Prophecy

"Predict how future events will turn out, so we might know you are gods." — Isaiah 41:23, NET¹

"He did not suppress them, but let them stand in their contradiction to events." — Moshe Greenberg, on Ezekiel's failed oracles, 1997²

Summary

This section asks one question.  Does the Bible contain predictions that came true in a way that ordinary foresight, editing after the event, or reading in after the fact cannot explain?  It answers in ten articles.  Most test particular texts:  the messianic prophecies, the visions of Daniel, the oracles against the nations, and the predictions about the end and about Israel.

Every article uses the same eight tests, set out in How to Test a Fulfilled Prophecy.  Each test answers one rival explanation.  Test 1 asks whether the words can be shown to exist before the event:  by a manuscript, a translation, a datable quotation, or internal evidence that does not argue from the prediction's accuracy.  That answers editing after the event.  Test 2 asks what the text committed to, so that a reader at the time could have said what would count as a failure, and it counts every free choice of start, unit and end.  That answers vagueness.  Test 3 asks whether anyone read the text as a prediction of that kind before the event.  A pattern noticed afterwards, which scholars call typology, may be good theology, but it is reading in after the fact.  Test 4 asks whether the outcome lay beyond the reach of everyone who knew the text, and Test 5 whether someone with no stake in it reports the fulfilment.  Test 6 counts every miss from the same source beside every hit, with a symmetry rule:  a miss excused as conditional costs the hits the same discount.  Test 7 allows a probability only from independent items that pass the other tests.  Test 8 applies the same rules to non-biblical oracles.  A test loose enough to pass Nostradamus is too loose, and one that no real prediction could pass is too strict.

The section's verdict:  the Bible contains a few predictions that the three rival explanations do not easily account for, none that they plainly cannot, and a real record of misses in the same books.

1. Two texts survive as evidence a skeptic must weigh: Isaiah 53 and Daniel 9. Both were written centuries before Jesus. The Great Isaiah Scroll was copied between 180 and 100 BC,³ and Daniel 9 predates Jesus by about two centuries even on the latest date any scholar gives the book. The central thing each is said to foretell, an innocent man's death for many and an anointed one cut off at a countable time, lay beyond any follower's power to arrange. But neither was read before Jesus, in any dated text, as foretelling a dying messiah. Isaiah 53 does not mention crucifixion, and most of its matching details could be shaped by a writer who knew it. Counted from any Persian decree, Daniel 9 lands within about a century of Jesus, and its exact year depends on choices of start, year length and end.
2. Several smaller hits add a little. Micah 5:2 expected a ruler from Bethlehem, though a birthplace can be supplied. Genesis 49:10 was an old hope of a ruler from Judah, though it sets no date. Nahum foretold the permanent ruin of Nineveh, which fell in 612 BC, though no one can show how long before 612 he wrote. Babylon fell to the Medes and Persians, and Edom ended as a nation. Jesus's prediction of the temple's fall is worth about seven to one if Mark was written before AD 70.
3. The misses are real, and they count in the same ledger. Ezekiel named Nebuchadnezzar as Tyre's destroyer and said the city would never be built again. Ezekiel himself records that the siege won nothing, and Tyre stands today. His forty-year desolation of Egypt never happened. Isaiah and Jeremiah describe a violent sack of Babylon, and Cyrus's army entered it without battle. Daniel 11 follows the history closely down to the persecution of 167 BC and then describes a campaign and a death that did not happen. Daniel 7, if its fourth kingdom is Greece, expected God's kingdom when Antiochus IV fell. "This generation will not pass away" either failed or was fulfilled only in the temple's fall, and the text does not settle which. Conservative answers exist for each: the prophecy was conditional, or not meant literally, or still to come. Each answer, by the symmetry rule, costs the matching hit.
4. The misses also show that the books were not rewritten to fit. A writer working after the events would not have kept Ezekiel 29:18 beside the Tyre oracle, or left Jeremiah's violent Babylon after Cyrus's peaceful entry. For the oracles against the nations, the misses are the best evidence that the words came first. Daniel 11 is the exception. Its accuracy stops at a dated point, which is the mark of writing after the event, and the case against that reading rests on one radiocarbon sample.
5. The popular case fails. Josh McDowell's "over 300" fulfilled prophecies come down to two that must be weighed and a few that add a little.⁴ Peter Stoner's odds of one in 10¹⁷ fail Test 7. Hosea 11:1 as prediction, the donkey of Zechariah 9:9 as foreknowledge, the "pierced" of Psalm 22:16, Isaiah 7:14 as a prediction of Jesus, and Tyre as a showpiece should all be retired. Conservative scholars have already conceded most of these points in print.
6. One finding cuts both ways. No dated text before Jesus has a messiah who suffers, dies and atones. So the apologist who says Jesus fulfilled what every Jew expected is wrong. And the skeptic who says the first Christians borrowed a dying-messiah story is wrong too, because there was none to borrow.

