Justified True Belief

Mapping the Landscape of Good Reasons for the Truth of Christianity

Christian Evidences

Evidence & Arguments for Christian Theism

Divinity of Christ

Biblical Evidence for High-Christology: Jesus is God

TBD

Old Testament Criticism

Evidence for the Reliability of the OT Bible

Common Objections

Objection Analyses to Christian Theism

World Religions

Critical Analyses of Non-Christian Religions

Philosophy

Phileō Sophia - to Love Wisdom.

Epistemology

Study of Knowledge

Dwindling Probabilities

(●) Alvin Plantinga argues that the historical case for Christianity loses probability at every step, so that it cannot support the firm belief faith involves. + A family is planning a Saturday trip to the coast. To make the noon ferry, the car has to start, the traffic on the highway has to be light, the bridge has to be open, and there has to be parking at the dock. Each of those is likely, maybe nine chances in ten. But the father, who does the math in his head, frowns. Four things that are each ninety percent likely, all of which have to happen, come to only about sixty-six percent together. Add two more and they are down near half. That frown is the idea behind one of the most discussed arguments in Christian philosophy of the last generation. It comes from Alvin Plantinga, in Warranted Christian Belief (2000), and it is called the principle of dwindling probabilities. A few terms carry it, and they are worth having in hand before going any further. - A probability is a measure of how likely something is, from zero (impossible) to one (certain). Ninety percent is 0.9. - A conjunction is two or more claims joined by "and": the car starts and the bridge is open. The conjunction rule says that a conjunction can never be more probable than any one of its parts. You cannot be likelier to make the ferry than to get the car started, because making the ferry includes getting the car started. - The multiplication rule says how much less probable the conjunction is: the probability of the first claim, times the probability of the second given the first, and so on. "Given the first" means taking the first as settled: given that the car has started, how likely is light traffic? Multiply the father's four nines, 0.9 × 0.9 × 0.9 × 0.9, and you get his arithmetic, about 0.66. Plantinga applied this to the historical argument for Christianity, the case that the historical evidence, above all the evidence for the resurrection, shows the Christian faith to be true. It has been made by apologists, Christians who give a reasoned defense of the faith (the word comes from the Greek for a defense speech in court), from the early church, through Joseph Butler and William Paley in the 1700s, to the Oxford philosopher Richard Swinburne today. Plantinga broke the case into a chain of steps, roughly these: - God exists; - God would want to reveal himself and provide a way of salvation for human beings; - Jesus taught what the Gospels say he taught, including his claims about himself; - Jesus rose from the dead; - in raising him, God endorsed what he taught; - so the central teachings of the Christian faith are true. Each step, Plantinga argued, is at best probable given the one before, so the probability shrinks as the chain goes on, exactly as the ferry odds did. And the first step sets the ceiling. Plantinga started from the probability that God exists on background knowledge, meaning what we know about the world in general. Since a chain of claims joined by "and" can never be more probable than its first link, the whole chain could never rise above that first figure. - That first figure is the kind natural theology gives. Natural theology is the case for God built from nature and reason alone, such as the existence of the cosmos and the moral law, without appeal to Scripture. - Timothy McGrew's 2004 reply works an example that shows how fast the shrinking goes. Give six steps the figures .9, .9, .7, .6, .9, and .7, and they multiply to about .2143, roughly one chance in five, even though no single step falls below six in ten. His conclusion was that the historical argument, whatever its value, cannot give the high probability that the firm belief of faith involves. Where, then, does faith get its footing? On Plantinga's view, the warrant of faith (his word for whatever turns a true belief into knowledge) comes instead from the work of the Holy Spirit, without needing the historical case at all.

