In Defense of an Evidential Argument from Evil: A Reply to William Lane Craig (2016)
On his Reasonable Faith website, William Lane Craig published a popular article about general arguments from evil against the existence of an all-powerful, all-knowing, and perfectly good God. Although the article was not specifically aimed at philosopher Paul Draper's famous evidential argument from evil based on facts about the biological role of pain and pleasure, it is consistent with the objections Craig raised against that argument in an oral debate with Draper on God's existence in 1998. In this article I review the points Craig raises in order to evaluate whether they have any force in rebutting Draper's writings on his pain and pleasure version of the evidential argument from evil. I conclude that Craig's points do not in fact have any force whatever against Draper's robust abductive argument from evil.
The first section provides several examples of what Draper means by "pain and pleasure." The second section reviews the argument's logical structure. Section three critically assesses Craig's response to the argument's premises. Finally, the fourth section critically assesses Craig's objection to the inference made in the argument's conclusion.
1. Why Pain and Pleasure Is a Problem for Theism
Because Craig's essay is written for a popular audience, it occasionally blurs distinctions that he might not have made if he were writing for a more academic audience. For example, he refers to 'the probabilistic problem of evil' (italics mine) even though Craig knows that there are many different kinds of probabilistic (or evidential) arguments from evil. Even if he successfully identifies a flaw in one version of the evidential argument from evil, it doesn't follow that said flaw will apply to all versions of the evidential argument from evil. Consider an analogy. There is no 'the' cosmological argument but instead a family of arguments known as "cosmological arguments" consisting of multiple, independent arguments, such as the kalam cosmological argument, the Leibnizian cosmological argument, and the Thomistic cosmological argument. Each of these arguments are logically independent in the sense that each argument succeeds or fails independently of the others, and the different arguments could be combined together to create a cumulative case for theism.
Similarly, there is no 'the' evidential argument from evil but instead a family of arguments known as "evidential arguments from evil," consisting of multiple, independent arguments, such as pain and pleasure, flourishing and languishing, virtue and vice, triumph and tragedy, autonomy and heteronomy, divine silence in the face of tragedies, social evil, and the failure of theodicies. Each of these arguments are logically independent in the sense that each argument succeeds or fails independently of the others, and the different arguments could be combined together to create a cumulative case against theism.
I will say no more about these other evidential arguments from evil and instead will devote the rest of this article to Draper's 1989 evidential argument from pain and pleasure. So let's begin by reviewing what Draper means by "pain and pleasure." As I read him, Draper means three facts which he calls O1, O2, and O3. These three facts are most easily explained through the use of examples.
1.1 Observation 1: Moral Agents Experiencing Biologically Useful Pain and Pleasure
Suppose you are a teenager sleeping in a hotel that has caught on fire. The hotel is old and doesn't have any smoke alarms. The fire gets closer and closer to you until you are actually in pain from the smoke and the intense heat. Your pain wakes you up in time to escape, so you go on to survive and start a family in your twenties. In this case your pain was biologically useful because it contributed to the biological "goal" of survival. The naturalistic explanation for the unfolding of this scenario is obvious. If human beings are the products of evolution by natural selection, we would expect physical pain and pleasure to motivate human behavior in ways that aid survival and reproduction.
1.2 Observation 2: Sentient Beings Experiencing Biologically Useful Pain and Pleasure
Most human beings are what philosophers call moral agents, people who can be held responsible for their actions and the consequences of their actions. But some human beings (such as young children and humans with certain mental disabilities), as well as nonhuman sentient animals (such as primates and dolphins), are moral patients—sentient beings who can be harmed from their own point of view, but are not responsible for their actions.
On naturalism, among biological sentient beings we would expect both moral patients and moral agents to experience biologically useful pain and pleasure. For biological moral patients (such as nonhuman primates and dolphins) are biologically similar to biological moral agents (such as human beings). On theism, however, we would predict that biological moral patients do not suffer the same kind of pain as biological moral agents. Such pain plays no known moral role in the lives of the biological moral patients who experience it. For example, such pain isn't necessary for free will, doesn't seem to influence moral patients to freely choose right actions over wrong ones, doesn't enable moral patients to acquire moral virtue, and doesn't usually increase their knowledge of God. Indeed, as if to underscore the point, theists typically emphasize that concepts like moral freedom, moral obligation, moral virtue, and salvation do not even apply to nonhuman animals, and thus do not apply to the majority of moral patients.
