VI. Why Necessary?

Why is this elaborate scheme necessary? Why these supernaturally inspired writings and this individually applied supernatural testimony of the Holy Spirit? Or rather (because God could have accomplished his aim of enabling human beings of many times and places to know about the possibility and means of salvation in many different ways), what might recommend this particular scheme? Wouldn’t some less extravagant means suffice? Couldn’t this information come to us just as well by way of ordinary human testimony, for example? Perhaps (as Locke thought) God could have revealed the great truths of the gospel in some direct way only to certain human beings. They could then write them down for the benefit of the rest of us, who are then supposed to be able to see in the ordinary way that these writings do, indeed, constitute divine revelation (and are accordingly both true and to be believed). Why have any truck with special faculties or supernatural belief-producing processes like faith and the internal instigation of the Holy Spirit?

Well, first of all, we have no reason to think God either specially prizes ontological economy or specially dislikes supernatural processes. But the main problem with Locke’s appealingly simple device is that it wouldn’t work. First, according to the extended A/C model, 269we human beings, apart from God’s special and gracious activity, are sunk in sin; we are prone to hate God and our neighbor; our hearts, as Jeremiah said, are deceitful above all things and desperately corrupt. In this context, that fact is of great importance: without some special activity on the part of the Lord, we wouldn’t believe. As the apostle Paul says, “The man without the Spirit does not accept the things that come from the Spirit of God, for they are foolishness to him, and he cannot understand them, because they are spiritually discerned.”³⁴¹³⁴¹   1 Corinthians 2:14. Compare 1 Corinthians 1:23–24: “We preach Christ crucified, a stumbling block to Jews and foolishness to Gentiles, but to those whom God has called, both Jews and Greeks, Christ the power of God and the wisdom of God.” We human beings won’t come to see the depth of our own sin and our need for salvation without regeneration, rebirth; according to Jesus himself, we need the testimony of the Holy Spirit to come to believe the great truths of the gospel.³⁴²³⁴²   “No one can come to me unless the Father who sent me draws him” (John 6:44); and “I will pray the Father, and he will give you another Counselor, to be with you forever, even the Spirit of truth, whom the world cannot receive, because it neither sees him nor knows him; you know him for he dwells with you, and will be in you” (John 14:16–17). Given our fallen nature and our natural antipathy to the message of the gospel, faith will have to be a gift, not in the way a glorious autumn day is a gift, but a special gift, one that wouldn’t come to us in the ordinary run of things, one that requires supernatural and extraordinary activity on the part of God.³⁴³³⁴³   This, once more, is a point on which Calvin and Aquinas concur: “for since, by assenting to what belongs to faith, man is raised above his nature, this must needs come to him from some supernatural principle moving him inwardly; and this is God” (STII-II, q.6, a.1, respondeo; see also article 2). When I speak of supernatural activity on the part of God, I don’t mean to suggest a sort of incursion into or intervention in the natural order. The fact is, God is constantly active in the world: apart from his upholding activity, the world would disappear like a candle flame in a high wind. Supernatural activity on the part of God (as well as miracles) must be understood, instead, in terms of God’s special activity, as opposed to the way in which he ordinarily treats the things he has created. There are depths and problems here; they will have to await another occasion.

Furthermore, suppose someone did come to believe, just by way of historical investigation that Jesus was indeed the divine son of God, that he died for our sins and rose again, and that through him we can have eternal life. Merely believing this—as an interesting fact about the world, rather like the fact that the universe began in a Big Bang some twelve to sixteen billion years ago—is insufficient. These truths must be sealed to the heart, as well as revealed to the mind. This sealing is the topic of the next chapter; now we note only that coming to faith includes more than a change of opinion. It also (and 270crucially) includes a change of heart, a change in affection, in what one loves and hates, approves and disdains, seeks and avoids. According to the present model, faith is, indeed, a belief-producing process; it is also an affection-producing process, a process issuing in alteration of affection as well as change of opinion. Given our constitution, this alteration of affection can’t be accomplished just by coming to believe, as a historical fact, the main lines of the gospel.³⁴⁴³⁴⁴   “If they wouldn’t believe Moses and the prophets, they will not be convinced even if someone rises from the dead” (Luke 16:31).