(●) The dwindling calculation leaves out the very evidence the historical argument rests on. + In 2004 Timothy McGrew asked, in the journal Philosophia Christi, "Has Plantinga Refuted the Historical Argument?" Plantinga replied in the same journal in 2006, and in that same issue Timothy and Lydia McGrew answered him with a rejoinder, which is simply a reply to a reply. Richard Swinburne, whose version of the historical argument had been Plantinga's main target, also replied, in 2004. McGrew's central point is about what the probabilities are probabilities of, and given what. In probability talk, "given" names what you are taking into account. The chance of rain given dark clouds overhead is higher than the chance of rain given a clear sky. Same question, different information, different answer. - In Plantinga's calculation, each step is assessed on background knowledge, written K: what we know in general about the world, before looking at the specific historical evidence about Jesus. - But the historical argument is an argument from evidence: the testimony of the disciples, the empty tomb, the appearances, the conversion of Paul, the rise of the church. Call that evidence E. - McGrew argued that Plantinga kept the evidence fixed at K while the hypothesis (the claim being tested) was refined, step by step. So the calculation measures how probable Christianity is before the historical evidence is considered, not how probable it is after. - The evidence for the resurrection bears on every step in the chain, including the first. Testimony and an empty tomb that are hard to explain unless God raised Jesus are, for that very reason, evidence that there is a God. In the rejoinder, the McGrews add a further point: there is no single "historical argument" to refute. Apologists have built many, and an objection that fits one way of setting out the case need not touch the others. McGrew's reply can be set out as a deductive argument, one in which, if the premises are true, the conclusion has to be true. Think of the premises as a locked chain: if each link holds, the conclusion cannot slip out. - You might wonder whether a chain of premises is just the father's ferry trip again, with doubt piling up at every link. It is not, because validity does not dwindle. A valid step is one where the conclusion really does follow, and it passes truth from premises to conclusion without loss. - Confidence can dwindle, though. Three premises that are each only a little more plausible than not can be doubtful taken together, because the conclusion needs all three at once. So the usual standard for a good deductive argument, that each premise be more plausible than its denial, is only the minimum. Each premise here is defended as far more plausible than that, so that the three together still leave the conclusion well supported. • The numbers show why. If each of three premises is at least ninety percent likely, all three together are at least seventy percent likely, however they are related, because the three ten-percent chances of failure add up to no more than thirty. - The three premises rest on different kinds of ground: • (P1) describes what the historical argument is. • (P2) is an analysis of Plantinga's calculation that, by Craig's report, Plantinga himself came to accept. • (P3) rests on the rules of probability.

(P1) The historical argument reasons from the historical evidence about Jesus. + The historical argument is an argument from evidence. Its strength depends on that evidence. (1) The argument has always started from testimony and events. - Paul lists the witnesses: Jesus "appeared to more than five hundred brothers at one time, most of whom are still alive, though some have fallen asleep" (1 Corinthians 15:6). The natural reading of "still alive" is that they could still be asked. - Luke tells Theophilus that he has written "that you may have certainty concerning the things you have been taught" (Luke 1:4), on the basis of what eyewitnesses delivered (Luke 1:2). - The apologists who built the historical argument, from Butler and Paley to Swinburne and the McGrews, all made the evidence its engine: the testimony, the empty tomb, the appearances, and the transformation of the disciples. (2) Without the evidence there is no historical argument. - Take away the testimony and the events, and what is left is not a weaker historical argument. It is no historical argument at all, just the general question of how likely Christianity is before anyone has looked. - Think of a court case. The prosecution's case is the evidence: the fingerprints, the receipt, the witness. Judging that case "on background knowledge" means judging it before the trial starts, while the jury is still finding its seats. Therefore, the historical argument's strength is the strength the historical evidence gives it.

(P2) Plantinga's calculation never adds the historical evidence to background knowledge. + The probabilities Plantinga multiplies are all probabilities given K, general background knowledge. The testimony about Jesus never joins K. (1) The ceiling is the probability of God on background knowledge. - Plantinga's first step is the probability that God exists given K. That is the kind of figure natural theology gives, reasoning from the cosmos and the moral law before anyone opens the Gospels, and he treated a high figure, around .9, as generous to the argument. - Every later step is then assessed given K and the earlier steps, and that first figure caps them all. (2) The evidence is held constant while the hypothesis grows. - On McGrew's analysis, the testimony, the empty tomb, and the appearances are never added to what the probabilities are "given." The hypothesis gets longer at every step, while the evidence stays where it started. - Think of checking a restaurant bill by estimating what dinner for four usually costs, without ever looking at the receipt. The estimate may be perfectly sensible. But the receipt is the evidence, and it never got read. (3) By Craig's report, Plantinga accepted the analysis. - William Lane Craig, reporting the 2006 exchange, says that Plantinga agreed with McGrew's analysis and explained that his objection had been to the way Swinburne first set out the argument. - That is Craig's summary, not a quotation from Plantinga, so it should be weighed as a report. Nor does (P2) rest on it alone. The premise rests on what the calculation takes as given, and the report simply shows Plantinga reading his own calculation the same way. (4) The result is a prior, not a verdict. - A prior probability is how likely a claim is before a particular piece of evidence is considered. A posterior probability is how likely it is after. - Plantinga's product is, in effect, a prior: how likely the full Christian story is on general knowledge alone. It is not the probability the historical argument claims to deliver, which is the posterior, the probability given the evidence. - Think of a doctor who estimates how likely a patient is to have a rare disease before running any tests. That estimate is useful, but it is not the diagnosis. Therefore, Plantinga's calculation measures the probability of Christianity before the historical evidence is weighed.