The articles in this section:

1. How to Test a Fulfilled Prophecy. The eight tests. Applied evenly, they reject Nostradamus and the ancient oracles written after the fact, and also Stoner's odds and several favourite Christian proof-texts.
2. Messianic Prophecy: An Overview. Of thirty texts applied to Jesus, two must be weighed, two add a little, and twenty-six cannot serve as evidence that anyone foresaw him.
3. Isaiah 53. Real but inconclusive evidence. The text is pre-Christian and most naturally describes one man suffering for others, but it does not mention crucifixion, and most of its details could be shaped.
4. Psalm 22. No manuscript says "pierced," and the New Testament never quotes that verse. The rest of the psalm is the better argument, and it turns on whether the Gospels told the crucifixion in its words because it happened that way.
5. Second Temple Messianic Expectations. Jews before Jesus expected a conquering son of David, sometimes with a priest, a prophet or a heavenly judge. No pre-Christian text has a messiah who suffers, dies and atones.
6. Messianic Birth Texts. Isaiah 7:14 carries no weight as a prediction, Micah 5:2 a little, and Genesis 49:10 a little as an old hope and none as a clock.
7. Daniel 9 and the Seventy Weeks. It argues that the seventy weeks predict the year and month of Jesus's crucifixion, counting 483 years of 360 days from 444 BC to the spring of AD 33.
8. Daniel 2, 7, and 11. Daniel 11 matches history closely down to 167 BC and then fails. The empire visions offer no fulfilled prediction that a skeptic must accept, unless a radiocarbon date settles the book's date in their favour. The date itself is argued in When Was the Book of Daniel Written?
9. Prophecies About Nations and Cities. Nineveh does best, Egypt worst, and Tyre is a failed prediction that its own book kept on the record. Cyrus by name rests on the disputed date of Isaiah 40–48.
10. Eschatology and Israel. The temple's fall is a modest hit if Mark is early, "this generation" is either a miss or a hit on the temple alone, all fifteen dated end-time forecasts it counts failed, and Israel in 1948 fails the out-of-reach test.

Part 1: How to Test a Fulfilled Prophecy

How to Test a Fulfilled Prophecy supplies the rules every other article applies.  The Bible sets the first rule itself:  Deuteronomy 18:22 says that a prophet whose prediction "is not fulfilled" has spoken presumptuously.¹  Applied to non-biblical controls, the tests do what a skeptic would expect.  The Egyptian Prophecy of Neferti, the Babylonian Dynastic Prophecy and much of the Sibylline Oracles are accurate up to their own time and no further, so they fail Test 1.  Nostradamus's quatrain on the death of Henri II passes Test 1, because it was printed in 1555, four years before the joust, and fails Tests 2 and 6, because most of its details do not fit and it is one of about 940 quatrains.⁵⁶  Applied to Christian apologetics, the tests remove Peter Stoner's one in 10¹⁷, whose estimates Stoner said outright he would not defend:  "No attempt will be made, however, to defend these estimates."⁷  They also remove Hosea 11:1 as a prediction, which the evangelical R. T. France calls "not a prediction, but an account of Israel’s origin,"⁸ the donkey of Zechariah 9:9, which the Gospels say Jesus sent for, Psalm 22's "pierced," and Ezekiel's Tyre.  What the tests leave is a short list of texts that were certainly written first and lay beyond any follower's reach, each to be judged on what it actually commits to.

