Does Collins's response to the low probability objection in Hawthorne and Manson's formulation succeed, or does the statistical problem remain?

The fine-tuning argument for physical constants is one of the strongest contemporary cosmological arguments for theism. Robin Collins developed its most influential Bayesian formulation in analytic philosophy. However, John Hawthorne and Neil Manson in their famous paper (2009) raised the "low probability objection" which targets the probabilistic structure of the argument. The debate between Collins and his critics reveals the depth of statistical challenges in applying probability theory to cosmology.

Inadequate responses to avoid

From some defenders of the fine-tuning argument:

"Fine-tuning is obvious to observation; we don't need probabilistic calculations." This ignores the technical nature of the argument. Collins himself emphasizes that the argument's strength lies in its precise Bayesian formulation, not in general intuition. Refusing to engage with probabilistic problems weakens the argument rather than strengthening it.

"Hawthorne and Manson are just atheists looking for loopholes." Non-academic ad hominem. Hawthorne is a respected philosopher in epistemology and metaphysics, and Manson is a specialist in philosophy of science. Their criticism is published in peer-reviewed journals and taught in major universities. A serious response requires engaging with their technical arguments.

"Collins responded and the matter is settled." Oversimplification. Collins's response (2009, 2012) is strong but did not end the debate. Hawthorne and Manson developed counter-responses, and the discussion remains active in conferences and specialized journals.

From some critics:

"Probability is meaningless in cosmology." Too strong a claim. Even Hawthorne and Manson don't go this far, but rather raise specific problems in how to calculate probabilities. Rejecting all probabilistic thinking in cosmology goes beyond what technical criticism warrants.

"The problem refutes all fine-tuning arguments." Imprecise. The Hawthorne-Manson problem targets specific Bayesian formulations, especially those relying on a "natural measure" for the parameter space. Other formulations may avoid the problem.

Why these responses are inadequate

They fail to grasp the technical nature of the debate. This is not a general philosophical discussion, but a debate in the foundations of probability theory and its application to infinite spaces. Serious evaluation requires precise understanding of the mathematics involved.

Structure of the Hawthorne-Manson objection

The basic problem: Collins calculates the probability of fine-tuning based on "naturalism" versus "design." But calculating P(fine-tuning | naturalism) requires specifying a probability distribution over the space of possible constants. The problem: this space is infinite, and there is no unique "natural measure" for distributing probabilities over it.

Technical detail: Take the gravitational constant G. The theoretical range of possible values for G is (0, ∞). To calculate the probability that G falls within the narrow range permitting life, we need a probability distribution over (0, ∞). But:

- A uniform distribution over (0, ∞) is mathematically undefined.
- Any specific distribution (like log-uniform) appears arbitrary.
- The choice of distribution determines the result: different distributions give radically different probabilities.

Hawthorne and Manson: "Without objective justification for choosing a particular distribution, probability calculation becomes arbitrary. The argument therefore lacks solid mathematical foundation."

The second problem: the normalization problem. Even if we choose a distribution, how do we normalize it? Normalization requires that the sum/integral of probabilities = 1. But over an infinite space, this requires arbitrary decisions about "boundaries" and "relative weights."

Collins's detailed response

Collins in "The Teleological Argument" (2009) and "The Fine-Tuning Evidence is Convincing" (2012) provides a multi-level response:

First: the Epistemic Constraint Principle. We don't need an "absolutely objective" probability distribution. An "epistemically justified" distribution based on what we know suffices. In the case of physical constants, we have:
- Knowledge of theoretically allowed ranges (from theoretical physics).
- Knowledge of life-permitting ranges (from calculations).
- The principle of indifference within specified ranges.

This gives a "quasi-natural" distribution sufficient for the argument.

Second: the Comparison Range Argument. We don't need to calculate absolute probability. Comparison suffices: the probability of fine-tuning is much higher on design than on naturalism. Even with ambiguity in distributions, the relative difference remains enormous.

Example: Even if we disagree on the precise distribution for G, P(G in permitted range | design) >> P(G in permitted range | naturalism) remains true for any "reasonable" distribution.