One objection to this point goes as follows. If naturalism is true, the existence of biological sentient beings is surprising, and so it's false to say that naturalism 'predicts' that biological sentient beings would experience biologically useful pain and pleasure. I agree that naturalism, by itself, does not predict the existence of biological sentient beings and so, for that reason, doesn't predict biological sentient beings who would experience biologically useful pain and pleasure. That is not of obvious relevance to Draper's argument from pain and pleasure, however, since that argument includes the proposition "Biological sentient beings exist" in the relevant background information. If there is a successful evidential argument against naturalism based upon that proposition, then so be it. But, given the truth of that and other background propositions, naturalism does make it probable that such beings would experience biologically useful pain and pleasure. Take B1 to be a proposition in our background knowledge, E1 to be an item of evidence to be explained, and H1 and H2 to be rival explanatory hypotheses. The fact (if it is a fact) that
B1 is more probable on the assumption that H1 is true than on the assumption that H2 is true
does not in any way undercut
E1 is more probable on the assumption that H1 and B1 are true than on the assumption H2 and B1 are true.
1.3 Observation 3: Sentient Beings Experiencing Pain and Pleasure Not Known to be Useful
But not all physical pain and pleasure is biologically useful. For example, consider an animal trapped in a forest fire that suffers horrific pain as it slowly burns to death. On the one hand, this kind of pain is biologically appropriate: it is biologically useful in general that animals feel pain when they come in contact with fire. But, on the other hand, this specific instance of pain is not biologically useful because it does not contribute to the biological goals of survival or reproduction.
On naturalism, this is just what we would expect. If naturalism is true, all animals are the byproducts of unguided evolution by natural selection, which is both indifferent to suffering and incapable of fine-tuning animals to prevent such pain. Thus, the kind of pain and pleasure that we actually find is what we would expect if naturalism were true.
But if theism were true, God could "fine tune" animals so that they only experience physical pain and pleasure when it is morally necessary. So theism leads us to expect that pain and pleasure are fundamentally moral phenomena that just happen to be connected to the biological goals of survival and reproduction. That's a huge coincidence that naturalism doesn't need.
2. Draper's Evidential Argument from Pain and Pleasure Summarized
While my examples above refer to naturalism, in his 1989 article Draper instead referred to "the hypothesis of indifference" (HI), which he defines as the proposition that "neither the nature nor the condition of sentient beings on earth results from benevolent or malevolent actions performed by supernatural persons." I will say more about that in a moment.
2.1 Formalizing the Argument
For the sake of brevity, I will adopt the following notation throughout the remainder of this essay:
|Pr(x)||the epistemic probability of x|
|Pr(x | y)|| the epistemic probability of x conditional upon y
(i.e., the epistemic probability that proposition X is true,
on the assumption that proposition y is true)
|Pr(|H|)||the intrinsic probability of H|
|>!||much greater than|
Let O (for "observations") represent the evidence to be explained, in this case observations about pain and pleasure. More formally, Draper defines O as a statement about "the kinds, amounts, and distribution of pain and pleasure in the world." In Draper's formulation, O is the conjunction (or combination) of the following:
O1 = a statement about facts concerning "moral agents experiencing pain or pleasure that we know to be biologically useful";
O2 = a statement about facts concerning "sentient beings that are not moral agents experiencing pain or pleasure that we know to be biologically useful"; and
O3 = a statement about facts concerning "sentient beings experiencing pain or pleasure that we do not know to be biologically useful."
So defined, Draper's argument from the biological role of pain and pleasure runs as follows:
- O is known to be true.
- Theism (T) is not much more probable intrinsically than the hypothesis of indifference (HI) [i.e., Pr(|T|) is not much greater than Pr(|HI|)].
- O is much more likely on the assumption that the hypothesis of indifference is true than it is on the assumption that theism is true [i.e., Pr(O | HI) >! Pr(O | T)].
- So, other evidence held equal, theism is probably false.
Let us now return to the distinction between naturalism and HI. While HI is, like naturalism, logically incompatible with theism, HI is logically consistent with both supernaturalism and naturalism. Indeed, HI has no positive ontological commitments. This is important because it strengthens the argument by strengthening its second premise. HI is a much more modest hypothesis than naturalism. Accordingly, HI is much more intrinsically probable than naturalism.
In his popular essay, Craig writes nothing in response to premises (1) or (2). Instead, all of his objections respond to either premises (3) or (4). Let's turn to those objections now.
