The Probabilistic Argument from Gratuitous Evil and the Bracketing of Evidence

Elliott Crozat


December 10, 2019


In this article, I argue that when considering the existence of God, it is unreasonable to bracket background evidence in the manner described below. First, this move is methodologically insufficient. Second, it is conceptually questionable.

The Argument from Gratuitous Evil and the Bracketing Method

The argument from gratuitous evil is usually articulated as a version of the probabilistic argument from evil against the existence of God.[i] The argument can be articulated as follows:

  1. If there is a God, then there are no instances of gratuitous evil (i.e., God would not permit gratuitous evil).
  2. Probably, there is at least one instance of gratuitous evil.
  3. Therefore, probably, there is no God.

Some advocates of this and other versions of the probabilistic argument use a method of reasoning such that all background evidence for and against the existence of God is bracketed, except the data of evil. On the Bracketing Method, the likelihood of God’s existence is judged solely in relation to data regarding the occurrence of evil; such data, considered alone, decreases the probability of God. For example, William Rowe writes:

“The specific question assigned to us for discussion is this: Grounds for belief in God aside, do the evils in our world make atheistic belief more reasonable than theistic belief? The initial clause in this question is important… If we put aside grounds for belief in the existence of God, the likelihood that God exists cannot reasonably be assigned any probability beyond 0.5—where 1 represents God’s existence as certain, and 0 represents certainty that God does not exist. So, if we start from an initial point of God’s existence having a likelihood of 0.5 or less, and restrict ourselves to the evidence generated by the enormous amount of horrendous evil that occurs daily in our world, it should strike anyone that the likelihood of God’s existence can only go downward from 0.5… Apart from taking into account the positive reasons for thinking that God exists, do the evils that occur in our world make atheistic belief more reasonable than theistic belief? I shall argue that they do.”[ii]

Problems with the Bracketing Method

The Method is problematic for two reasons. First, it is too narrow. Plausibly, one should account for all available and relevant evidence before reaching a verdict on a particular case. For instance, in a court of law, the entire set of available evidence that is relevant, reliable, and otherwise admissible ought to be considered. In the sciences, investigators weigh all accessible, pertinent, and dependable evidence; if new information is collected, it can reinforce, weaken, or overturn a scientific theory. And when working a crime scene, detectives secure the scene and gather all germane evidence, not just some. As J. P. Moreland and William Lane Craig note, probabilities are relative to the background evidence under consideration. Concerning theism, they add: “Relative to the full scope of evidence, God’s existence is probable.”[iii] Michael Peterson expounds,

“Rowe’s artificial epistemic restraints set the stage for an invigorating academic exercise for both the theists and nontheists who accept them. However, the really important question is not whether the evidence of intense suffering counts against a limited form of theism considered against a highly controlled information set but how this evidence affects the evaluation of full-orbed versions of religion that involve theism…To make the complete position of Christian theism vulnerable to criticism of a subset of the position when that criticism is based on a piece of evidence, no matter how impressive, does not tell us nearly as much as critics claim.”[iv]

In short, using the Method is like asking a soccer team to ignore most of the playing field and use only a small section thereof. Although this might be an effective way to practice a subset of soccer skills, on game day the whole field is playable and the entire set of skills are needed.

Second, the Method is conceptually inadequate. Suppose we follow Rowe and try to bracket all evidence except the data of evil. It is doubtful that such a move is wholly achievable. For, as Rowe notes, we are addressing the following question: setting aside evidence for belief in God, do the evils that occur make atheism more probable than theism? (Call this question “Q.”)

Notice that Q refers to God. Immediately, a Socratically-inquisitive reader should ask: What is meant by “God?” Are we discussing the greatest conceivable being, the one who exists in all possible worlds if any? Are we asking if the data of evil lowers the probability that a being exists who is either necessary or impossible?

These questions raise a problem: given Q, we cannot bracket the concept of God and its entailments. If the proposition “God exists” is logically consistent and thus possible, then (assuming “God” refers to a necessary being), it follows via standard modal logic that God exists necessarily. But the existence of a necessary being is compatible with the existence of any other entity: merely possible, actual, necessary, or contingent. Furthermore, the probability that a necessary being exists is 1, regardless of whatever else exists, including evil in any possible distribution or degree.