We therefore need a change of attitude in addition to a change of opinion, and won’t sustain the latter without the former. Well then, why couldn’t God (supernaturally if he feels that is necessary) just arrange for a change of attitude and affections? Why do we need that supernatural source for a change in opinion? Given the right affections, wouldn’t Scripture and our ordinary faculties (reason, memory, perception, sympathy, induction, etc.) be sufficient to enable us to see the truth of the message of the gospel?

I doubt it. First, what is proposed is such that by virtue of the ordinary faculties employed in historical investigation, only a few people would acquire the knowledge in question, and they only after a great deal of effort and much time; furthermore, their belief would be both uncertain and shot through with falsehood.³⁴⁵³⁴⁵   As Aquinas says about the existence of God if accepted on the basis of the theistic proofs (ST I, q.1, a.1, respondeo; Summa contra Gentiles, Bk. I, ch. 4; ST II-II, q.2, a.4). It is for this reason, Aquinas says, that it was entirely appropriate for the existence of God to be proposed as an object of belief or faith, even though it could in fact be proved by reason. What is being taught, after all, is not something that chimes straightforwardly with our ordinary experience. It isn’t like an account of an ancient war, or of the cruelty of the Athenians to the Melians, or of the overweening pride of some ancient despot. That sort of thing would be easy enough to believe. What we have instead, however, is the claim that a certain human being—Jesus of Nazareth—is also, astonishingly, the unique divine Son of God who has existed from eternity. Furthermore, this man died, which is not uncommon, but then three days later rose from the dead, which is uncommon indeed. Still further, it is by way of his atoning suffering and death and resurrection that we are justified, that our sins are forgiven, and that we may have life and have it more abundantly. This is heady stuff indeed, and the mere fact that some ancient authors believed it would certainly be insufficient for a sensible conviction on our part. As biblical scholars remind us, there are many ancient books with stories more or less (in my opinion, mostly less) like the biblical ones; how many of those ancient books do we in fact believe?

Still (comes the reply), can’t we discover for ourselves, without any special divine aid or assistance, that the Bible (the New Testament, say) is in fact “from God”: divinely inspired in such a way that God speaks to us in it and through it,³⁴⁶³⁴⁶   Perhaps in the way proposed by Nicholas Wolterstorff’s Divine Discourse. As I understand Wolterstorff, however, his account of how it could be that God speaks presupposes the main lines of Christian teaching, and hence wouldn’t offer a way in which we could come to see that that teaching is in fact true (i.e., wouldn’t provide the materials for an argument for the truth of that teaching). and hence wholly reliable?³⁴⁷³⁴⁷   This would substantially be the Locke-Swinburne model; however, it isn’t clear whether they would concur with the current proposal in the suggestion that a change in affection and attitude is necessary to a proper appreciation of the historical case. Can’t we come to see this in the same way that we can learn that Herodotus and Xenophon are reasonably reliable reporters of what they hear and see? And once we see that, couldn’t we then infer that the Bible’s central message of incarnation and atonement is true? Can’t we see and appreciate the historical case for the truth of the main lines of Christian belief without any special work of the Holy Spirit? “You must be born again” all right—your affections, aims, and intentions must be recalibrated, redirected, reversed—and that requires special divine help. But given that recalibration, couldn’t you then see and appreciate the historical case for the truth of the main lines of Christianity without any special work of the Holy Spirit?