3. Premise (3)
3.3.1 The Skeptical Theism Defense
Craig's first objection to probabilistic arguments from evil is that the probability of known facts about evil on theism is inscrutable (unknowable):
1. We are not in a good position to assess the probability of whether God has morally sufficient reasons for the evils that occur. As finite persons, we are limited in time, space, intelligence, and insight. But the transcendent and sovereign God sees the end from the beginning and providentially orders history so that His purposes are ultimately achieved through human free decisions. In order to achieve His ends, God may have to put up with certain evils along the way. Evils which appear pointless to us within our limited framework may be seen to have been justly permitted within God's wider framework. To borrow an illustration from a developing field of science, Chaos Theory, scientists have discovered that certain macroscopic systems, for example, weather systems or insect populations, are extraordinarily sensitive to the tiniest perturbations. A butterfly fluttering on a branch in West Africa may set in motion forces which would eventually issue in a hurricane over the Atlantic Ocean. Yet it is impossible in principle for anyone observing that butterfly palpitating on a branch to predict such an outcome. The brutal murder of an innocent man or a child's dying of leukemia could produce a sort of ripple effect through history such that God's morally sufficient reason for permitting it might not emerge until centuries later and perhaps in another land. When you think of God's providence over the whole of history, I think you can see how hopeless it is for limited observers to speculate on the probability that God could have a morally sufficient reason for permitting a certain evil. We're just not in a good position to assess such probabilities.
But Draper refuted this objection in his 1989 essay. To sum up: it's possible that God has unknown reasons for allowing evil. But it's also possible—and antecedently just as likely—that God has unknown reasons for preventing evil. So God's preventing the "brutal murder of an innocent man or a child's dying of leukemia could produce a sort of ripple effect through history such that God's morally sufficient reason for" preventing such evils "might not emerge until centuries later and perhaps in another land." So the possibilities of unknown reasons for allowing evil and unknown reasons for preventing evil "cancel out." We're right back where we started, namely, working with what we do know: O. In fact, this is pretty much the point of using epistemic probabilities. If we had perfect, complete information, then we wouldn't need to use probabilities at all. So human ignorance is not a good objection to comparing Pr (O | HI) to Pr(O | T). As far as I know, Craig has never addressed this portion of Draper's original 1989 essay in any of his writings.
Furthermore, as numerous philosophers (nontheists and theists alike) have pointed out, logically consistent natural theologians cannot appeal to the limitations of human cognitive abilities to defeat evidential arguments from evil. Allow me to explain. If human cognitive limitations really did prevent us from assessing whether God has morally sufficient reasons for allowing evil, including facts about pain and pleasure, then Craig can kiss goodbye all of his arguments from natural theology for God's existence. Consistent skeptical theists should also insist that human cognitive limitations prevent us from assessing:
- the antecedent probability of our universe beginning to exist on theism, i.e., Pr(beginning | theism);
- the antecedent probability of so-called cosmological 'fine-tuning' on theism, i.e., Pr('tuning' | theism); and
- the antecedent probability of the Resurrection on theism, i.e., Pr(Resurrection | theism).
This is why logically consistent natural theologians, like Oxford University philosopher Richard Swinburne, don't rely upon skeptical theism. Instead, they attempt to provide theodicies—explanations for why God, if He exists, would allow facts about the kinds, amounts, and distribution of evil in the world to obtain.
3.3.2 Craig's Theodicies
Craig's second objection is that Christian doctrines increase the antecedent probability of evil on theism; that is, they increase the value of Pr(O | T). In his words:
2. The Christian faith entails doctrines that increase the probability of the co-existence of God and evil. In so doing, these doctrines decrease any improbability of God's existence thought to issue from the existence of evil.
But Craig's second objection contradicts his first one. If we can determine that the Christian faith entails doctrines that increase the probability of the coexistence of God and evil, then it follows that we are in a good position to assess whether God has morally sufficient reasons for the evils that occur. But let's put that to the side. Has Craig shown that Christian doctrines increase the antecedent probability of evil on theism?
We can formalize his second objection as follows.
(5) Evil is antecedently more probable given Christian theism (CT) than it is given HI, i.e., Pr(evil | CT) > Pr(evil | HI).
But Draper's argument is not about all types of evil; rather, it concerns known facts about a specific subset of evils (and goods): pain and pleasure. So let's revise (5) to become:
(5') Facts about pain and pleasure are antecedently more probable given Christian theism (CT) than they are given HI, i.e., Pr(O | CT) > Pr(O | HI).
Even if (5') were true, however, it would be irrelevant. Why? Because it wouldn't follow that:
(6) O is antecedently more probable given T than it is given HI, i.e., Pr(O | T) > Pr(O | HI).
As Draper insightfully notes in an unrelated context, his evidential argument from evil compares HI to theism simpliciter, not to theism conjoined with Christian doctrines—and the relevant issue is whether O is just as probable given theism as it is given HI, not whether theism, with the assistance of several auxiliary hypotheses, can explain O as well as HI. Notice also that Christian theism is not known to be true. So again, one must wonder: what relevance could Christian doctrines possibly have to Draper's evidential argument from facts about pain and pleasure?