(C1) Therefore, Plantinga's calculation leaves out the evidence the historical argument rests on. + Put (P1) and (P2) side by side and this follows directly. The historical argument's strength is the evidence (P1), and the calculation never adds the evidence (P2). (1) The calculation answers a different question. - Plantinga's figure answers "How likely is Christianity on general background knowledge?" The historical argument asks "How likely is Christianity given everything we know, including the testimony about Jesus?" - A low answer to the first question is no answer to the second, any more than the doctor's estimate before the tests settles what the tests will show. Therefore, the dwindling calculation, as Plantinga set it out, does not assess the historical argument.

(P3) Evidence for a later step can raise the probability of the earlier steps. + The probability of God on background knowledge alone is not a ceiling on everything that comes after it. Evidence for a later step can reach back and raise the earlier ones. (1) Evidence for a later step can be evidence for an earlier one. - A fact is evidence for a claim when it is more probable if the claim is true than if it is false. The testimony and the empty tomb are far more probable if God exists than if he does not. - Here is why. If there is no God, a resurrection is all but ruled out, and every report has to be explained some other way: as error, as legend, or as something worse. If there is a God, a resurrection is one live explanation among them, a real candidate rather than a nonstarter. - So the resurrection evidence is evidence for God. It raises the first step in the chain, not just the fourth. - Think of a postcard from Rome in a friend's handwriting, all about a restaurant she found near the Pantheon. It is a postcard about dinner, but it is also, without trying to be, good evidence that she made it to Italy at all. (2) The resurrection sets a floor under God's existence. - One claim entails another when the first cannot be true without the second. "God raised Jesus" entails "God exists," since there is no raising without a God to do it. - That gives a floor. On any evidence at all, God's existence is at least as probable as the resurrection. Whatever probability the evidence gives the resurrection, it gives God's existence at least that much. - Think of a detective who finds a man's fingerprint on the safe. Before the print turned up, it seemed unlikely the man had even been in town. But he cannot have opened the safe without being in town, so as the print makes the later claim probable, it lifts the earlier claim with it. (3) The ceiling is set by the posterior, not the prior. - The conjunction rule holds at every stage. Before the evidence, the whole chain is capped by the prior probability of God. After the evidence, it is capped by the probability of God given the evidence, and that second cap is the one the historical argument is about. - Since the evidence can raise the probability of God, the ceiling rises with it. The probability of God on background knowledge alone is no ceiling on the probability of the resurrection once the evidence is in. - Picture the numbers. Suppose, just for illustration, that God's existence on background knowledge alone were a coin toss, 0.5, and the evidence lifted it to 0.95. The ceiling on the whole story is then 0.95, not 0.5. (4) Strong evidence can make a whole chain highly probable. - When a piece of evidence is far more likely if a story is true than if it is false, it can lift the probability of the whole story, every step at once. - A video of a robbery, showing a man's face, his car, and his license plate, does not make each link of the story a little more probable one at a time. It makes the whole story probable together. Therefore, evidence for the resurrection can raise the probability of every step in the chain, including the existence of God.

(C2) Therefore, the dwindling of probabilities on background knowledge does not show that the historical argument fails. + This follows from (C1) and (P3), with one link between them that is worth saying out loud. An estimate that leaves out evidence, when that evidence could raise it, is no verdict on the evidence. The calculation leaves the evidence out (C1), and the evidence can raise every step (P3), so a low product on background knowledge tells us nothing about how probable Christianity is given the evidence. (1) The historical argument has to be judged on its evidence. - Whether the historical argument succeeds depends on how strong the evidence actually is: on the testimony, its independence, its early date, and the alternatives. That is a question for history and probability together, and it cannot be settled by multiplying priors. - The doctor's diagnosis comes from the tests, not from the odds she quoted before running them. - None of this shows that the historical argument succeeds. That is a separate case, and it is made on its own evidence. (2) The conjunction rule itself stands. - The conjunction rule is a theorem of probability: a statement proved from the basic rules of probability, the way the angles of a triangle are proved to add up to 180 degrees. (C2) does not deny it. - What (C2) denies is that the rule, applied to probabilities before the evidence, measures the argument. Applied after the evidence, the rule still binds, and the case has to meet it there. Therefore, if (P1), (P2), and (P3) hold, Plantinga's principle of dwindling probabilities does not refute the historical case for Christianity, which stands or falls on its evidence. See also: Philosophy / Epistemology: Warrant Philosophy / Epistemology: Internalism Philosophical Theology / Prolegomena: Meta-Apologetics: Evidentialist Christian Evidences / Resurrection of Jesus: Maximal Data Method (McGrew)