I don’t think so. Even discounting the effects of sin on our apprehension of the historical case, that case isn’t strong enough to produce warranted belief that the main lines of Christian teaching are true—at most, it could produce the warranted belief that the main lines of Christian teaching aren’t particularly improbable. For how could such a case go?³⁴⁸³⁴⁸   What follows is roughly based on Richard Swinburne’s argument for a similar conclusion in his Revelation (Oxford: Clarendon Press, 1992), chapters 5, 7, and 8. A difference is that Swinburne thinks one believes that p if and only if one believes that p is more probable than not. (Faith and Reason [Oxford: Clarendon Press, 1981], p. 32); I take it that belief that p is more probable than not is nowhere nearly sufficient for belief that p. (I am about to throw an ordinary die: I believe it is more likely than not that it won’t come up showing face 2 or 3, but I certainly don’t believe that it won’t; what I actually believe on this head is only that it will come up showing one of faces 1 through 6 (and not, for example, wind up delicately balanced on one of its points or edges). First, of course, the case in question couldn’t in any way rely on the thought that the Bible is in some special way inspired by God; for these purposes, we should have to treat it exactly as we would any other ancient volume. We should have to follow the example of those Scripture scholars who try to determine (for 272example) what actually happened with Jesus—what he preached, whether he rose from the dead—without making any special theological assumptions about the reliability of the Bible or the person of Jesus.³⁴⁹³⁴⁹   See below, pp. 390ff. They bracket any such theological beliefs they may have and then try to assess the historical case or evidence for such claims as that Jesus actually asserted that he was the divine redeemer, or the claim that he died and came back to life. Such a case for the truth of the main lines of Christianity could be at most a case for the probability that these teachings are true.

What would such a case be like? How could it be constructed? The conclusion of the case (or argument) would be that the central Christian claims are probable. Now a proposition is probable only with respect to some other proposition or propositions.³⁵⁰³⁵⁰   The absolute or logical probability of a proposition would then be its probability with respect to a necessary truth. In this case, the relevant other propositions would be some body of background knowledge K—what we all or nearly all know or take for granted or firmly believe, or what at any rate those conducting the inquiry know or take for granted or believe.³⁵¹³⁵¹   These probabilities would not be Bayesian measures of degrees of belief, but something much more objective—Richard Swinburne’s epistemic probability, or the objective probability of WPF (pp. 161ff.). And the aim would be to show that the claims of the Christian gospel are probable with respect to K—that is, probable with respect to what we know or take for granted. For simplicity, take the central Christian claims to be sin (human beings are in need of salvation), incarnation (Jesus is the incarnate second person of the trinity), atonement (by virtue of his suffering and death, he atoned for our sin and enables us to attain eternal salvation), and general availability (salvation isn’t restricted to just one group of people, for example, the Jews³⁵²³⁵²   Peter’s vision in Acts 10.); and let’s use ‘G’ to name their conjunction. Our aim, therefore, is to argue that G is reasonably probable on K; we can employ the usual symbolism for probability and put this by saying that P(G/K) is reasonably high.

How can we construct such a case—argue that P(G/K) is reasonably high? The usual way (and the method followed by Swinburne) is to try to find some proposition (or group of propositions) P which is probable with respect to K, and which is such that G is probable with respect to its conjunction with K: that is, a proposition P such that P(P/K) and P(G/P&K) are both high. For example, you might argue first that T, the existence of God, is probable on K, our background knowledge. Then you might argue that given our background knowledge K and the existence of God (T), it is probable that God 273would reveal certain crucial truths (truths we need to know) to humankind.³⁵³³⁵³   Thus Swinburne: “So if there is other evidence which makes it quite likely that there is a God, all powerful and all good, who made the Earth and its inhabitants, then perhaps it becomes to some extent likely that he would intervene in human history to reveal things to them” (Revelation, p. 70). Call that proposition R. Then you might continue arguing in the same vein (repeating the same form of argument), finally winding up with some propositions with respect to which it is likely that God raised Jesus from the dead, thus authorizing and validating the message of the New Testament. That message could then be taken as authorized by God and hence true; and the message contains those propositions G to whose probability we are trying to argue. So you might then conclude that G is in fact probable with respect to what we know.

To illustrate and explain this procedure, suppose you are interested in the probability that Eleonore is at the party. It is very probable, on your background knowledge K, that Paul is at the party (call that proposition ‘P’): P(P/K) is high—for definiteness, say it’s .9. It is also very likely that Eleonore is at the party (call that proposition ‘E’), given that Paul is (she ordinarily goes to every party he goes to); so P(E/P&K) is also high—say it is also .9. There is a formula from the probability calculus that enables you to conclude that it is likely that Eleonore is there, too:

P(E/K) equals or exceeds P(P/K) × P(E/P&K).

