecies from the Old Testament which deal with the first advent of Christ have been listed. Every one of them was completely fulfilled by Jesus Christ." If so, "you could set your estimates ridiculously low on the whole three hundred prophecies and still obtain tremendous evidence of inspiration." Put every one at one in four, he says, down to "even that one in four rose from the dead on the third day," and you still beat the forty-eight. On the arithmetic he is right: three hundred factors of one in four multiply to roughly 1 in 10¹⁸⁰.
The strongest modern form of the argument is not Peter Stoner's but his grandson's. The online edition, revised by Don W. Stoner in November 2005, adds Appendix 3, "A Challenge to Critics of Chapter 3." It names the two standard objections head-on: that the estimates may be too high, "or that the events are statistically connected - so that fulfilling one prophecy will virtually guarantee fulfillment of another." It then reruns the numbers to show "what sort of numbers a person playing the devil's advocate might obtain." Don Stoner restricts the candidates to males born in Bethlehem within a 107-year window derived from Daniel 9:25, generously rounded to 25,000 men. He separates what a conspiracy could stage from what it could not. Of the colt, "we might expect the perpetrators to deliberately set this one up," whereas "Betrayal is, by definition, not an element subject to advance planning." After ten prophecies and deliberately generous odds, he still arrives at "less than one in a million," and concludes: "We must still conclude the presence of divine inspiration."
That is a serious argument, and it deserves a serious answer. But notice what it answers. Every refinement in it defends against a conspiracy: people acting out the prophecies. None of it defends against a narrator.
Where the Math Actually Dies: Conditioning and a Double Count
Start with the probability theory, on Stoner's own terms. His calculation depends on getting each factor's conditional probability right. It needs the chance of each event given the others, and for most of his list that is where the math dies.
Critics usually put this as an independence objection. If you flip a fair coin, getting heads on the first flip doesn't change the probability of getting heads on the second flip. Each flip is independent, so you can multiply probabilities: the chance of two heads in a row is 1/2 × 1/2 = 1/4. Prophecy fulfillments, the critic says, are not like coin flips, so Stoner can't multiply. As mathematics, that is wrong. The chain rule, P(A and B) = P(A) × P(B given A), multiplies dependent probabilities perfectly validly, provided each factor is the chance of that event given everything before it. Stoner knows this. He words several of his estimates as conditional questions: "Of the people who have been betrayed, one in how many has been betrayed for exactly thirty pieces of silver?" Where he drops the conditioning, he says why. Of his fourth prophecy he writes, "There seems to be no relation between the fulfillment of this prophecy and those which we have previously considered."
So the real question is whether his conditional probabilities are the right ones, and in three places they are not.
- One event is priced twice. His fourth prophecy asks, "One man in how many, the world over, has been betrayed by a friend, and that betrayal has resulted in his being wounded in his hands?" His gloss is that the betrayal led to Christ being "put to death, wounds being made in His hands." His eighth asks how many men have been crucified. But the wounds in the hands are the crucifixion. Once the fourth factor is granted, the eighth is close to certain, yet it enters the product again at 1 in 10⁴, on top of a 1 in 10³ that already includes those wounds. You need not take my word for it. McDowell prints cross-references beside the calculation, keying Stoner's fourth prophecy to items 33 and 44 of his own list and the eighth to items 44 and 45. Item 44 is "Hands and Feet Pierced," so the same entry sits under both. And Don Stoner's Appendix 3 merges the two, without comment, into a single step numbered 5 and 5b.
- One passage is split in two. The fifth and sixth prophecies are two consecutive verses, Zechariah 11:12-13, describing a single sign-act, and in Matthew they are a single episode (Matthew 27:3-10). Together they contribute 10⁸ of the 10²⁸.
- The conditioning is dropped where it matters. On the forerunner, Stoner says that "the question can just as well be general"; on the colt, the properly conditioned question "becomes so restrictive" that he substitutes "an equivalent question." But a man already being hailed as the promised king is not a random man. The chance that his followers would give him a herald and a royal entry is far higher than the world-wide base rate, as Don Stoner himself concedes for the colt.