Plantinga's argument and its revival: Crisp, Thomas M. "On Believing That the Scriptures Are Divinely Inspired." In Analytic Theology: New Essays in the Philosophy of Theology, edited by Oliver D. Crisp and Michael C. Rea, 187-213. Oxford: Oxford University Press, 2009. Plantinga, Alvin. "Historical Arguments and Dwindling Probabilities: A Response to Timothy McGrew." Philosophia Christi 8 (2006): 7-22. Plantinga, Alvin. "Rationality and Public Evidence: A Reply to Richard Swinburne." Religious Studies 37 (2001): 215-222. Plantinga, Alvin. Warranted Christian Belief. New York: Oxford University Press, 2000. The McGrews' replies and the historical argument: Craig, William Lane. "Problem of Dwindling Probabilities." Reasonable Faith, Question of the Week #691, 2020. reasonablefaith.org. McGrew, Lydia. "Of Generic Gods and Generic Men: The Limits of Armchair Philosophy of Religion." Journal of Analytic Theology 6 (2018): 183-203. McGrew, Timothy. "Has Plantinga Refuted the Historical Argument?" Philosophia Christi 6 (2004): 7-26. McGrew, Timothy, and Lydia McGrew. "On the Historical Argument: A Rejoinder to Plantinga." Philosophia Christi 8 (2006): 23-38. McGrew, Timothy, and Lydia McGrew. "The Argument from Miracles: A Cumulative Case for the Resurrection of Jesus of Nazareth." In The Blackwell Companion to Natural Theology, edited by William Lane Craig and J. P. Moreland, 593-662. Chichester: Wiley-Blackwell, 2009. Swinburne, Richard. "Natural Theology, Its 'Dwindling Probabilities' and 'Lack of Rapport'." Faith and Philosophy 21 (2004): 533-546. Swinburne, Richard. The Resurrection of God Incarnate. Oxford: Clarendon Press, 2003. Historical apologists: Butler, Joseph. The Analogy of Religion, Natural and Revealed. 1736. Paley, William. A View of the Evidences of Christianity. 1794. On probability and confirmation: Babbage, Charles. The Ninth Bridgewater Treatise: A Fragment. London: John Murray, 1837. Carnap, Rudolf. Logical Foundations of Probability. Chicago: University of Chicago Press, 1950. Earman, John. Hume's Abject Failure: The Argument Against Miracles. New York: Oxford University Press, 2000. Hempel, Carl G. "Studies in the Logic of Confirmation." Mind 54 (1945): 1-26, 97-121. Critic sources engaged: Hume, David. An Enquiry Concerning Human Understanding. First published 1748 as Philosophical Essays Concerning Human Understanding. Section 10, "Of Miracles." Biblical citations: Luke 1:2; Luke 1:4; Acts 17:31; 1 Corinthians 15:6; Titus 1:2. Scripture quotations are from the English Standard Version.
+ Plantinga replied that he never claimed to refute every historical argument. His target was a particular way of setting out the case, and the dwindling problem is real for that version.
1. Plantinga's narrower claim should be granted. In his 2006 response, Plantinga narrowed his target, and a brother's clarification of his own meaning should be taken at his word. By William Lane Craig's account of the exchange (P2), Plantinga agreed with McGrew's analysis, and said his objection was to the way Swinburne had originally formulated the argument, not to historical arguments in general. If that report is right, the two sides ended closer together than the 2000 book suggested. 2. The narrower claim is the McGrews' point. The McGrews' rejoinder answers that there are many historical arguments, not one. If the dwindling problem fits only a version in which the probabilities are assessed before the evidence, then it does not touch the versions that weigh the evidence, and weighing the evidence is what historical apologists have always done. Think of the family heading for the ferry again. Showing that one road to the coast floods at high tide says nothing about the other roads. 3. Swinburne denied the structure Plantinga described. That leaves one question: was even Swinburne's version the flooded road? Swinburne, whose case was the main target, replied in 2004 that his argument does not have the step-by-step structure the dwindling principle needs. - He feeds different pieces of evidence into the argument at different stages, so the evidence raises the probability of the later steps as it goes. - Run that way in his 2003 book, his own estimate, on all the evidence, that Jesus was God incarnate and rose from the dead came to about 97 percent. - Craig reports that Swinburne said he had intended a probability argument along McGrew's lines. So the principle, read as Plantinga clarified it, leaves the historical argument standing as its defenders actually make it.
+ The conjunction rule is a theorem. However much evidence there is, the whole Christian story can never be more probable than the bare claim that God exists.
1. The rule is true. This should be granted without hesitation: a conjunction can never be more probable than any of its parts. Nothing in the McGrews' reply denies it. 2. The rule applies to the probabilities after the evidence. Once the evidence is in, the whole story can be no more probable than God's existence given that evidence. Here is the thing worth noticing, though. God's existence given the evidence can be far higher than God's existence on background knowledge alone, because the resurrection evidence is far more probable if there is a God than if there is not (P3). - A man's alibi collapses, and the probability that he was at the scene rises. The probability that he committed the crime can never exceed the probability that he was there, but both have just gone up together. 3. The ceiling moves with the evidence. So the theorem does set a ceiling. That ceiling is the posterior probability of God, though, not the prior, and the dwindling argument looks like a refutation only when the prior ceiling is taken for the posterior one.