Part 2: Messianic Prophecy: An Overview

Messianic Prophecy:  An Overview sorts the thirty Old Testament texts that the New Testament or popular apologetics applies to Jesus, using Tests 1, 3 and 4.  Every text passes Test 1.  Two, Isaiah 53 and Daniel 9, survive as evidence a skeptic must weigh.  Nine were read as messianic before Christianity, but the match lay within someone's reach or cannot be checked;  Micah 5:2 and Genesis 49:10 carry a little weight among them.  Seven were not read as messianic before Jesus and depend on his own choices or on the Gospel writers' reports, seven are typology, and five are not predictions in their own setting or are not in the text at all.  The page also scores Daniel 7:13, the "one like a son of man," which no other article scores:  probably read as messianic near Jesus's time, and committed to nothing a historian can yet check.  Its conclusion is that the case is far smaller than the lists claim, and that it does not rest on their weak items.

Part 3: Isaiah 53

Isaiah 53 finds the passage real but inconclusive evidence.  The text predates Jesus beyond dispute, and two Jewish scholars, Marc Brettler and Amy-Jill Levine, grant that "'my servant' in 52:13 and 53:11 cannot refer to the collective Israel."⁹  But the national reading is ancient too:  Origen met Jews arguing it around AD 248.¹⁰  Whether the servant actually dies is still disputed, and most of the details apologists lead with, such as the silence before accusers and the grave with the rich, could be shaped by a writer who knew the text or enacted by a man who did.  What survives is the shape of the fit.  The chapter does not mention crucifixion, and crucifixion was the one thing about Jesus that Jews could not accept.  In Justin's dialogue of about AD 160, the Jew Trypho grants the suffering and balks at the cross:  "For we cannot bring ourselves even to think of this."¹¹  The cross came first and the text was recruited to explain it, so the passage passes Test 4 at its centre while failing it in its details.

Part 4: Psalm 22

Psalm 22 finds that the famous clause "they pierced my hands and my feet" will not carry the weight put on it.  The standard Hebrew reads "like a lion," the Greek "they dug," and no manuscript in any language says "pierced."  The Dead Sea scroll usually cited for "pierced," from Nahal Hever, was copied in AD 50–68, after the crucifixion, and the charge that either side tampered with the verse fails both ways.  The decisive fact is that the New Testament never quotes the verse, though it quotes verse 18 word for word.  The rest of the psalm is the better argument:  Jesus cries its first verse from the cross, the mockers use the words of verses 7–8, and John quotes verse 18 for the soldiers dividing his clothes.  No pre-Christian Jewish text reads the psalm as messianic, so the case turns on the direction of dependence:  did the Gospel writers tell the crucifixion in the psalm's words because it happened that way, or did the psalm supply the story?  The article finds that open, with John's garment scene, which splits one line of Hebrew poetry into two acts, the clearest sign of shaping.

Part 5: Second Temple Messianic Expectations

Second Temple Messianic Expectations supplies the Test 3 evidence for the messianic texts, from Dead Sea Scroll copies made before Jesus, some of them dated by radiocarbon.  The common hope was a royal son of David who would drive the gentiles out of Jerusalem and rule by the law.  John Collins, the leading specialist, counts "four basic messianic paradigms (king, priest, prophet, and heavenly messiah)," and says they "were not equally widespread."¹²  A list of scriptures has a datable messianic reading older than Jesus, among them Numbers 24:17, Isaiah 11, 2 Samuel 7, Amos 9:11, Psalm 2, Deuteronomy 18 and Isaiah 61.  Isaiah 53, Psalm 22 and Zechariah 12:10 are not on it.  Every text offered as a pre-Christian suffering messiah fails on its reading or its date, and Martin Hengel, the strongest defender of the opposite view, concedes that such an expectation "cannot therefore be proven to exist with absolute certainty and in a clearly outlined form in pre-Christian Judaism."¹³  The article draws the conclusion both ways:  the apologist who says Jews expected a suffering messiah is wrong, and so is the skeptic who says the first Christians borrowed one.

Part 6: Messianic Birth Texts

Messianic Birth Texts reads three texts in their own settings.  Isaiah 7:14 is a sign to King Ahaz with a deadline:  before the child can tell evil from good, the lands of the two kings Ahaz feared will be desolate, and Damascus fell in 732 BC and Samaria in 722.  Its word for the mother means a young woman, and the evangelical Craig Blomberg concedes that "there are no demonstrably pre-Christian Jewish texts that reflect on the meaning of Isa. 7:14."¹⁴  Micah 5:2 is a real prediction of a king from "Bethlehem Ephrathah,"¹ and Blomberg calls it the one text in Matthew's birth story that fits "a very straightforward scheme of prediction and fulfillment."¹⁴  But a birthplace is within reach of a family and of a storyteller, so its weight is small.  Genesis 49:10 was read before Christianity as the promise of a ruler from Judah.  A Qumran commentary reads "Until the messiah of righteousness comes, the branch of David."¹⁵  But the verse sets no date, and the popular claim that its scepter departed in AD 7 rests on a mistranslated year and a rabbinic cry that no rabbinic text contains.