Again, paraphrasing Draper, the only conceivable way that Christian theism could be relevant would be if it showed that the third premise of the argument is false—that O is not antecedently much more probable on HI than on T. But how could it do this? Since we don't know that Christianity is true, we cannot simply equate Pr(O | T) with Pr(O | T & CT). Rather, in order to assess CT's effect on Pr(O | T), we must use what Draper calls the weighted average principle (WAP), which gives the following equation:
Pr(O | T) = Pr(CT | T) x Pr(O | T & CT) + Pr(~CT | T) x Pr(O | T & ~CT)
This formula is an average because Pr(CT | T) + Pr(~CT | T) = 1. It is not a simple straight average, however, since those two values may not equal 1/2. The second half of the right-hand side of that equation, Pr(~CT | T) x Pr(O | T & ~CT), is not going to be useful for deriving a high value for Pr(O | T). Otherwise, there would be no need to introduce CT in the first place! In order for CT to be a good defeater of premise 3 of Draper's argument from the biological role of pain and pleasure, then, Pr(CT | T) x Pr(O | T & CT) needs to be high—the higher the better.
Paraphrasing Draper, this equation clearly exposes the weakness of Christian doctrines as an objection to Draper's evidential argument from evil. For Christianity in particular is antecedently very unlikely on theism in general for three reasons. First, CT is a very specific hypothesis, and so will be very improbable in the absence of evidence for it regardless of how little evidence there is against it. The specificity of CT means that it makes many more assumptions than T alone, so there are many more places where CT can get the facts wrong. The more assumptions that a hypothesis makes, the greater the risk of error.
Second, we have very little "independent" evidence for the truth of CT. At best, it is weakly supported by the historicity of Jesus and by several events relating to his life, death, and alleged resurrection, such as his crucifixion, burial, and empty tomb, as well as postmortem appearances of someone who appeared to be Jesus. Granted, CT entails that "God raised Jesus from the dead" (R) and Christian scholars, including Craig, argue that R is the best explanation for those events. But R remains highly speculative. Contrary to what Craig and other defenders of R assert, R does not imply the events of Jesus' crucifixion, burial, empty tomb, and postmortem appearances.
Third, the assumption that theism is true gives us very little reason to expect that R is true. Thus, apart from independent evidence for CT, a very specific hypothesis like CT is very unlikely on theism, and hence Pr(O | T) is very close to Pr(O | T & ~CT). In short, CT gives us no reason at all to think that the third premise of Draper's evidential argument from evil (premise 3) is false.
Following Draper's notation, we will let T1, ..., Tn represent "expansions" of theism, i.e., auxiliary assumptions which entail theism. This notation enables us to put the WAP in a generic form:
Pr(O | T) = Pr(Tn | T) x Pr(O | Tn) + Pr(~Tn | T) x Pr(O | T & ~Tn)
Let us now turn to Craig's specific points about Christian doctrines, which function as theodicies. Since Draper used T1, T2, and T3 in his 1989 article to refer to specific theodicies, in what follows I will pick up where Draper left off and number Craig's theodicies starting at T4.
220.127.116.11 T4: The Knowledge of God Theodicy
First, Craig offers the following explanation for human suffering:
a. The chief purpose of life is not happiness, but the knowledge of God. One reason that the problem of evil seems so puzzling is that we tend to think that if God exists, then His goal for human life is happiness in this world. God's role is to provide comfortable environment for His human pets. But on the Christian view this is false. We are not God's pets, and man's end is not happiness in this world, but the knowledge of God, which will ultimately bring true and everlasting human fulfillment. Many evils occur in life which maybe utterly pointless with respect to the goal of producing human happiness in this world, but they may not be unjustified with respect to producing the knowledge of God. Innocent human suffering provides an occasion for deeper dependency and trust in God, either on the part of the sufferer or those around him. Of course, whether God's purpose is achieved through our suffering will depend on our response. Do we respond with anger and bitterness toward God, or do we turn to Him in faith for strength to endure?
This paragraph suggests the following theodicy:
T4: God exists, and one of His final ends is human knowledge of God.
It seems to me that this theodicy suffers from essentially the same problems as theodicy T1 in Draper's original 1989 article, however. To see why, I will charitably assume for the sake of argument that Pr(T4 | T) is high. Nevertheless, Pr(O | T4) is still not significantly greater than Pr(O | T & ~T4). Craig is surely correct that when humans experience undeserved biological pain, it "provides an occasion for deeper dependency and trust in God." So T4 does provide some reason to expect O (and, specifically, O1). But O also reports biological pain experienced by sentient beings that are not moral agents—such as dolphins, elephants, and nonhuman primates—and so are incapable of deeper dependency and trust in God. This is much more surprising on T4 than on theism & ~T4.