+ Thomas Crisp has restated the dwindling argument with more care, aimed at whether historical argument can show the Scriptures are divinely inspired. That claim has more links than the resurrection, and its probabilities still dwindle.
1. Crisp's version is more careful. In a 2009 essay, Thomas Crisp set out a refined version of the dwindling problem. His question is whether a person can come to know, through historical arguments, that the Bible is divinely inspired. The version deserves to be taken seriously, and its target is harder than the resurrection. A doctrine of Scripture is a longer claim than "God raised Jesus," with more links, and each of them needs support. 2. The same structural point applies. Lydia McGrew answered Crisp in 2018, arguing that his version evaluates a concrete historical argument a priori. The phrase is Latin for "from what comes before," and here it means judged in advance: each step is given its probability before the evidence is weighed. If that criticism is right, Crisp's version inherits the same problem as Plantinga's (C1). Whatever evidence a person actually has, that evidence has to go into what the probabilities are "given." 3. A longer claim needs more evidence, not a different method. This much should be granted outright: the more links a claim has, the lower its ceiling, and the resurrection evidence does not by itself support every link in a doctrine of Scripture. Each further link needs evidence of its own, such as what the risen Jesus taught about the Scriptures and whom he commissioned to speak for him. Think of a family proving an old will. Showing that the signature is genuine does not show that no sheet was swapped, so they check every sheet too. That makes the job longer. It does not make the will unprovable. 4. A thinner case is not a refuted one. It is true that a person who knows little history has less historical evidence than a scholar does. That shows the strength of the case varies with what a person has, which everyone grants. It does not show that the historical case, once the evidence is in, dwindles away. Think of a neighbor who read only a two-paragraph newspaper summary of a trial. She has less to go on than the juror who heard every witness, so her own case for the verdict is thinner. The evidence the jury heard has not shrunk.
+ Numerical probabilities cannot be assigned to historical and religious claims, so the whole exchange, on both sides, is an exercise in false precision.
1. Precise numbers are not needed. The McGrews' point does not depend on any number being exactly right. It depends on a structural fact: probabilities assessed without the evidence cannot measure an argument from the evidence. That holds whatever the numbers turn out to be. Guessing a restaurant bill without looking at the receipt is the wrong method whether the guess is forty dollars or four hundred. 2. Comparative judgments do real work. Historians and juries reason about how much more likely a piece of evidence is on one story than another without ever writing down a number. A juror can judge that a confession, a fingerprint, and a witness are far more likely if the defendant is guilty than if he is innocent. 3. Without numbers, the dwindling shrinks to what everyone grants. The conjunction rule needs no numbers at all. A long story of uncertain links is no more probable than its weakest link, and a skeptic of numbers can still say so. What goes missing without numbers is the size of the shrinkage. That matters, because Plantinga's conclusion was about size: the product, he argued, falls too low for faith. That conclusion needs the very numbers the objection forbids. What the objection leaves standing is the qualitative rule, the rule stated in words rather than figures. The historical argument already accepts that rule. It still has to be applied to the probabilities given the evidence (P3), with numbers or without them. Think of the father with no calculator. He can still tell that four things all going right is less likely than any one of them, but he cannot say whether they will make the ferry three times in four or one time in four.
+ Faith requires firm conviction, but historical evidence can only make a claim probable. So the historical argument could never ground faith, whatever the probabilities.