Thus P(E/K) will be at least .81.³⁵⁴³⁵⁴   “At least”: that is because there could also be some probability that Eleonore would be there even if Paul were not. The probability of Eleonore’s being there (E) will be the weighted average of the probabilities of E given Paul’s being there (P) and the probability of E given -P—weighted by the probabilities of P and -P. The relevant formula is
   P(E/K) = [P(E/(P&K)) × P(P/K)] + [P(E/(-P&K)) × P(-P/K)].

This sort of argument can be reiterated. Perhaps you also know that there is a pretty good probability—.8, say—that Vonnie will be there, given that Eleonore is: in that case, you can conclude that the probability of Vonnie’s being there is at least .648; and perhaps you know that the probability of Jim’s being there, given that Vonnie is there, is .95; then the probability that he’s there will be at least .616.

Now suppose we try along these lines to construct a case for the probability of G with respect to that background knowledge K. We should first have to find the probability that T (theism) is true: what is the probability (on our background knowledge, or the totality of what we know apart from theism) that there is an omnipotent, omniscient, wholly good being who has created the world? In his book The 274Existence of God,³⁵⁵³⁵⁵   Oxford: Clarendon Press, 1979. Swinburne considers this probability and concludes on the last page of the book, “On our total evidence theism is more probable than not.” The argument is complex and at many points controversial.³⁵⁶³⁵⁶   Especially, perhaps, with respect to the judgments of comparative simplicity involved, and the judgment that simplicity is, in fact, a good guide to probability. From the present perspective, however, an even more vexing problem is that its conclusion is only that theism is more probable than not on the relevant body of knowledge or information K: it lies somewhere in the (half open) interval .5 to 1. Even if all the other probabilities involved in our historical case were as high as 1, we could conclude no more than that the probability of the truth of Christian teaching lies somewhere in that same interval.

But if my only ground for Christian teaching is its probability with respect to K, and all I know about that probability is that it is greater than .5, then I can’t rationally believe that teaching. Suppose I know that the coin you are about to toss is loaded. I don’t know just how heavily it is loaded, so I don’t know what the probability is that it will come up heads, but I do know that this probability is greater than .5. Under those conditions I do not believe that the next toss of this coin will come up heads. (Of course I also don’t believe that it will come up tails; and I suspect that it will come up heads.) All I know is that it is more likely than not to come up heads; and that’s not sufficient for my sensibly believing that it will. The same goes in this case: if what I know is only that the probability of Christian belief (with respect to K) is greater than .5, I can’t sensibly believe it.³⁵⁷³⁵⁷   Note, of course, that we can’t simply add theism to the relevant body of knowledge K on the grounds that it is more probable than not on what we know; that way lies contradiction. It is more probable than not that this die will not turn up ace; the same, of course, for each of the other five possibilities; so if we could add each of these propositions (it won’t come up 1, it won’t come up 2 . . . ) to K, we wind up with the contradiction that the die will come up showing some number between 1 and 6 (inclusive) and also that it will not. I can hope that it is true, and think it rather likely that it is; I can’t believe it. To give the historical case for G a run for its money, therefore, suppose we arbitrarily assign T a much higher probability on K—let’s say that it is at least .9. Many will howl with indignation at such a high assignment; let us ignore them for the moment.

We must next consider the probability, given T&K, that

A God would make some kind of revelation (of himself, or perhaps of what we need to know about him) to humankind.

Well, that seems quite likely, although of course it’s very hard to predict a priori what God would or wouldn’t do. Again, let’s be generous and estimate this probability as also lying in the interval .9 to 1.