Furthermore, in addition to the facts reported in O, we also know that the problem of evil by itself, as well as divine silence during suffering, causes many people to doubt or deny God's existence. But the belief that God exists is a prerequisite for trust in God. So, channeling Draper, we may conclude that T4's 'predictive' advantage is greatly outweighed by T4's 'predictive' disadvantages, and for this reason Pr(O | T4) is not significantly greater than Pr(O | T & ~T4).
18.104.22.168 T5: The Rebellion Theodicy
Craig next explains human suffering along these lines:
b. Mankind is in a state of rebellion against God and His purpose. Rather than submit to and worship God, people rebel against God and go their own way and so find themselves alienated from God, morally guilty before Him, and groping in spiritual darkness, pursuing false gods of their own making. The terrible human evils in the world are testimony to man's depravity in this state of spiritual alienation from God. The Christian is not surprised at the human evil in the world; on the contrary, he expects it. The Bible says that God has given mankind over to the sin it has chosen; He does not interfere to stop it, but lets human depravity run its course. This only serves to heighten mankind's moral responsibility before God, as well as our wickedness and our need of forgiveness and moral cleansing.
This paragraph suggests this additional theodicy:
T5: God exists, and mankind is in a state of rebellion against God and His purpose.
This is not a successful theodicy, however. Two responses immediately suggest themselves. First, Pr(T5 | T) is not high. While the content of T (arguably) provides some reason to expect the existence of nondivine persons, and so to that extent does provide some reason to expect the existence of mankind, it provides no reason at all to expect mankind in a state of rebellion against God and His purpose. In fact, the situation is even worse than that. Based solely on the content of T, the existence of nondivine persons in rebellion against God is very surprising. T entails that if nondivine persons exist, then they were created by a morally perfect being. Everything else held equal, the proposition that "Nondivine persons were created by a morally perfect being" provides strong reason to expect that nondivine persons would not be in rebellion against God. So Pr(T5 | T) is low, i.e., Pr(T5 | T) << 1/2. In accordance with the WAP, it follows that Pr(O | T) will be much closer to Pr(O | T & ~T5). (Remember, at this point we are assessing Pr(T5 | T), the antecedent probability of T5 on T, not Pr(T5 | CT), the antecedent probability of T5 on CT. So Christian doctrines about the rebellion of mankind are not relevant here.)
Second, Pr(O | T5) is not significantly greater than Pr(O | T & ~T5). This can be seen by considering O1, O2, and O3 individually. Regarding O1, T5 is irrelevant to the vast majority of human pain and pleasure reported in O1. Turning to O2, T5 is by definition irrelevant to the pain and pleasure experienced by moral patients reported in O2. This leaves O3. O3 includes biologically gratuitous human pain and pleasure. T5 is arguably relevant to that portion of O3. But while it is relevant, T5 does not provide a strong reason to expect biologically gratuitous human pain and pleasure, i.e., Pr("biologically gratuitous human pain and pleasure" | T5) is not > 1/2. Furthermore, T5 also seems irrelevant to biologically gratuitous pain and pleasure experienced by moral patients who are not also moral agents; at the very least, T5 provides no reason to think such pain and pleasure is more probable than not. For all of these reasons, then, T5 fails to significantly "boost" the probability of O on the assumption that T is true.
22.214.171.124 T6: The Afterlife Compensation Theodicy
Third, Craig offers yet another explanation for human suffering, to wit:
c. The knowledge of God spills over into eternal life. In the Christian view, this life is not all there is. Jesus promised eternal life to all who place their trust in him as their Savior and Lord. In the afterlife God will reward those who have borne their suffering in courage and trust with an eternal life of unspeakable joy. The apostle Paul, who wrote much of the New Testament, lived a life of incredible suffering. Yet he wrote, "We do not lose heart. For this slight, momentary affliction is preparing us for an eternal weight of glory beyond all comparison, because we look not to the things that are seen, but to the things that are unseen, for the things that are seen are transient, but the things that are unseen are eternal" (II Cor. 4:16-18). Paul imagines a scale, as it were, in which all the sufferings of this life are placed on one side, while on the other side is placed the glory that God will bestow on his children in heaven. The weight of glory is so great that it is literally beyond comparison with the suffering. Moreover, the longer we spend in eternity the more the sufferings of this life shrink toward an infinitesimal moment. That's why Paul could call them "a slight and momentary affliction"—they were simply overwhelmed by the ocean of divine eternity and joy which God lavishes on those who trust Him.
This paragraph suggests the following theodicy:
T6: God exists, and will reward believers in the afterlife with glorious pleasure that will outweigh any horrific suffering they experienced in this life.