1. This is the real motive behind Plantinga's argument. Plantinga's deeper concern was that Christian belief is firm, and a firm belief cannot rest on an argument whose probability falls well short of it. Christians who share that concern are in good company, and it deserves an answer. Plantinga's own version of the worry leaned on the dwindled figure, the low product of his calculation. If the evidence makes the whole story highly probable (P3), most of the gap between that figure and firm belief closes. What remains is the stronger form of the worry: that no probability short of certainty will do. 2. High probability is how we know most of what we know. We know that Julius Caesar was killed in Rome and that Lincoln was president during the Civil War. Neither is certain in the way that two plus two is four. Both are known, and known firmly, on evidence. Joseph Butler, one of the historical argument's founders, put it in one line in 1736: "to us, probability is the very guide of life." Luke's word "certainty" (Luke 1:4) is the firm assurance that comes from investigated testimony, not mathematical proof. 3. Firm commitment can rest on strong evidence. A woman who marries a man after years of knowing him commits herself completely, though she cannot prove he will be faithful. Her commitment is wholehearted because her evidence is strong. Faith in Christ can be the same: a full trust resting on evidence that is very strong, not on a proof. 4. The evidence does not compete with the work of the Spirit. Plantinga holds that the Holy Spirit gives believers warrant directly, and the historical argument does not deny that the Spirit works in believers. The dispute is narrower: whether the historical case, taken on its own evidence, can reach a high probability. A Christian can say yes to both. When your phone rings and you know your sister's voice at once, the caller ID showing her number does not compete with your recognition. Two grounds for the same belief strengthen each other. See also: Philosophical Theology / Prolegomena: Meta-Apologetics: Reformed
+ The prior probability of a resurrection is so low that no amount of testimony could overcome it. Plantinga's figures were, if anything, too generous to the historical argument.
1. This is Hume's objection in a new setting. The Scottish philosopher David Hume argued in 1748 that our uniform experience of the dead staying dead weighs so heavily against a miracle that testimony could establish one only if the testimony's being false would be even more of a miracle, and that in practice it never is. It is a serious objection with a long history. The McGrews have answered it at length in their 2009 essay on the argument from miracles. Notice where it moves the debate. Plantinga granted God a high probability and argued from the shape of the chain. Hume attacks the evidence itself, and that is exactly where (C2) says the question belongs. 2. Independent testimonies compound. The McGrews' reply is that the evidential force of several independent witnesses (witnesses who did not copy one another or draw on the same source) multiplies. If each witness's testimony is far more likely if the event happened than if it did not, and the witnesses are independent, their combined force can overcome a very low prior. - A single smoke detector going off might be a fault. Five detectors in five rooms, wired separately, going off together is almost certainly a fire. - The numbers can be very large. In their 2009 essay the McGrews estimated, conservatively by their own account, that the testimony of the women, of the disciples, and of Paul, taken together, is more likely if Jesus rose than if he did not by a factor on the order of a one followed by forty-four zeros. - The mathematics is not special pleading, a rule invented to get one's own side off the hook. Charles Babbage, the mathematician remembered as a pioneer of the computer, made the same calculation against Hume in 1837, and John Earman, a philosopher of science writing with no apologetic aim, defended it in 2000. - Independence is the hinge, and critics contest it: if the Gospels all drew on one source, they count as fewer witnesses. That is a historical question with a substantial case behind independence, and it is argued on its own evidence. 3. The prior must include what is known about God. If there is a God who has reasons to act in history, the prior probability of a resurrection is not the same as the probability of a dead man rising by natural causes. Background knowledge that includes the case for God's existence raises it. A knock at the door at midnight is startling, unless a friend phoned an hour ago to say she was driving over. What you already know changes how surprising the event is. See also: Christian Evidences / Resurrection of Jesus: Maximal Data Method (McGrew) Christian Evidences / New Testament Criticism: Convergence of Independent New Testament Traditions Christian Evidences / New Testament Criticism: Undesigned Coincidences as Evidence of Gospel Reliability
+ The cumulative case for the resurrection multiplies many probabilities too. If multiplication dwindles for Plantinga, it dwindles for the cumulative case.