Thus, the Method is conceptually questionable. It seems to take God as possible but improbable. Rowe himself seems to admit that God is logically possible.[v] Does the bracketeer believe that God is the sort of being that is necessary if possible? If the bracketeer admits that God is possible, shouldn’t he conclude that God is necessary regardless of evil?

The bracketeer can avoid this problem by claiming no position on whether or not God is possible. However, to avoid the charge of ad hoc maneuvering, he should explain this move. Given the work by philosophers on the coherence of theism, he needs to account for his non-commitment. Moreover, the bracketeer stills needs to address the problem of methodological narrowness described above.

Suggestions for Further Dialectic

Bracketeers (and perhaps all who support probabilistic versions of the argument from evil) ought to explain what they mean by “God.” They should clarify whether or not they hold that the concept of God is that of a necessary being and whether or not they find God possible in the first place. They should also explain what they mean by “probable,” “improbable,” and “possible.”

If bracketeers admit God as necessary if possible, and if they believe God is possible, then their use of the Method breaks down. It is not only artificial, as Peterson notes. It is inconsistent. Indeed, if God is possible, it seems the entire family of probabilistic arguments against God is in trouble[vi], since each of the family members attempts to conclude that the existence of a necessary-if-possible being is improbable to some degree less than .5 but greater than 0. (Regarding this attempt, we might note that each member follows the family warning: don’t ever take sides against the family.)

However, if the bracketeer takes God’s existence as impossible, he ought to say why, since God’s impossibility is not obvious. And if our bracketeer construes God as a contingent being, he should provide an explanation. For, as Brian Leftow has noted, most philosophers hold that if God exists, God exists necessarily.[vii]


In sum, there are at least two reasons to doubt the Bracketing Method: it is methodologically constrained and it is conceptually inconsistent. As such, bracketeers have some explaining to do.

Citations and Notes

*Photo by João Silas on Unsplash

[i] The term ‘gratuitous evil’ refers to an evil which is pointless in the sense that permitting its occurrence is not necessary to bring about a greater good which would justify its permission, nor is allowing the evil required to prevent the occurrence of an equally bad or worse evil. The probabilistic argument from evil is commonly articulated in something like the following way: although it is logically possible that God and evil coexist, the occurrence of evil renders the existence of God improbable to some degree less than .5 and greater than 0. See Note vi below for further commentary.

It would be profitable to read this article along with Tim Stratton’s Gratuitous Evil and Animal Suffering at

[ii] William Rowe, “Evil is Evidence against Theistic Belief,” in The Problem of Evil: Selected Readings, Second Edition, ed. Michael L. Peterson (Notre Dame: University of Notre Dame Press, 2017), 130-131.

[iii] J. P. Moreland and William Lane Craig, Philosophical Foundations for a Christian Worldview, First Edition, (Downer’s Grove: InterVarsity Press, 2003), 542.

[iv] Michael L. Peterson, “Christian Theism and the Evidential Argument from Evil,” in The Problem of Evil: Selected Readings, Second Edition, ed. Michael L. Peterson (Notre Dame: University of Notre Dame Press, 2017), 175-176.

[v] William Rowe, “Evil is Evidence against Theistic Belief,” in The Problem of Evil: Selected Readings, Second Edition, ed. Michael L. Peterson (Notre Dame: University of Notre Dame Press, 2017), 139.

[vi] At least, the family is in need of some good explaining. The two reasons to doubt the Bracketing Method are also reasons to doubt the whole enterprise of probabilistic arguments against God. Concerning the second reason, consider a common articulation of the probabilistic argument, provided by Justin P. McBrayer: “while it is logically possible that both God and evil coexist [sic], the latter is evidence against the former.” (See ) Notice that, in this formulation, the logical possibility of God is admitted, yet evil is said to make God’s existence improbable. But consider: if God is logically possible, then God is necessary and therefore not improbable.

[vii] Brian Leftow, “Swinburne on divine necessity,” in Religious Studies, Vol. 46, No. 2 (june 2010), 141, available at


About the Author

Elliott Crozat

Elliott R. Crozat (Ph.D., M.A.) is a full-time professor of philosophy and the humanities at Purdue University Global. His philosophical interests include metaphysics, epistemology, philosophy of religion, ethics, and the meaning of life. He lives in Sarasota, FL.

Learn More

More from this author