And for the biggest dependency of all, the fact that the people telling the story knew the prophecies, Stoner has no factor at all.
The Deeper Problem: Who Reported the Fulfillment?
Now switch lenses, from probability to the literary and redaction-critical study of the Gospels. The fulfillment events are not independent of the texts. Every factor in Stoner's product prices an event we know about only because a Gospel says it happened, and the Gospel authors knew the prophecies they were writing about. This is a deeper problem than independence. It is not about how the events relate to one another. It is about the evidence: how likely is it that a fulfillment would be reported, given that the person reporting it knew the prophecy? That question undercuts every factor at once, and it would do so even if there were only one.
Matthew is the most transparent case, because he shows his working. He uses the formula "that it might be fulfilled which was spoken through the prophet" (World English Bible) at least ten times by scholarly count. He structures entire episodes around Old Testament passages. The birth narrative in Matthew 1-2 is a case study in literary construction: the virgin birth mapped onto Isaiah 7:14 (which isn't even a messianic prophecy in context, as I argue in the Septuagint problem), the Bethlehem birth mapped onto Micah 5:2, the flight to Egypt mapped onto Hosea 11:1, the slaughter of the innocents mapped onto Jeremiah 31:15. Raymond Brown's The Birth of the Messiah, the standard critical commentary on the infancy narratives, reads Matthew as having fitted his formula citations to stories that were already being told, stories already shaped by the pattern of Moses' birth (pp. 100, 104, 108). On that reading, Scripture shaped how the story was told and what was made of it, so the story cannot be treated as an independent historical report that happens to match.
Take Bethlehem. If the people telling the story already believed, as John 7:42 shows some did, that "the Christ comes of the offspring of David, and from Bethlehem," then a Bethlehem birth is exactly what they would report. Luke has Bethlehem too, by a different route and without citing Micah (Luke 2:4-7), so the claim is older than either Gospel. But that only pushes the question back to the believers who first told the story. The "probability" of a Bethlehem report is not a demographic ratio. It is a function of what the tellers believed a Messiah had to be.
Now apply the same question to the two factors that carry the most weight after Bethlehem. The thirty pieces of silver appear in only one Gospel. Mark says only that the priests "promised to give him money" (Mark 14:11). Matthew alone has "They weighed out for him thirty pieces of silver" (Matthew 26:15), echoing Zechariah's "they weighed for my wages thirty pieces of silver." The potter's field is Matthew's alone as well, and he credits the text to "Jeremiah the prophet" (Matthew 27:9). Acts tells a different story, in which Judas himself "obtained a field with the reward for his wickedness" (Acts 1:18). Those two factors are 10³ × 10⁵ = 10⁸ of Stoner's 10²⁸. The probability that a writer who knew Zechariah, and who was writing to show that Scripture had been fulfilled, would report a fulfillment of Zechariah is not 1 in 10⁸. It is close to 1.
The same point disposes of the three-hundred-prophecy fallback. That move works only if each factor is genuinely a chance event, even at one in four. For a fulfillment the narrator wrote to match the text, the right number isn't one in four. It is close to one, and factors close to one barely move the product: three hundred of them at 0.99 each multiply to about 1 in 20, not 1 in 10¹⁸⁰. It is also where Don Stoner's appendix, for all its care, misses. It answers staging by pricing what the actors could not control: the betrayal, the price, the thrown silver. It never prices composition. A narrator can write the price of a betrayal, and where the silver landed, as easily as he can write the donkey. (The staging question itself, for the one event Don Stoner concedes could have been set up, is the subject of a separate essay.)
A defender will reply that this begs the question: if the Gospels are reliable reports, the events happened and the numbers stand. Fair enough, as far as it goes. But then the number is borrowed. All of its evidential weight rests on the reliability of the reports, and Stoner's conclusion, that "the Bible is inspired," takes in the very Gospels whose reports it assumes. Even granting the reports, the selection and guesswork problems that follow still apply. And the weight is thinnest precisely where the reports fall silent or diverge, as Mark's unnamed sum and Acts' field show.