T6 fails for two reasons. First, it's far from obvious that Pr(T6 | T) is high. In addition to the dubious assumption that theism 'predicts' a physical universe designed for the evolution of embodied moral agents in the form of human beings, T6 also makes a variety of theological assumptions, none of which are antecedently more probable than not on mere theism. For example, T6 assumes:
(a) that God, having created embodied moral agents, intends for them to have an eternal afterlife (rather than an eternal biologically embodied existence or no eternal existence whatsoever);
(b) that the only or primary factor God uses to decide one's fate in the afterlife is one's religious beliefs during this life (rather than using one's behavior as the only or primary factor);
(c) that compensating believers in the afterlife somehow morally justifies God in allowing the types, quantity, and distribution of evil we find in the world; and
(d) that God is somehow morally justified in not compensating nonbelievers who have suffered in this life.
For all of these reasons, it's far from obvious that Pr(T6 | T) is high.
Second, Pr(O | T6) is not significantly greater than Pr(O | T & ~T6). In fact, it appears that these two values are either equal or very close to equal, which again implies that T6 is irrelevant. As was the case with T5, T6 is irrelevant to the vast majority of human pain and pleasure reported in O1 (since it was or is experienced by nonbelievers who, according to CT, will not be compensated in the afterlife). Furthermore, T6 is, by definition, irrelevant to the vast majority of pain and pleasure experienced by sentient beings (as reported in O2 and O3).
126.96.36.199 T7: The Infinite Goodness Theodicy
Finally, Craig's last explanation for human suffering runs as follows:
d. The knowledge of God is an incommensurable good. To know God, the source of infinite goodness and love, is an incomparable good, the fulfillment of human existence. The sufferings of this life cannot even be compared to it. Thus, the person who knows God, no matter what he suffers, no matter how awful his pain, can still say, "God is good to me," simply by virtue of the fact that he knows God, an incomparable good.
This paragraph suggests the following theodicy:
T7: God exists, and the knowledge of God is an incommensurable good.
There is no doubt that, if God exists, the knowledge of God is an enormous good. But to go from there to the claim that "The sufferings of this life cannot even be compared to it" is a surprisingly tone deaf comment to make for someone like Craig. Indeed, one is tempted to retort, "Try telling that to the Holocaust survivors who lost their faith in God as a result of their experience," and to dismiss this entire point with the contemptuous sneer it deserves. But to do so would be a mistake, for that would fail to point out that there is no good reason to think that knowledge of God requires anyone to suffer pain, and therefore the knowledge of God does not in any way increase Pr(O | T) because Pr(O | T7) is not significantly greater than Pr(O | T & ~T7). So this doctrine is irrelevant.
In summary, then, none of Craig's four Christian doctrines increase Pr(O | T). Accordingly, the central claim of Draper's evidential argument from evil—that facts about the biological role (and apparent moral randomness) of pain and pleasure are much more probable on HI than on T—emerges unscathed.
4. Conclusion (4)
Craig's final strategy is to argue that the total evidence makes theism more probable than not:
3. Relative to the full scope of the evidence, God's existence is probable. Probabilities are relative to what background information you consider. For example, suppose Joe is a student at the University of Colorado. Now suppose that we are informed that 95% of University of Colorado students ski. Relative to this information it is highly probable that Joe skis. But then suppose we also learn that Joe is an amputee and that 95% of amputees at the University of Colorado do not ski. Suddenly the probability of Joe's being a skier has diminished drastically!
Similarly, if all you consider for background information is the evil in the world, then it's hardly surprising that God's existence appears improbable relative to that. But that's not the real question. The real question is whether God's existence is improbable relative to the total evidence available. I'm persuaded that when you consider the total evidence, then God's existence is quite probable.
Craig then proceeds to summarize his purported evidence for God's existence.
Craig's Joe example is a perfect example for illustrating why Draper's argument succeeds, whereas Craig's theistic arguments do not. In the Joe example we are asked to consider the following statistical syllogism:
- 95% of University of Colorado students ski.
- Joe is a University of Colorado student.
- [95% probable] Joe skis.
As Craig correctly points out, that statistical syllogism is logically incorrect because it violates the rule of total evidence. We know more about Joe than the fact that he is a University of Colorado student. We also know that he is an amputee. So once the evidence about Joe is fully stated, it is improbable that Joe skis.
Although Draper's argument is inductive, it has a different logical form than a statistical syllogism; it is what I have elsewhere categorized as either an "explanatory argument" or, in the spirit of Richard Swinburne's work, an "F-inductive argument." Nevertheless, the rule of total evidence is satisfied because the conclusion of Draper's argument is that "All other evidence held equal, theism is probably false"—not simply "Theism is probably false."