But now we come to the hard parts. Somehow we have to make a probabilistic argument for the proposition that such a revelation would contain G, the great claims of the gospel. Of course a revelation from God would include G only if G is true; so what we really need here is a probabilistic argument for a conclusion sufficient to entail G. One common way to do this would be to argue that it is likely that Jesus taught G, and that by raising Jesus from the dead God endorsed or ratified that teaching. But just on the basis of ordinary historical scholarship, without the assumption that the Bible is, in fact, a divine revelation, it really isn’t likely that Jesus taught anything nearly as definite as G—that is, sin, incarnation, atonement, and general availability. Scripture scholars argue at length about what precisely Jesus taught, but those who approach the matter ‘from below’ (i.e., without employing any special theological assumptions) for the most part are not at all prepared to assert that Jesus taught G. Indeed, even if we do accept the Bible as authoritative, it still won’t be clear that Jesus taught G; much of our grasp of the central claims of Christian faith comes from other parts of the Bible (e.g., the Pauline epistles) and later reflection (e.g., the Nicene Creed).

Perhaps, though, it is likely just on historical grounds that the teachings of Jesus were such that by sensible interpretation and extrapolation one could arrive at G. So we must ask after the probability of

B Jesus’ teachings were such that they could be sensibly interpreted and extrapolated to G

given K&T and A; that is, we must ask after the value of P(B/(K&T&A)). B is fairly vague, but let’s suppose it’s rather likely, just on the basis of historical scholarship. Of course there will be many who would demur—those who think Jesus was a homosexual magician,³⁵⁸³⁵⁸   See Morton Smith, Jesus the Magician (New York: Harper and Row, 1978). for example, not to mention those who think he was the first Christian atheist.³⁵⁹³⁵⁹   See Thomas Sheehan, The First Coming (New York: Random House, 1976). Let’s say they are wrong and that this probability is high—for definiteness, in the interval .7 to .9.

But now things get harder yet. We must next consider the proposition that God endorsed Jesus’ teachings by performing a great miracle and raising him from the dead. What is the probability, just on historical grounds, that

C Jesus rose from the dead?

Of course, C must be taken in a literal and bodily sense; it is not to be glossed as, for example, the mere thought that the followers of Jesus underwent some experience so impressive and revivifying that they 276acquired the energy and determination necessary to start a new religion.³⁶⁰³⁶⁰   As in much contemporary liberal and quasi-liberal theology. See, e.g., Norman Perrin, The Resurrection according to Matthew, Mark, and Luke (Philadelphia: Fortress Press, 1977), p. 83: of the witnesses to the resurrection appearances of Jesus, he says, “in some way they were granted a vision of Jesus which convinced them that God had vindicated Jesus out of His death and that therefore the death of Jesus was by no means the end of the impact of Jesus upon their lives.”
   Here we can bracket the question what sort of body Jesus had upon resurrection: was the body he had upon resurrection (whether or not it was numerically the same body he had before his death) a glorified body with supernatural powers? The latter would be still harder to establish by historical argument (see Robert Cavin’s “Is There Sufficient Historical Evidence to Establish the Resurrection of Jesus?” Faith and Philosophy [July 1995]). And again, what we need to consider is the conditional probability of C on K, T, and A&B—that is, P(C/(K&T&A&B)). What is this probability? One hesitates to say much here, given the enormous controversies and disagreements among Scripture scholars. How many people are there who believe on strictly historical grounds together with theism (no help from theology, or the internal instigation of the Holy Spirit, or anything like that), that Jesus Christ arose from the dead (in the strict and literal sense)? Even if you had a fine command of the vast literature and thought there was rather a good historical case here, you would presumably think it pretty speculative and chancy. I’d guess that it is likely that the disciples believed that Jesus arose from the dead, but on sheerly historical grounds (together with the assumption that there really is such a person as God, who is rather likely to make a revelation to us) it is considerably less likely that this actually did happen. Given all the controversy among the experts, we should probably declare this probability inscrutable—that is, such that we can’t really say with any confidence what it is. Again, let’s be generous: let’s say that this proposition is more probable than not—for definiteness, say it lies in the interval .6 to .8.