There is a second, quieter dependency too: the prophecies are all drawn from the same cultural and textual pool. A first-century Jew who claimed to be the Messiah would naturally be associated with Davidic lineage (because everyone knew the Messiah was supposed to be Davidic), would teach in the Temple (because that's what Jewish religious figures did), and would attract attention in Jerusalem (because that's where messianic expectations were centered). These aren't independent variables. They're correlated features of a single social and theological context.
The Texas Sharpshooter Fallacy: Shooting First, Drawing the Target Later
Stoner's calculation commits what's known as the Texas Sharpshooter fallacy — named after the joke about a Texan who fires his gun at the side of a barn, then walks over and paints a bullseye around the tightest cluster of bullet holes. It looks like incredible marksmanship, but the target was drawn after the shots were fired. It's an informal fallacy, not a formal theorem of statistics — the everyday face of what statisticians treat more rigorously as the problem of selecting your hypothesis after seeing the data. David Hand, whom we'll meet in a moment, calls the underlying mechanism the "law of selection."
This is exactly what prophecy-fulfillment apologetics does. There are thousands of passages in the Hebrew Bible. After Jesus lived and died, early Christians went through these texts and selected the passages that seemed to match events in his life. They chose the hits and ignored the misses. Then apologists come along centuries later, treat the selected passages as though they were a predetermined list of criteria, and calculate the odds of someone matching all of them. But the selection happened after the outcome was known. That's not prediction — it's post-hoc pattern matching.
To see why this matters, consider an analogy. Shuffle a standard deck of 52 cards and deal them out. The probability of getting that exact sequence is 1 in 52 factorial — roughly 1 in 10⁶⁷. That's a number so large it dwarfs Stoner's 10¹⁷ by fifty orders of magnitude. Does this mean a miracle just occurred? Of course not. Some sequence had to come up, and every possible sequence is equally improbable. The astronomical improbability only seems meaningful because you're calculating it after the fact for the specific sequence you happened to observe.
David Hand addresses this in The Improbability Principle: Why Coincidences, Miracles, and Rare Events Happen Every Day. Hand identifies several laws that explain why apparently impossible coincidences are not only possible but, given enough chances, to be expected. His "law of truly large numbers" is particularly relevant here: given enough opportunities, even a very improbable specified event becomes likely to happen. Stoner does make one correction of this kind: he divides by everyone who has ever lived. But he counts the opportunities as people. The opportunities that matter are the thousands of verses early Christians could search after the fact, and those he never counts. And Hand's "law of selection" directly applies to the prophecy game — when you get to choose which events to focus on after they've occurred, you can always construct a narrative of incredible improbability.
The apologist might respond: "But the prophecies were written before Jesus, so the target was drawn first!" This is superficially plausible but collapses under examination. The prophecies weren't a predetermined list. There was no canonical document in Second Temple Judaism that said, "Here are the 48 things the Messiah must do, numbered 1 through 48, and anyone who does all of them is the real deal." Different Jewish groups had different messianic expectations. The Dead Sea Scrolls community expected multiple messiahs — a priestly one and a kingly one. The Psalms of Solomon describe a warrior king. Some texts anticipate a prophetic figure. Others expect no personal messiah at all, just divine intervention. John J. Collins documents this diversity thoroughly in The Scepter and the Star: Messianism in Light of the Dead Sea Scrolls.
What Christians did — and this is the critical move — was select from this diverse pool exactly those passages that could be mapped onto Jesus' life, while ignoring or reinterpreting the passages that couldn't. That's target-painting. That's the Texas Sharpshooter.
Stoner's Numbers: Questions Without Data
Back to probability, and to the inputs. Even if we granted the conditioning and overlooked the post-hoc selection problem (which we shouldn't, but let's pretend), Stoner's individual estimates cannot bear the weight he puts on them. They are not fabricated, and they are not arbitrary in the sense of being pulled from the air by one man. But they are guesses, and a calculation is only as good as its inputs.