By contrast, Craig's arguments from natural theology contain no such ceteris paribus clause, and so are vulnerable to a parallel objection. Putting aside the fact Craig appears to have confused "background information" with "evidence to be explained", imagine if we were to paraphrase Craig's own words against him as part of a reply to his kalam cosmological argument in the following manner:
Similarly, if all you consider for background information is the beginning of the universe, then it's hardly surprising that God's existence appears probable relative to that. But that's not the real question. The real question is whether God's existence is probable relative to the total evidence available. I'm persuaded that when you consider the total evidence, then God's existence is quite improbable.
In fact, Craig does neglect to consider additional evidence which favors naturalism over theism, including but not limited to the other known facts about evil mentioned in section 1. So, contrary to what he claims, Craig has not even attempted to assess the total evidence, much less justified the conclusion that it favors theism over naturalism.
 To be precise, I am referring to Draper's most famous evidential argument from evil, his evidential argument from pain and pleasure. See Paul Draper, "Pain and Pleasure: An Evidential Problem for Theists." Nous Vol. 23, No. 3 (June 1989): 331-350. Reprinted in The Evidential Argument from Evil ed. Daniel Howard-Snyder (Indianapolis, IA: Indiana University Press, 1996): 12-29. This paper relies on the 1996 reprint.
 Jeffery Jay Lowder, "Summary and Assessment of the Craig-Draper Debate on the Existence of God (1998)" (October 15, 2011). The Secular Outpost blog. <http://www.patheos.com/blogs/secularoutpost/2016/02/20/summary-and-assessment-of-the-craig-draper-debate-on-the-existence-of-god-1998/>.
 Paul Draper, "Christian Theism and Life on Earth" in The Blackwell Companion to Science and Christianity ed. J. B. Stump and Alan G. Padgett (Malden, MA: Blackwell, 2012): 306-315; Paul Draper, "Humean Arguments from Evil" in The Blackwell Companion to the Problem of Evil ed. Justin P. McBrayer and Daniel Howard-Snyder (Malden, MA: Blackwell, 2013): 67-74, p. 72.
 Paul Draper, "Humean Arguments from Evil," p. 73; Paul Draper, "God and the Mystery of Evil," paper delivered at the Society of Christian Philosophers conference, September 26, 2015, pp. 29-34.
 Paul Draper, "Humean Arguments from Evil," p. 73.
 Paul Draper, "God and the Mystery of Evil," pp. 18-19, 26-29.
 William Rowe, "The Evidential Argument from Evil: A Second Look" in The Evidential Argument from Evil ed. Daniel Howard-Snyder (Indianapolis, IA: Indiana University Press, 1996): 262-285, p. 276; Theodore M. Drange, Nonbelief and Evil (Amherst, NY: Prometheus Books, 1998), pp. 208-211, 223-224.
 Cf. William Rowe, "The Problem of Evil and Some Varieties of Atheism." American Philosophical Quarterly Vol. 16, No. 4 (October 1979): 335-341. Reprinted in The Evidential Argument from Evil ed. Daniel Howard-Snyder (Indianapolis, IA: Indiana University Press, 1996): 1-11.
 Paul Draper, "More Pain and Pleasure: A Reply to Otte" in Christian Faith and the Problem of Evil ed. Peter van Inwagen (Grand Rapids, MI: Eerdmans, 2004): 41-54, p. 47.
 Paul Draper, "Pain and Pleasure: An Evidential Problem for Theists," pp. 24-25. Cf. Jeffery Jay Lowder, "Draper on Pain and Pleasure: Part 3" (December 7, 2014). The Secular Outpost blog. <http://www.patheos.com/blogs/secularoutpost/2014/12/07/app3/>.
 See, for example, Paul Draper, "The Skeptical Theist" in The Evidential Argument from Evil ed. Daniel Howard-Snyder (Indianapolis, IA: Indiana University Press, 1996): 175-192, p. 188.
 Paul Draper, "God and Evil: A Philosophical Inquiry" (October 1, 2010), p. 18. Talk presented at the University of Notre Dame Ninth Annual Plantinga Lecture, Notre Dame, Indiana. <https://philreligion.nd.edu/assets/44795/1011lecture.pdf>
 Cf. Draper's parallel response to the multiverse objection to an evidential argument from moral agency in Paul Draper, "Cosmic Fine Tuning and Terrestrial Suffering: Parallel Problems for Naturalism and Theism." American Philosophical Quarterly Vol. 41, No. 4 (October 2004): 311-321.
 Paul Draper, "More Pain and Pleasure: A Reply to Otte."
 Paul Draper, "Pain and Pleasure: An Evidential Problem for Theists," p. 20.
 William Lane Craig, Assessing the New Testament Evidence for the Historicity of Jesus (Lewiston, NY: Edwin Mellen Press, 1989).