Third: the Limited Sensitivity Argument. Collins conducts sensitivity analysis: how does the result change with different assumed distributions? He shows that for most "reasonable" distributions, the result remains in favor of design by a large margin. Only distributions "specially designed" to defeat the argument give different results.

Fourth: reliance on "natural distributions" from physics. In some cases, physics itself suggests "natural" distributions. For instance, in string theory, the solution space has mathematical structure that determines a natural measure. Collins leverages these cases to strengthen his argument.

Counter-responses from Hawthorne, Manson, and others

"Epistemic constraint is insufficient" response. Even with epistemic constraint, choice remains between multiple "epistemically reasonable" distributions. For example: should we use uniform or log-uniform over ranges? Both are "reasonable," but give different results.

"Comparison needs numbers" response. The claim that P(tuning | design) >> P(tuning | naturalism) needs quantitative specification. Without specific numbers, how do we know the difference is "large enough" to support the conclusion?

"Distributions from physics are not neutral" response. Distributions derived from specific physical theories (like string theory) assume the correctness of those theories. This introduces additional assumptions into the argument.

Recent developments (2020-2026)

The "Objective Bayesianism" current. Attempts to develop "objective" methods for choosing prior distributions, based on information theory principles (maximum entropy, etc). Luke Barnes and others apply this to fine-tuning.

The "Topological Analysis" current. Instead of focusing on numerical probabilities, studying the topological structure of parameter space. Are life-permitting regions "topologically rare" regardless of the specific measure?

The "Non-Bayesian Arguments" current. Developing formulations of fine-tuning that don't depend on precise probabilistic calculations, but on other concepts (complexity, simplicity, coherence).

Assessment from the perspective of rational plausibility (rajḥān ʿaqlī)

The Hawthorne-Manson objection reveals genuine difficulty in the rigorous Bayesian formulation of the fine-tuning argument. However:

1. Technical difficulty doesn't invalidate the basic intuition: observed fine-tuning needs explanation.
2. Collins's responses, despite not being completely decisive, show that the argument retains considerable strength even with probabilistic ambiguity.
3. The existence of alternative formulations (non-rigidly Bayesian) strengthens the argument's general position.

In the framework of rational plausibility: fine-tuning remains strong evidence contributing to the cumulative balance in favor of design, even if we cannot determine its precise strength numerically. The technical problem reduces claims of certainty, but doesn't invalidate the evidence's indicative force.

Where we stand on this debate today

The debate between Collins and his critics hasn't been closed, but it transformed qualitatively in the 2020-2026 period. There are three notable developments:

First, Luke Barnes's work (2020-2023) strengthened the design position through detailed physical analyses showing that the "rarity" of life-permitting regions in parameter space remains constant across multiple and diverse measures, thus reducing the severity of the measure-selection problem. Second, the development of objective Bayesianism provided more rigorous tools for justifying prior distributions, especially through maximum entropy principles, though their application to infinite spaces remains subject to technical disagreement. Third, alternative formulations emerged that transcend the strictly Bayesian framework entirely, such as arguments based on explanatory simplicity or the topological structure of parameter space, meaning that the fate of the fine-tuning argument doesn't depend entirely on resolving the Hawthorne-Manson problem.

Current status: The low probability problem hasn't been definitively resolved within its narrow Bayesian framework, but the general fine-tuning argument has diversified its tools, making it more flexible and harder to reduce to a single point of weakness. Today's debate is less polarized and more technical, occurring at the intersection of mathematical physics, philosophy of probability, and philosophy of religion.

For further reading

- John Hawthorne & Neil Manson, "The Fine-Tuning Argument, the Anthropic Principle, and the Problem of Priors" (2009)
- Robin Collins, "The Teleological Argument" in The Blackwell Companion to Natural Theology (2009)
- Robin Collins, "The Fine-Tuning Evidence is Convincing" (2012)
- Luke Barnes, "A Reasonable Little Question: Fine-Tuning and Naturalism" (2019)
- William Lane Craig & Sean Carroll Debate (2014) - The fine-tuning section
- "Formulation: Fine-Tuning Argument" page on the website
- "Objection: Probability Assignment Problem" page on the website