Take his most famous example: the probability that the Messiah would be born in Bethlehem. Stoner assigns this a probability of 1 in 2.8 × 10⁵, based on the ratio of Bethlehem's average population, from Micah's day to his own, to the average population of the earth over the same period. Set aside the fact that we don't have reliable population data for first-century Bethlehem (or first-century anywhere, really), let alone an average across twenty-seven centuries. The deeper problem is that this calculation assumes a random person is equally likely to be born in any location on earth proportional to population. But messianic candidates weren't random people drawn from a uniform global distribution. They were Jewish men from the land of Israel, which drastically changes the sample space. And if someone wanted to claim Davidic lineage, as a messianic claimant would, being associated with Bethlehem, the city of David, was practically a requirement. It's not an independent variable with a probability of 1 in 280,000. It's a correlated feature of the messianic claim itself.
Or consider his probability for "betrayed for thirty pieces of silver," pegged at 1 in 10³. Where does that number come from? Stoner is candid: his estimates were pooled from more than 600 students in about a dozen college classes, and he lowered several of them to be "more conservative." He even defines the question: of all the people who have been betrayed, how many were betrayed for exactly thirty pieces of silver? But no one has ever counted that. There is no record of ancient betrayals and their prices from which a frequency could be drawn, so the 10³ has a question behind it but no data. That's a question, not a statistic, and averaged student guesses are still guesses.
This is a pattern throughout Science Speaks. Each estimate is a considered guess with no data behind it. Stoner's answer to the obvious objection, that readers should supply their own numbers or set them all at one in four, changes the guesses without supplying any data, and I've already said why the one-in-four fallback fails. If Stoner had submitted these calculations to a statistics journal, they would almost certainly have been rejected before reaching peer review, not because the conclusion was theologically inconvenient, but because elicited guesses with no base rate behind them are not measurements.
At this point an apologist usually plays a trump card: Science Speaks carries an endorsement from the American Scientific Affiliation, and McDowell quotes it in full. The foreword, by H. Harold Hartzler of Goshen College, states that the manuscript was "carefully reviewed by a committee of the American Scientific Affiliation members and by the Executive Council," and found "in general, to be dependable and accurate in regard to the scientific material presented," and that the probability principles were "thoroughly sound" and applied "in a proper and convincing way." Doesn't that settle it? No, and it's worth being precise about why.
First, consider what the ASA is. It's not a body of statisticians, and this wasn't anonymous peer review. The American Scientific Affiliation is a fellowship of evangelical Christians in the sciences, founded in 1941 around what it calls "a common fidelity to the Word of God," an organization whose express purpose is harmonizing Christian faith and science. Stoner was himself a co-founder. So the "review" is an in-house endorsement from a faith-based society the author helped establish, evaluating a book whose conclusion that society was already committed to. That is not what peer review means, and it's not what gives a result evidentiary weight. Notice, too, the hedges the popular retellings drop: "in general," and "in regard to the scientific material presented."
Second, and this is the deeper point, a favorable review can't manufacture an empirical basis that the numbers never had. Read closely, the committee's statement is about principles: the probability rules are "thoroughly sound," and the scientific material is accurate "in general." Its last clause does vouch for the application. My inference, and it is only an inference, is that the committee could vouch for the application only in the way a referee checks arithmetic. It says nothing about where the inputs come from, and there is nothing behind them to check. The 1 in 2.8 × 10⁵ for Bethlehem, the 1 in 10³ for thirty pieces of silver: these have questions behind them, but no counts. A reviewer can confirm that you multiplied your figures together correctly while the figures themselves remain guesses. Endorsing the arithmetic doesn't validate the premises, and the premises are guesses.
Several "Prophecies" Aren't Prophecies
Now the third lens: history and philology, reading the prophecies in their own settings. Stoner's calculation inherits a deeper problem from the apologetic tradition it serves: several of the "prophecies" it counts aren't prophecies at all, including two of Stoner's own eight. I've covered this in the Septuagint problem and Matthew's fulfillment formula, but the short version bears repeating here.