Part 7: Daniel 9 and the Seventy Weeks

Daniel 9 and the Seventy Weeks states its conclusion directly:  "Daniel 9:24-26 uniquely predicts the year and month of Jesus’ crucifixion, with no credible alternative explanation."  It dates Artaxerxes's order to rebuild Jerusalem, in Nehemiah 2, to March or April 444 BC.  It counts 483 years of 360 days, which is 173,880 days, and reaches March or April of AD 33.  It holds that the Septuagint and the Dead Sea Scrolls show the prophecy was written centuries before Jesus, and it assigns the last half of the seventieth week to the future.  The method article passes Daniel 9 on Test 1 even at 165 BC, the latest date any scholar gives the book.  Under Test 2 it asks that the start, the length of the year and the end be fixed before counting:  the four usual starting decrees and the two year lengths give eight landing years, from 62 BC to AD 40, and three of them fall in AD 26–36, the span of Jesus's adult ministry and death.¹⁶  So the count lands in the right century from every Persian decree, which is a real feature of a text that predates Jesus by at least two centuries, and its exact year rests on the choices the Daniel 9 article argues for.  The main rival reading makes the anointed one the high priest Onias III, murdered in 171 BC, and When Was the Book of Daniel Written? finds that on that reading the 483 years overshoot the real interval by fifty to seventy years.  Neither reading fits without choices.

Part 8: Daniel 2, 7, and 11

Daniel 2, 7, and 11 tests Daniel's visions of world empires.  Daniel 8 names its empires, Media-Persia and Greece, and its small horn is Antiochus IV, who desecrated the Jerusalem temple in 167 BC;  commentators of every school agree.  Daniel 11:2–35 matches Persian and Hellenistic history closely.  By the article's own count it makes 48 claims that an event could have falsified, none contradicted by an independent source and most confirmed by Polybius, Appian, Livy, Diodorus, Babylonian tablets or 1 Maccabees.¹⁷  Then the match stops.  Verses 40–45 describe a third conquest of Egypt and a death near Jerusalem, but Antiochus died in Persia in late 164 BC, and the conservative Joyce Baldwin grants that the conquest and the battle "never took place."¹⁸  Daniel 2 and 7 offer no checkable fulfilled prediction.  If their fourth kingdom is Greece, chapter 7 expected God's kingdom to follow Antiochus's fall, which Hartman and Di Lella call "a prediction not borne out by events."¹⁹  If it is Rome, Rome's rise was guessable in the 160s BC, and no surviving reader identifies the fourth kingdom with Rome before Rome ruled Judea.

The date of the book decides how much of this counts, and When Was the Book of Daniel Written? finds it undecided.  The court tales of chapters 1–6 carry sixth-century Babylonian knowledge.  The visions carry second-century detail and an ending that failed in 164 BC.  The one physical date is a 2025 radiocarbon measurement of 4Q114, a Qumran copy that preserves part of Daniel 11.  It gives two ranges of nearly equal probability, 355–285 BC and 230–160 BC, and the dating team rejects the older one by comparing the handwriting of dated Aramaic manuscripts of that period.³  On the dating article's reading of the figures, three quarters of the younger range falls between 205 and 175 BC, before Antiochus IV took the throne.³  If that one sample holds, Daniel 11:21–35 is a remarkable prediction.  If the turning point holds, verses 2–39 are history written as prophecy.  On neither view are verses 40–45 a fulfilled prediction.