Next, we must consider the probability of

D In raising Jesus from the dead, God endorsed his teachings

on the previous propositions; that is, we must consider P(D/(K&T&A&B&C)). From C we have only that Jesus arose from the dead, not that God raised him from the dead, thereby endorsing his teaching. Given T, though, it does seem likely that God raised him from the dead—how else would it happen? Still, did he, in so doing, ratify what Jesus taught? Not necessarily: there are other reasons why he might have done it. Perhaps it was to endorse the teaching of the Pharisees as opposed to the Sadducees (Matthew 22:23), or perhaps as a reward for special devotion and a holy life, or perhaps for some reason of which we have no knowledge. Still, this probability should probably be pegged fairly high: let’s say, for definiteness, .9.

But there is still another probability to be evaluated here: the probability that in raising Jesus from the dead and endorsing his teachings, he was also endorsing their extrapolation to G, the central teaching of Christianity: we must look into the probability of

E The extension and extrapolation of Jesus’ teachings to G is true

That is, we must look into P(E/(K&T&A&B&C&D)). Here the issues are more complex than they appear at first. Suppose you were completely convinced, on merely historical grounds, that Jesus rose from the dead: wouldn’t it be an enormous further step to conclude G, that he was, in fact, the divine and unique son of God, the second person of the trinity, and that his suffering and death is a propitiatory sacrifice, whereby we can have eternal life? It isn’t easy to see how a powerful historical case for all this could be made; perhaps it could go as follows. In accord with B, above, Jesus’ teachings can naturally be extrapolated or extended to G; and perhaps God endorsed this extension of Jesus’ teachings in raising him from the dead. But why think so? Why think that extrapolation (as opposed to all the other possibilities) has it right? Well, perhaps Jesus intended to (and did) found a church to interpret and preserve his teachings; God ratified that intention too; the church he founded is still extant, preserved (by the Holy Spirit, perhaps) from error, and teaches G. Here we really have five further propositions that together constitute our historical case for E; so, instead of E, we must consider the probability of the conjunction of

(1) Jesus intended to (and did) found a church to interpret and preserve his teachings,

(2) God ratified that intention in raising him from the dead,

(3) The church Jesus founded is still extant,

(4) God has preserved that church from error,

and

(5) That church teaches G

on K&T&A&B&C&D.

Now in the context of the present argument, we can take the conjunction of (1) to (5) as E*. Our present project, then, is to evaluate the probability of E*, so construed, on K&T&A&B&C&D. It seems sensible to estimate P((1)&(2)&(3)&(4)/(K&T&A&B&C&D)) as very 278high:³⁶¹³⁶¹   Celsus, an early critic of Christianity, apparently thought this probability fairly low, not much greater than .5 (see Origen, Contra Celsum, 1.68); let’s suppose Celsus was wrong. to be generous (and keep things as simple as possible) let’s say this probability is 1. That still leaves us with (5), however: what is its probability on K&T&A&B&C&D plus the conjunction of (1) to (4)? This is not easy to estimate. Given that there is a church that God has preserved from error, which church is it? Is it one that teaches G? At present, many mainline Protestant churches (and some Roman Catholic clergy), for example, don’t seem really to teach G at all. These churches (and their members) display a very wide spectrum of opinion, ranging all the way from extremely liberal views, according to which very little of classical Christianity is actually true (though much of it perhaps warmly inspiring), to full-blooded classical Christian belief. Which of these opinions did Christ mean to endorse? Which of these most faithfully conforms to his intentions? Is it a group that actually teaches G?³⁶²³⁶²   Swinburne (Revelation, chapter 8) proposes two criteria for determining what is to count as the church: continuity of aim and continuity of organization. The first depends on continuity of doctrinal teaching; but then to apply it we would already have to know what the true church teaches. That is, we would have to know what Jesus intended his church to teach; but then we can’t use this test to determine what Jesus intended his church to teach. That’s not easy to say on historical grounds; once again, let’s be generous and estimate this probability (i.e., P((5)/(K&T&A&B&C&D)) as somewhere in the interval .7 to .9. This means that P(E/(K&T&A&B&C&D)) will lie in that same interval.