 By itself, R tells us nothing about whether there was an empty tomb, for R is compatible with a wide variety of auxiliary hypotheses concerning the status of Jesus' corpse between the time of his death and the time of his alleged resurrection. For all we know antecedently—that is, prior to considering the specific evidence—Jesus could have been denied burial and could have risen from the cross, not the grave. Or perhaps individuals who were resurrected in a tomb might decide to stay in the tomb, eternally marveling at their own resurrection body. In order to guarantee an empty tomb, the resurrection hypothesis must be combined with at least one other hypothesis, such as the honorable burial hypothesis. In other words, given only the truth of the resurrection hypothesis, the probability of an empty tomb is less than 100%. See Robert Greg Cavin, "Is There Sufficient Historical Evidence to Establish the Resurrection of Jesus?" Faith and Philosophy Vol. 12, No. 3 (July 1995): 361-379. Reprinted in The Empty Tomb: Jesus Beyond the Grave ed. Robert M. Price and Jeffery Jay Lowder (Amherst, NY: Prometheus Books, 2005): 19-41.
 In fact, it seems to me that Pr(O | T4) is actually less than Pr(O | T & ~T4), but the success of Draper's argument doesn't depend upon that claim.
 See, for example, William L. Rowe, "The Problem of Evil: A Second Look" in The Evidential Argument from Evil ed. Daniel Howard-Snyder (Indianapolis, IA: Indiana University Press, 1996): 262-310, p. 276.
 Paul Draper, "Pain and Pleasure: An Evidential Problem for Theists," p. 22.
 I owe this point to Paul Draper.
 I owe this point to Paul Draper.
 Paul Draper, "Pain and Pleasure: An Evidential Problem for Theists," p. 20.
 As Paul Draper has pointed out in private correspondence, T5 is not completely irrelevant to O because O3 includes biologically gratuitous human pain.
 To avoid any misunderstandings, I again want to clarify the point made above. The content of theism (arguably) does provide some reason for expecting the existence of nondivine persons, and so to that extent provides some reason for expecting the existence of mankind, i.e., Pr(mankind | theism) > 0. But the content of theism does not provide reason to expect that the existence of mankind is more probable than not; in other words, there is no good reason to think Pr(mankind | theism) > 1/2.
 I owe this point to Paul Draper.
 Jeffery Jay Lowder, "F-Inductive Arguments: A New Type of Inductive Argument (March 21, 2014). The Secular Outpost blog. <http://www.patheos.com/blogs/secularoutpost/2014/03/21/f-inductive-arguments-a-new-type-of-inductive-argument/>.
 Craig's words suggest that evidential arguments from evil treat some fact about evil as "background information" when in fact they do the opposite—they treat some fact about evil as evidence (odd or puzzling facts to be explained). What actually constitutes the relevant background information for Draper's arguments is summarized in the following five propositions:
B1: Pain and pleasure, if they exist, have intrinsic moral value.
B2: A physical universe—which operates according to natural laws, is intelligible, and which supports the possibility of intelligent life—exists.
B3: Living things, including sentient beings, exist on Earth. These sentient beings include, but are not limited to, human beings.
B4: Some (Earthly) sentient beings are not moral agents, but are biologically very similar to embodied moral agents.
B5: Humans are goal-directed organic systems, composed of parts that systematically contribute to the biological goals of these systems.
 At this point, Craig might object that the kalam cosmological argument is deductive, not inductive. But, at the end of the day, this turns out to be irrelevant since at least one of the premises is not known with certainty to be true. In fact, I think neither premise is known with certainty to be true, i.e., Pr("anything which begins to exist has a cause") < 1 and Pr("the universe began to exist") < 1. Therefore, the kalam's conclusion, "Therefore, the universe has a cause," is also uncertain, i.e., Pr("the universe has a cause") < 1. So, regardless of whether we categorize the kalam cosmological argument as deductive or inductive, we're still dealing with probabilities.
Furthermore, the probability of the kalam cosmological argument's conclusion is decreased by additional, relevant evidence. Even if we knew with certainty that the universe began to exist, that evidence would be offset by the evidence that the universe began to exist with time, not in time. If we include in our background information the fact that the universe began to exist, the fact that it begin with time is more probable on naturalism than on theism. But that entails that the kalam cosmological argument commits the fallacy of understated evidence. It understates the relevant cosmological evidence, analogous to how Craig's Joe example understates the relevant physiological evidence about Joe.
 I am grateful to Paul Draper for helpful comments on a previous draft of this essay.
Copyright ©2016 Jeffery Jay Lowder. The electronic version is copyright ©2016 by Internet Infidels, Inc. with the written permission of Jeffery Jay Lowder. All rights reserved.