Part 9: Prophecies About Nations and Cities

Prophecies About Nations and Cities tests six cases clause by clause, counting every clause and not only the ones that came true.  Nineveh does best.  Nahum names the city, mentions the fall of Egyptian Thebes in 663 BC as past, and predicts the permanent ruin of the capital of the strongest empire on earth.  A Babylonian chronicle records that it was taken in 612 BC and turned into ruin-hills, and Xenophon passed it two centuries later without knowing its name.  The weakness is that no one can show how long before 612 Nahum wrote.  Egypt does worst.  No clause of Ezekiel's forty-year desolation happened, and Pharaoh Amasis reigned four decades in prosperity.  Tyre is a failed prediction that its own book kept.  Ezekiel 26 names Nebuchadnezzar, and Ezekiel 29:18 says that "he and his army received no wages from Tyre for the work he carried out against it."¹  The evangelical Robert Chisholm concedes that Nebuchadnezzar "did not devastate the city in the manner described by Ezekiel,"²⁰ and Tyre stands today;  a separate article weighs it in full.  Babylon fell to the Medes as foretold, but without battle, and it emptied over centuries.  Edom lost its independence, but its land later held Petra.  Cyrus by name would outweigh them all if Isaiah wrote it in the eighth century, and nothing in the manuscripts can show that he did.  Two patterns run through the cases.  The oracles that came true went with the odds, since the royal cities of Mesopotamia were all eventually abandoned while the Phoenician ports lived on.  And the books kept their misses, so the same evidence that weakens these oracles as proof of foresight shows that they were written when they claim.

Part 10: Eschatology and Israel

Eschatology and Israel tests three claims.  The fall of the temple.  Jesus's prediction that the temple would be torn down came true in AD 70 and passes Tests 4 and 5, but it passes Test 1 only if Mark was written before the war, which the site's Gospel dating article leaves undecided.  A control limits its weight:  another Jesus, the son of Ananias, cried woe on Jerusalem from four years before the war until a Roman stone killed him in the siege.²¹  The article prices the razing at about one chance in seven for a well-informed forecaster in AD 30, so even with an early Mark it is a good forecast worth about seven to one.  "This generation."  "This generation will not pass away until all these things take place"¹ either failed, if the coming of the Son of Man in the same chapter was meant within that generation, or was fulfilled only in the temple's fall.  The symmetry rule says it cannot be both.  The dated forecasts and Israel.  Fifteen dated end-time predictions by Christian interpreters, from Hippolytus's AD 500 to Harold Camping's 2011, all failed.  The article charges them to the interpreters' arithmetic, not to the texts, which say "as for that day or hour no one knows it."¹  Israel's founding in 1948 fits the promised gathering, but the return was sought and lobbied for by readers of the prophecies, so it fails Test 4.

Part 11: The section side by side

The two tables score the strongest and the weakest case from each article by the tests, in each article's own grades.  "Pass" and "Fail" are for a reader who does not already accept the Bible's authority.  A dash means the article did not score that test for that case.  Test 8 is applied by the non-biblical control at the foot of each table:  the same columns must treat it fairly.  Second Temple Messianic Expectations scores no prediction of its own;  it supplies the "read before" column for the messianic texts.

The strongest case in each article

The weakest case in each article

Four things stand out.  First, the two cases that must be weighed, Isaiah 53 and Daniel 9, pass Test 1 beyond dispute and draw their strength from Test 4:  the death at the centre of each lay beyond the reach of the people who knew the text.  Neither passes Test 3 with a reading dated before the event, so neither can claim that its fulfilment was recognised in advance.  Second, the next three, Daniel 11, Cyrus and the temple's fall, each turn on an open Test 1:  one radiocarbon sample, the date of Isaiah 40–48, and the date of Mark.  Third, every weak case fails on Test 3 or Test 4, a pattern read in afterwards or a match someone could supply, or on Test 6, clauses that did not happen.  Fourth, the controls behave as the method promises.  Nostradamus fails, and Josephus's prediction to Vespasian passes with its limits stated, so the tests are neither rigged against prophecy nor loose enough to pass anything.

Part 12: Questions that cross articles

Seven questions run across more than one article.  Each is settled here from the sources the articles themselves cite, and the articles follow these answers.