Take Stoner's fourth. Zechariah 13:6 reads, in the World English Bible, "What are these wounds between your arms?" The man answers, "Those with which I was wounded in the house of my friends." The Hebrew is literally "between your hands" (the word is yad), which is where the King James "wounds in thine hands," the rendering Stoner quotes, comes from. But read the paragraph it sits in (Zechariah 13:2-6). It is about the end of prophecy itself: on that day, the passage says, a prophet will be ashamed of his vision and will insist, "I am no prophet, I am a tiller of the ground." The man with the wounds is a prophet caught out and disowning his calling, and his wounds may well be the kind the prophets of Baal inflicted on themselves when they "cut themselves in their way with knives and lances" (1 Kings 18:28). Stoner prices this at 1 in 10³ as a prediction of the crucifixion. His fifth and sixth, the thirty pieces and the potter, come from Zechariah 11:12-13, where the prophet describes being paid off as a shepherd and throwing the silver "to the potter, in Yahweh's house." That is a sign-act told in the past tense, about the prophet's own wages. Matthew, who quotes it, credits it to Jeremiah.
The lists popularizers work from are worse. McDowell's list of prophecies "totally beyond the human control of Jesus," printed just before the calculation, includes Isaiah 7:14, the famous "virgin will conceive" passage. It is addressed to King Ahaz in the eighth century BCE about a sign that will occur in his own lifetime. The Hebrew word almah means a young woman of marriageable age; it says nothing either way about virginity. The Septuagint translation used parthenos, which Matthew then picked up and ran with. Most critical commentators read the sign within Ahaz's own crisis. Joseph Blenkinsopp's Anchor Bible commentary makes Immanuel a child of the royal Davidic house, possibly Hezekiah (pp. 233-34). Brevard Childs is the instructive exception, and he cannot be counted in that majority: he finds "the messianic hope associated with Immanuel" introduced in chapter 7 and developed into "a clearly messianic interpretation" of the child's role in chapter 8 (pp. 62, 80). But Childs is reading the shape of the finished book of Isaiah, not what the sign meant to Ahaz, and that is a different kind of claim from the one a probability calculation needs.
Hosea 11:1, "called my son out of Egypt," is explicitly about the Exodus. The full verse reads: "When Israel was a child, then I loved him, and called my son out of Egypt." The "son" is Israel. It's not a prediction about anything. It's a retrospective statement about the Exodus. Matthew 2:15 attaches it to the holy family's stay in Egypt, but this is typological reuse, not prediction-fulfillment.
If you're calculating the probability of someone fulfilling "prophecies" that aren't predictions in the first place, your calculation isn't wrong. It's meaningless. You're computing the odds of hitting a target that doesn't exist.
The Method Proves Too Much
The same method, applied consistently, would "prove" other messianic claimants with equal mathematical force.
Consider Simon bar Kokhba, who led the Jewish revolt against Rome in 132-135 CE. Rabbi Akiva, one of the most prominent rabbinic authorities of the era, declared bar Kokhba the Messiah. And bar Kokhba fulfilled messianic criteria: later tradition made him a descendant of David, he led an independent Judean state for some three years, his coins proclaimed "the freedom of Jerusalem," and his name was associated with the "star" prophecy of Numbers 24:17 — "A star will come out of Jacob." The name "bar Kokhba" literally means "son of the star." If you ran Stoner's calculation on bar Kokhba using a different selection of prophecies — the ones that emphasize military victory, national liberation, and Davidic kingship — you'd get numbers that are equally astronomically improbable. Would that prove bar Kokhba was the Messiah? Obviously not. He lost the war and died.
The same logic applies to Sabbatai Zevi, the seventeenth-century rabbi and mystic who declared himself the Messiah and who also matched a curated set of prophetic expectations. His followers performed exactly the same interpretive moves that Christian apologists perform: they took ancient texts, applied creative hermeneutics, and declared fulfillment. When Zevi converted to Islam under Ottoman pressure — a rather conspicuous failure of messianic expectations — his followers developed elaborate theological explanations for why the apparent failure was actually a deeper fulfillment. Sound familiar?
The point is not that bar Kokhba or Sabbatai Zevi was the Messiah. The point is that the probabilistic method doesn't distinguish between genuine fulfillment and post-hoc rationalization. It can't. It's not designed to. It's designed to produce impressive numbers, and it produces them regardless of whether the underlying claims are true.
Confirmation Bias: The Engine Behind the Machine
What drives the entire prophecy-probability enterprise is confirmation bias: the well-documented cognitive tendency to seek, interpret, and remember information that confirms pre-existing beliefs while ignoring or discounting information that contradicts them.