Now how do we get a probability (on K) for G, given all this? Note that E entails G; so (following our present argument) to find the probability of G on K, what we need is to find the probability of E on K. How do we do that? Our argument followed the strategy of finding a series of propositions, T and A-E, such that the first is probable on K, the second on K together with the first, the third on K together with the first and second, and so on. A little arithmetic enables us to conclude that

P(E/K)

will be equal to or greater than

P(T/K) × (P(A/(K&T)) × P(B/(K&T&A)) × P(C/(K&T&A&B)) × P(D/(K&T&A&B&C)) × P(E/(K&T&A&B&C&D)).

The little arithmetic goes as follows.

By

(1) P(X/Y) ≥ P(X/Z&Y) × P(Z/Y)

we know that

(2) P(E/K) ≥ P(E/K&T&A&B&C&D) × (P(T&A&B&C&D/K).

Consider the right multiplicand. According to the probability calculus,

(3) P(X&Y/Z) = P(X/Z) × P(Y/X&Z);

hence

(4) P(T&A&B&C&D/K) = P(T&A&B&C/K) ×
P(D/T&A&B&C&K).

By substitution into (2) we have

(5) P(E/K) ≥ P(E/T&A&B&C&D&K) × P(T&A&B&C/K) ×
P(D/T&A&B&C&K).

Again, by (3) we know that

(6) P(T&A&B&C/K) = P(C/T&A&B&K) × P(T&A&B/K);

substituting into (5), we have

(7) P(E/K) ≥ P(E/T&A&B&C&D&K) × P(C/T&A&B&K) ×
P(T&A&B/K) × P(D/A&B&C&K).

Applying (3) and substitution a couple of more times and rearranging terms, we have

(8) P(E/K) ≥ P(T/K) × P(A/K&T) × P(B/K&T&A) ×
P(C/K&T&A&B) × P(D/K&T&A&B&C) ×
P(E/K&T&A&B&C&D),

which was to be demonstrated.

In some cases these values were intervals rather than real numbers, sharp probabilities. That’s no problem; since we are in any event winding up with the statement that P(G/K) is equal to or greater than some number, what we do is just use the lower bounds of the intervals. Doing the arithmetic, in the present case we wind up with the proposition that P(E/K) is at least .21. If instead of using just the lower bounds, we use the midpoints of the intervals assigned, we find that P(G/K) is at least .35. Suppose we stick with the midpoint (rather than the lower bound): then our argument entitles us to say only that the probability of G on K is at least .35. It could be higher, of course, but all we can say with confidence, given the argument, is that it is equal to or greater than .35.

Now of course it is ludicrous to assign real numbers to these probabilities: there is vagueness of many kinds here. Not only can’t we sensibly assign a real number to any of these probabilities, it also seems wrong to assign them intervals with sharp boundaries; our actual reasoning must be vaguer. Perhaps the best we can really say is that these probabilities are high, or low, or fairly near .5. Still, our reasoning, even if vague, would have to be guided here roughly and vaguely by the calculus of probabilities; and the best way to let it be thus guided is to assign probabilities (and intervals of probability) that comport with the vague estimates we seriously make, and then 280see what the consequent probabilities would be. When we do this in the present case, in our attempt to estimate the power of a historical argument for G, an argument that doesn’t rely on faith or any special theological assumptions, what we can say is only that this probability is at least high enough not to be a whole lot less likely than its denial. Of course we might quibble with the specific values I proposed. But I tried to err on the side of generosity; and even if we assigned somewhat higher probabilities, the result won’t change much. The conclusion to be drawn, I think, is that K, our background knowledge, historical and otherwise (excluding what we know by way of faith or revelation), isn’t anywhere nearly sufficient to support serious belief in G. If K were all we had to go on, the only sensible course would be agnosticism: “I don’t know whether G is true or not: all I can say for sure is that it is not terribly unlikely.” The main problem for such a historical case, as I see it, is what we can call the principle of dwindling probabilities: the fact that in giving such a historical argument, we can’t simply annex the intermediate propositions to K (as I’m afraid many who employ this sort of argument actually do) but must instead multiply the relevant probabilities.

It is for this reason that some such scheme as proposed in the testimonial model is necessary, if we human beings are to be able to know the great truths of the gospel.