1. Micah 5:2 on Test 3: partly, not pass. Matthew 2 and John 7 show the reading was current in Jesus's day. The method article's rule decides how much that counts: a text fails Test 3 if its predictive reading first appears after the event. Both witnesses are Gospels, written after Jesus's life by his followers. In Matthew the chief priests answer Herod with the verse, and in John a crowd objects that "the Christ is a descendant of David and comes from Bethlehem."¹ No Qumran text, apocryphal book, Philo or Josephus cites the verse, and the Targum's and the Talmud's Bethlehem readings were written down later. What keeps Micah above a plain fail is the form of John's report. The expectation is an objection to Jesus that John never answers, which a writer inventing evidence would not do. So the grade is "Partly."
2. Isaiah 53 and the Targum: early Jewish evidence, but not a reading before Jesus. The Aramaic Targum renders the servant as "my servant the messiah," but a Targum standing alone cannot date a reading before Jesus. Brettler and Levine write that the Targum "underwent several editions in the second and the fourth centuries CE, though it preserves earlier traditions," and that its rendering of this passage was "likely composed some point before the Bar-Kochba revolt (132–135)."⁹ That is a latest date, a century after the crucifixion, not an earliest one. Joseph Fitzmyer calls the Targums "a written record dating in general from the Amoraic period," the age of the rabbis who followed the Mishnah.²² Neither source puts the reading before Jesus, so it fails Test 3 as a reading made before the event. The Targum still matters. It shows that Jews read the servant as the messiah early on, which answers the charge that Christians invented that reading. It also moves every line of suffering onto someone else, which is why the Isaiah 53 article calls it a double-edged witness. Isaiah 53's verdict does not rest on the point.
3. Genesis 49:10 in Greek: ambiguous, so the pass rests on Qumran. The Septuagint looks messianic because it makes "the expectation of nations" a person. But John Collins says that "The majority reading of the LXX manuscripts translates siloh as 'the things laid away for him' and fails to associate the future glory of Judah with an individual ruler."²³ The Greek reads "A ruler shall not fail from Judah, nor a leader from his thighs, until the things laid up for him come; and he is the expectation of nations."²⁴ What comes is neuter and plural, ta apokeimena, "the things laid up." The "he" who is the nations' expectation is a person, but the nearest person in the verse is the ruler from Judah whose line does not fail. So the Greek has a person and does not have a single future ruler who comes, and it cannot carry Test 3 alone. The pre-Christian messianic reading rests on 4Q252, a Qumran commentary copied in the Herodian period,²⁵ which reads "Until the messiah of righteousness comes, the branch of David."¹⁵ So Genesis 49:10 passes Test 3 as a messianic hope, on Qumran's evidence.
4. Daniel 7:13 is scored in the Overview. Daniel 2, 7, and 11 and Second Temple Messianic Expectations both leave the "one like a son of man" to the Overview, which scores it: probably read as messianic near Jesus's time, through the Similitudes of Enoch, whose date is disputed, and committed to nothing a historian can yet check.
5. Did Stoner treat the prophecies as independent? No. The method article quotes his second chapter, "If the events are not entirely independent, a somewhat different theory of probability must be used,"²⁶ and shows that he framed the linked items conditionally. The fair complaints against Stoner are his undefended estimates and his choice of eight items from "more than three hundred."²⁶
6. The Gabriel stone. Israel Knohl read a stone inscription from about the turn of the era as telling a slain leader to live after three days. He gave up that reading in 2011: "I now accept the reading האות (“the sign”) of the third word of line 80."²⁷ The stone is not a pre-Christian witness to a dying and rising messiah.
7. Why Psalm 22:16 fails Test 3. Not because the New Testament never quotes the verse, which is a fact about Christian use, but for the reason the Psalm 22 article gives: no pre-Christian Jewish text reads the psalm as messianic.

Part 13: One ledger for the hits and the misses

Test 6 asks that the misses be counted beside the hits, from the same source.  On the plain reading of the texts, the section's articles find these misses.

The rescues are open to a believer, and conservative scholars use them.  Each one has a price under the symmetry rule.  A reader who calls Babylon's sack figurative cannot count Babylon's ruin as a literal fulfilment.  A reader who calls Tyre's oracle conditional cannot then count Alexander's causeway of 332 BC as its fulfilment.

The misses do one more thing, and it favours the books.  They show that Ezekiel and Jeremiah were not rewritten to fit events.  Moshe Greenberg draws the lesson from Ezekiel 29:17–21:  the prophet "did not suppress them, but let them stand in their contradiction to events."²  So for the oracles against the nations, what came true was said first.  The same argument does not help Daniel 11, where the accuracy itself stops at a dated point.