1. Two different things are being multiplied. Plantinga's calculation multiplies the probabilities of the steps in a hypothesis built from claims joined by "and," each assessed without the evidence. That product can never go up as steps are added. The McGrews' cumulative case (several independent lines of evidence weighed together) multiplies something else: the strength of independent pieces of evidence, meaning how many times more likely each piece is if the hypothesis is true than if it is false. That ratio is called a Bayes factor. The name comes from Bayes's theorem, the rule for updating a probability on new evidence, which the eighteenth-century minister Thomas Bayes worked out. A fingerprint a thousand times more likely to be on the safe if the suspect touched it than if he did not has a Bayes factor of a thousand. - Each Bayes factor multiplied in raises the evidential force only when it is greater than one, that is, only when that piece of evidence really is likelier if the hypothesis is true. A factor below one (evidence that fits the hypothesis worse than its denial) pulls the total down, and a factor of exactly one changes nothing. - The pieces must also be independent in the sense that matters: once you know whether the hypothesis is true, learning one piece does not make the next any likelier. Two witnesses who compared notes do not count twice. 2. The two products move in opposite directions. - Probabilities short of certainty are fractions less than one, and multiplying them is like taking a discount off a discount: ten percent off, then ten percent off again, leaves 81 percent of the price. - Bayes factors for evidence that favors a hypothesis are greater than one, and multiplying those is like interest compounding. Three independent pieces of evidence, each ten times more likely if the hypothesis is true than if it is false, are together a thousand times more likely on the hypothesis than off it (10 × 10 × 10). - Both are multiplication. One works like a discount and the other like interest. 3. Both products belong in a full assessment. A complete probability of the Christian story weighs the prior probability of the hypothesis, steps and all, against the combined strength of the evidence. Plantinga's calculation supplies only the first half. - Odds compare the chances for a claim with the chances against it: three to one in favor means three chances it is true for every one it is false. Put in odds, the rule is simple. The odds after the evidence equal the odds before it, times the combined Bayes factor. A hypothesis that starts at a thousand to one against, met by evidence with a combined factor of a million, ends at a thousand to one in favor. - So the McGrews' case does not escape the dwindling prior. It has to outweigh it, and whether the evidence is strong enough to do that is the historical question (C2).
+ Plantinga's background knowledge is everything we know, the historical evidence included. On that reading nothing was left out, and his figures are simply estimates given all the evidence.
1. McGrew saw this reading coming. This is the most natural way to defend the 2000 calculation against (P2), and it deserves a straight answer. If K contained everything we know, each figure in the calculation would already be a judgment on all the evidence, and nothing in its form would leave the testimony out. In outline, Timothy McGrew's 2004 paper answers with a dilemma: there are two ways to read K, and each one makes trouble for the calculation. - On a synchronic reading (from the Greek for "at the same time"), K is everything we know at one moment, all the evidence weighed at once. A judge who reads the whole file before deciding anything is reasoning synchronically. - On a diachronic reading ("through time"), K is where the argument starts, with general knowledge and natural theology, and the historical evidence is added in stages. That is how Swinburne meant his argument to run. - On the diachronic reading the calculation goes wrong in the way (P2) describes, because it never adds the evidence. The objection takes the synchronic reading instead, and that reading has troubles of its own. 2. Then the first figure loses its grounds. On the synchronic reading, McGrew argues, Plantinga has no grounds for calling his figure for God generous. The figure is now supposed to be the probability of God given everything, the testimony about Jesus included, so it can no longer be borrowed from natural theology. And the one thing it would now have to reflect, what the resurrection evidence does to the probability of God (P3), is exactly the thing nobody has yet weighed. Think of a forecaster who says, "Seventy percent chance of rain, given everything we know." If she set that figure before the radar image came in, and the radar shows a wall of storms twenty miles off, her seventy percent is not "everything we know." It is her forecast from before she looked. 3. Then the dwindling stops being a structural objection. If every figure is a judgment on all the evidence, then a low product is simply the claim that the evidence is weak at one step or another. That may be argued. But it has to be argued from the testimony, the tomb, and the rival explanations, which is the historical argument's own ground (C2), and the multiplication adds nothing to it. A lawyer who thinks a jury got the verdict wrong argues about the fingerprints and the witnesses. Pointing out that guilt requires several things to be true at once is no argument at all. 4. Plantinga did not take this route. By Craig's account of the 2006 reply (P2), Plantinga accepted McGrew's analysis rather than defending his figures as judgments on all the evidence.
+ Add the evidence and the steps still multiply. Grant each step a generous .9 given the evidence, and six steps still come to about .53, little better than a coin toss.
