Suppose I win the lottery today. I tell myself that this was because God helped me win. Suppose I play the lottery next year and lose. I tell myself that this was because God wanted me to lose.

Either way, I attribute both my win and loss to this God. I also tell myself that God only cares about me and the only time God ever chose to intervene in the world was in getting me to win and lose the lottery.

Now, intuitively, this seems like an excuse. My hypothesis only seems to explain these two lottery events and nothing else going on in the world. Arguably, it doesn’t even explain that. There is no mechanism or anything else showing how God would do so. The naturalistic hypothesis seems to explain much more about the rest of the world and there’s nothing within that hypothesis that rules out me winning and losing a lottery. My God hypothesis gets away from having to explain the rest of the world since the very definition of my God is that He chooses to intervene only for my two lottery events.

Something about the naturalistic hypothesis explaining more about the world seems to make it more likely to be true. But why? I can’t seem to find a non circular justification for this. Why exactly is a naturalistic hypothesis more likely to be true than my tailored God hypothesis?

Anticipating some potential responses, one might say that one can invent an infinite number of supernatural hypotheses to explain my lottery win and loss, and thus the likelihood of my specific one being true is low. But this assumes the principle of indifference which we all know has issues. There is no reason to assume that each supernatural hypothesis is equally likely or what that would even mean.

Another possible response might be that my hypothesis is not falsifiable. But one could argue that so is the naturalistic hypothesis. How would I verify or falsify that my lottery wins were caused just by blind natural laws? It seems that one cannot do so without addressing alternative hypotheses, such as supernatural ones. But those alternative hypotheses are unfalsifiable making the naturalistic hypothesis unfalsifiable as well.

What then would justify believing in the naturalistic hypothesis over mine?

  • "Explaining more" typically means cohering with other established "explanations", i.e. theories of "the rest of the world". As such, it is more likely because those established theories provide corroborating evidence for it. By the way, this is the reason why a Bayesian would not make outlandish hypotheses in your other question more likely than chance. They are severely downgraded by evidence against them from "the rest of the world", which is ample in virtue of them being outlandish.
    – Conifold
    Commented Aug 28, 2023 at 23:25
  • Thank you for your comment. So in other words, to a Bayesian, this may still confirm (in a likelihood sense) the outlandish hypothesis but not enough to make up for the very low prior, correct? This all makes sense except for the fact that doesn’t this imply that basically every observation can confirm any hypothesis? John winning the lottery today technically confirms the hypothesis that God wanted him to win over it occurring by chance.
    – user62907
    Commented Aug 28, 2023 at 23:33
  • The point of a scientific theory is not to be true but useful. If a given theory explains more observed facts while not being in contradiction with other observations, it helps you make more useful predictions about a wider variety of problems, which makes it better than competiting theories. Simply put, by learning only one formula you can solve more problems. Note that in your example the God hypothesis has tremendous explanatory power: there is simply nothing yu can't explain with it. But it has no predictive power, you can't use it to reliably evaluate the probability of winning next time.
    – armand
    Commented Aug 29, 2023 at 0:20
  • Confirm or disconfirm, yes, this is called holism:"Our statements about the external world face the tribunal of sense experience not individually but only as a corporate body", Quine. Of course, even Quine walked away from radical holism of everything on everything. There are autonomous chunks of theory/evidence that influence credibility of other chunks very little, if at all, and even within chunks the influence is a matter of degree. But moderate holism is a feature and not a bug, and largely accepted.
    – Conifold
    Commented Aug 29, 2023 at 0:45
  • The intuitive first principle though not logically proved justification is nothing but the razor of Ocame which on a whole is the best satisfactory justification either by internalist or externalist/reliabilist standard... Commented Aug 30, 2023 at 18:18

1 Answer 1


Either you've made predictions of future measurements, or all you've done is to rename the phenomenon.

The definition of God

Stop right there. God is a word in common use with all kinds of predictions of future measurements built into it. Defining words to mean things radically different from their common use obfuscates meaning. Let's use X.

The definition of X is that He

He is an implicit prediction about measurements of X.

He chooses to intervene only for my two lottery events

Capacity for choice and intervention are predictions about measurements of X.

If you don't want to make predictions (which we can then go try to measure, and hence can be falsified), you can take the predictive components out. What do we get?

The definition of X is the cause of my two lottery events.

You must log in to answer this question.