Objections and responses

1. "The misses prove the Bible is not inspired." The tests measure evidence of foreknowledge, not inspiration. On the plain reading some predictions failed, and conservative scholars say so. Robert Chisholm of Dallas Theological Seminary writes that "many prophecies were not fulfilled either in part or in whole."²⁰ The Bible's own answer is that much prophecy was conditional: Jeremiah 18 states the condition in general, Jonah 4:2 states it for Nineveh, and Jeremiah 26:19 records it for Micah. Chisholm argues that Deuteronomy's test "must apply to short-range prophecies, not prophecies of the distant future."²⁰ That answer is coherent, and it has a price: a reader who excuses a miss as conditional must discount the hits by the same rule. What the misses do rule out is the popular claim that every prediction was fulfilled in detail. What they mean for inspiration depends on what a reader thinks inspiration promises, which no historical test can settle.
2. "The hits were written after the event." For Isaiah 53 and Daniel 9 they were not. The Great Isaiah Scroll was copied between 180 and 100 BC, and Daniel 9 predates Jesus by about two centuries even at 165 BC. For the oracles against the nations, the misses are the evidence against the charge, because no later editor would have kept Ezekiel 29:18 or left Jeremiah's violent Babylon standing after Cyrus. The charge has real force in two places, and the section says so. Daniel 11 is accurate to a dated point and wrong after it, the mark of history written as prophecy, and only one radiocarbon sample stands against that reading. And Cyrus by name, Nahum's lead time and Mark 13 all rest on dates that no manuscript can settle.
3. "You only count the hits." The section counts the misses on purpose. The nations article scores every clause of each oracle, including the ones that failed. The Overview sorts thirty texts and finds that twenty-six cannot serve as evidence. The eschatology article lists every dated end-time forecast it could read in the forecaster's own words, and none came true. The selection problem is real for the popular lists. Stoner wrote that "More than three hundred prophecies" had been listed, then priced eight without saying how they were chosen.²⁶ That is one reason Test 7 refuses his number.
4. "The tests are built so that no prophecy can pass." Josephus's prediction to Vespasian passes Tests 1, 2 and 5. Isaiah 53 and Daniel 9 pass Test 1 beyond dispute, and Nahum's Nineveh passes on its main clauses. The tests also reject a skeptic's shortcut. Dating a text late only because it is accurate begs the question, as the method article says with the conservative John Oswalt; a late date has to rest on a turning point or on other evidence that does not assume the answer.
5. "The Gospel writers made the fulfilments up." The tests do not assume invention. Test 5 says only that a detail reported by believers alone cannot carry a probability argument, and the strongest cases do not depend on such details. Jesus's execution is reported by Tacitus, a Roman senator who despised the Christians: "Christus, from whom the name had its origin, suffered the extreme penalty during the reign of Tiberius at the hands of one of our procurators, Pontius Pilatus."²⁸ The shape of the fit also runs against invention. No one expected a crucified messiah, and Isaiah 53 does not mention crucifixion, so the cross was the fact the first Christians had to explain, not one they took from a text. Where shaping is visible, as in John's garment scene from Psalm 22, the articles say so.
6. "Jesus fulfilled the prophecies on purpose." Some he did. The Gospels say he sent for the donkey of Zechariah 9:9, which is why it fails Test 4, and the critic E. P. Sanders inclines "to the view that it was Jesus himself who read the prophecy and decided to fulfil it."²⁹ A man can choose how he enters a city and what he cries. He cannot arrange to be executed by Rome, which is the centre of Isaiah 53 and Daniel 9. Daniel 9's exact year is harder, because a man who knew the count could choose when to go to Jerusalem, and the Daniel 9 article raises that question itself.
7. "Jews never read these texts as messianic, so the Christian reading is hindsight." For the suffering texts, Second Temple Messianic Expectations agrees. No dated text before Jesus reads Isaiah 53, Psalm 22 or Zechariah 12:10 of the messiah, and John Collins writes that "There is no evidence that anyone in first-century Judaism expected such a figure."¹² For the royal texts the opposite is true: Numbers 24:17, Isaiah 11, 2 Samuel 7, Psalm 2 and Genesis 49:10 all have messianic readings older than Jesus. The finding cuts both ways. It removes the claim that Jesus fulfilled what every Jew expected, and it removes the claim that Christians borrowed a ready-made story of a dying messiah.
8. "Even weak items add up to a cumulative case." Only items that pass Tests 1 to 5 can be added, and only if they are independent of each other. Typology adds nothing however many examples there are. Matches reported by the same few writers, who knew the texts, are not independent. The two strongest texts do not need the weak items, and the weak items do not strengthen them.
9. "Daniel 9 lands on the very year of the crucifixion." On one set of choices it does, and the Daniel 9 article argues for that set: Artaxerxes's decree in 444 BC, years of 360 days, and AD 33. Test 2 asks that the start, the year and the end be fixed before counting and that the alternatives be counted. Of the eight combinations of the usual decrees and year lengths, three land in AD 26–36, and every Persian decree lands within about a century of Jesus. That makes Daniel 9 the strongest dated case in the section, and its weight turns on whether its choices can be shown to be the text's own rather than the interpreter's.
10. "Nahum wrote when Assyria was already collapsing." He may have. Nahum treats the fall of Thebes in 663 BC as past and wrote before 612, and no one can show where in those fifty-one years he stands. If he wrote in the 620s or 610s, the prediction was far less bold. The nations article ranks Nineveh first with that weakness stated.