1. The arithmetic is right. Six steps at .9 do multiply to just over .53. The conjunction rule binds after the evidence just as firmly as it did before, and (C2) never claimed otherwise. This is the honest form of the dwindling worry, and it has to be met on the evidence rather than waved away. 2. Steps that a later step entails cost nothing. "God raised Jesus" entails "God exists," so "God exists, and God raised Jesus" is exactly as probable as "God raised Jesus" alone. Any earlier step the resurrection entails drops out of the multiplication. It adds no risk that the resurrection claim did not already carry. Think of the ferry again. Once the family is standing on the deck, the question whether they reached the dock is not a further ninety-percent gamble. It is settled by their being there. 3. Later steps can be nearly certain given the earlier ones. The multiplication rule multiplies each step's probability given the steps before it, and some of those are very high. Jesus taught in God's name and made extraordinary claims about himself. If God then raised him from the dead, the act reads as God's endorsement, and it is hard to see what else it could be. - The New Testament reads it exactly that way: God "has given assurance to all by raising him from the dead" (Acts 17:31). Swinburne makes the same case philosophically: the resurrection authenticates Jesus' teaching the way a signature authenticates a letter. - The warrant is God's truthfulness. A God "who never lies" (Titus 1:2) would not perform the one act that most looks like endorsement for a man who spoke falsely in his name, and so mislead everyone who took the act at face value. - Very high is not certain, and this link leans on another: that the Gospels report what Jesus actually taught, a historical question with evidence of its own. A genuine notary's seal on a contract works the same way. If the seal is real, the probability that the notary witnessed the signing is not a separate gamble. That is what the seal means. 4. The real chain is short. What remains to be weighed is chiefly the probability, on the evidence, that God raised Jesus, and the link from that event to the truth of what Jesus taught. Dwindling bites hardest on long chains of loosely joined links, and least where each link is nearly certain given the one before. How high the probability of the resurrection stands on the evidence is the historical question (C2), and it is not settled by counting steps. See also: Christian Evidences / Resurrection of Jesus: The A.L.I.V.E. Argument (McGrew)
+ Evidence for a claim is not always evidence for what it entails. Confirmation theory has long shown that evidence can raise a hypothesis while lowering one of its consequences, so evidence need not flow backward as (P3) says.
1. The objection is correct as a general rule. Confirmation theory is the study of how evidence supports a theory. In 1945 the philosopher of science Carl Hempel proposed that whatever confirms a hypothesis also confirms everything the hypothesis entails. The philosopher Rudolf Carnap showed that this fails when "confirms" means "makes more probable." The objection is right about that, and it should be granted plainly. Think of drawing one card from a shuffled deck. Learning that the card is a heart raises the probability that it is the ace of hearts, from 1 in 52 to 1 in 13. Now, the ace of hearts entails "the card is an ace or a spade," since it is an ace. Yet learning that the card is a heart lowers the probability of "an ace or a spade," from 16 in 52 (the four aces plus the twelve other spades) to 1 in 13 (only the ace of hearts is left). 2. The first support for (P3) does not rest on that rule. (P3) has two supports, and each stands without the other. The first is direct (P3, point 1). The resurrection evidence counts for God because it is far more probable if there is a God than if there is not, and it would be so even if entailment carried nothing backward. - In the card case, the new fact rules out most of the ways the consequence could be true (the spades and the other three aces). That is why the consequence falls. - The resurrection evidence does nothing like that to God's existence. Suppose God exists but did not raise Jesus: then the testimony needs much the same natural explanation it would need if there were no God. So, taken as a whole, God's existence makes the evidence at least as likely as atheism does, and more likely to the degree that a resurrection is a live possibility if there is a God. 3. The second support, the floor, never fails. The second support is the floor (P3, point 2), and it needs no help from the first. Whatever happens to confirmation, a consequence is always at least as probable as anything that entails it. Given the evidence, "God exists" is at least as probable as "God raised Jesus." - Even in the card case, once you know the card is a heart, "an ace or a spade" stands at exactly 1 in 13, the same as the ace of hearts and never below it. - So once the evidence makes the resurrection more probable than God's existence was on background knowledge alone, God's existence must rise at least that far with it. The prior figure for God is then no ceiling, which is all (P3) needs.

Metaphysics: Free Will

Whether Our Choices Are Up to Us

Public Theology

Theological Analyses of Societal Issues

Theology of the Family

Where Faith and Family Intersect

Biographies

Notable Works & Great Quotes from Key Figures

Contemporary Voices

1960 AD – present
Playlist
Justified True Belief