Apologists count the "hits," passages that can be matched to Jesus' life, and ignore the "misses," messianic passages that Jesus conspicuously failed to fulfill. The Messiah was supposed to build the Temple: Zechariah 6:12-13 says of the Branch that "he shall build Yahweh's temple," and Maimonides makes it a test, declaring that a king who "builds the Temple in its place, and gathers the dispersed of Israel" is "definitely the Mashiach" (Laws of Kings 11:4, in Eliyahu Touger's translation). Jesus didn't. The Messiah was supposed to gather the exiles of Israel (Isaiah 11:12). Jesus didn't. The Messiah was supposed to establish universal peace (Isaiah 2:4, Micah 4:3). Two thousand years later, we're still waiting.
If you included these unfulfilled prophecies in your probability calculation, the numbers would look very different. But Stoner doesn't include them, and neither does McDowell. The standard reply is that these belong to a second coming, and Stoner himself restricts his count to prophecies of "the first advent." But that is the sharpshooter again: every hit goes in the first column, and every miss is moved to the second. Prophecy lists even split single verses this way. One widely circulated list of 356 prophecies (this site's prophecy catalog sets each of its entries beside the critical reading) counts Isaiah 2:4 as fulfilled under the heading "He shall judge among the nations," and says nothing about the same verse's promise that "They shall beat their swords into plowshares." The same list files Zechariah's Temple-builder under "Priest and King," fulfilled not by any building but by the heavenly priesthood of Hebrews 8:1.
What the Author's Grandson Now Says
The last lens is reception history, and it holds the most striking verdict of all. The most interesting reader of Science Speaks today may be the grandson who keeps it online. Don Stoner, whose Appendix 3 is the argument's strongest modern form, has since added a fourth appendix, "A Challenge to Apologists (including the younger version of myself in the previous section)," marked as a 2022 revision, with a matching warning on the book's front page. "At this point in time," he writes, "Science Speaks is more likely to be weighed for its historical significance than its usefulness as an evangelical tool."
His reason is a version of the objection this essay has been pressing. Each probability term has a numerator, the number of ways a prophecy might count as fulfilled, and as interpreters find more ways to read a passage, that numerator grows. It "was usually presumed to be" one in his grandfather's book, he writes, "but this is no longer a satisfactory assumption." He has not abandoned the argument. He writes as someone still defending the Gospel, and he tells apologists that they face "a hundred years worth of raised objections which must now be addressed." But the concession is real, and it comes from inside the tradition. The grandson who wrote the calculation's strongest modern defense now warns that its key simplifying assumption no longer holds.
Where This Leaves the Number
Those are my conclusions, and the arithmetic behind each is above. But I'd challenge any defender of the probability argument to address these objections directly rather than simply restating the impressive numbers. The numbers are irrelevant if the methodology that produces them is broken, and the methodology is broken in at least five separate ways: the conditional probabilities are wrong (one event is even counted twice), the selection is post-hoc, the individual probabilities are guesses without data, several of the prophecies aren't prophecies, and the same method would validate competing messianic claims. Underneath all five sits the problem no estimate can fix: every fulfillment in the product is attested only by a text written to show fulfillment.
The "1 in 10¹⁷" number isn't a discovery about reality. It's an artifact of a flawed method: a number produced by multiplying unmeasured guesses, one of them counted twice, for events selected after the fact from texts that aren't predictions, and reported by writers who knew what they needed to report. Like any calculation built on unmeasured inputs, it is impressive on the page and empty underneath.
The most parsimonious explanation for why Jesus appears to fulfill Old Testament prophecies is not that the laws of probability were supernaturally overridden. It's that early Christians, steeped in the Hebrew scriptures and needing to make theological sense of a crucified Messiah, searched those scriptures for passages that could be reinterpreted as predictions and then told their story in a way designed to show fulfillment. That's not a conspiracy theory. The texts show us the habit that produced it. Matthew says "that it might be fulfilled" over and over again, and each time he is showing you how he reads Scripture: backwards, from the events. Once you see the method, the improbable numbers stop needing a miracle.