Issues and unanswered questions

1. Daniel 9's choices. Its weight depends on fixing the decree, the length of the year and the end before counting. Its own article dates the decree to 444 BC; if the decree fell in 445 BC, the same 360-day count lands in AD 32.¹⁶ No dated pre-Christian reading of Daniel 9:25 as messianic has been shown, since the one Qumran text that seems to quote it, 11Q13, has only "Dan[" surviving.
2. One radiocarbon sample decides Daniel 11. If 4Q114's date is replicated, Daniel 11:21–35 is a remarkable prediction; if not, the turning point makes it history written as prophecy. The radiocarbon date is of the parchment, which the scribe used later by an unknown interval.
3. Isaiah 40–48 has no article. The Cyrus case rests entirely on when those chapters were written, and the section has not yet argued it.
4. Mark's date decides the temple prediction, and the site's Gospel dating article leaves it between the 40s and AD 70.
5. Nahum's decade is unknown, and with it how bold the Nineveh oracle was.
6. Texts without their own article. Zechariah 12:10, the piercing text the New Testament actually quotes, best handled with the rest of Zechariah 9–14; Psalms 2 and 110; Daniel 7:13, scored only in the Overview; Isaiah 9:1–7; and Genesis 3:15, which popular lists use and no article scores.
7. Priors are judgments. How likely an outcome was to an informed observer at the time, such as the chance that a Davidic claimant came from Bethlehem, can rarely be computed. The tests make the judgment visible; they do not remove it.
8. Two questions inside the strongest texts are open. Whether the servant of Isaiah 53 actually dies, or survives a near-death, is still argued by Jewish scholars. And whether the passion narratives were told in Psalm 22's words because events happened that way, or because the writers had the psalm open, is unsettled.

Verdict

Does the Bible contain predictions that came true in a way that ordinary foresight, editing after the event, or reading in after the fact cannot explain?  Tested from the manuscript date forward, it contains two that those explanations do not easily account for, and none that they plainly cannot.

Isaiah 53 and Daniel 9 were on record centuries before Jesus, so editing after the event is ruled out for them.  The thing at the centre of each, an innocent man's death for many and an anointed one cut off at a countable time, lay beyond the reach of anyone who knew the texts.  Reading in after the fact is not ruled out.  Neither text was read before Jesus, in any dated source, as foretelling a dying messiah.  Isaiah 53's details could be shaped by a writer who knew it, and Daniel 9's exact year depends on choices of decree, year and endpoint.  They are real evidence, and they are not proof.  Around them sit smaller hits:  Micah's Bethlehem, Judah's old hope of a ruler, Nineveh's ruin, Babylon's fall to the Medes, Edom's end as a nation, and the temple's fall if Mark is early.

The same ledger holds clear misses on the plain reading:  Tyre, Egypt, Babylon's violent end, Daniel 11:40–45, Daniel 7 if its fourth kingdom is Greece, and perhaps the generation saying.  Later interpreters went none for fifteen on their dated forecasts.  The conservative rescues are open, and each costs a matching hit.  The misses also do what the hits cannot:  they show that Ezekiel and Jeremiah were not rewritten to fit events, so what came true in those books was said first.

The popular case, hundreds of fulfilled prophecies and odds of one in 10¹⁷, does not survive the tests, and conservative scholars have conceded most of its weak items in print.  What survives is narrower and is stated here with its weaknesses.  A skeptic cannot dismiss it by pointing at the weak items, because it does not rest on them.

Sources