Hi everyone. I wonder what you may make of the following cosmological argument?

Consider some effect belonging to A, where "effect" signifies either a process going on in A – such as change – or A's coming-to-be. Because an effect is something of which we can ask ‘why?,’ it is something which we take to require a cause; and that it *must* have a cause is readily seen by the fact that nothing can cause itself, that is, that nothing can be its own cause. So, an effect is something that requires a cause distinct from itself. This cause does not have to be temporally prior, but it must be *logically prior* to the effect it produces.

It seems clear, then, that there must be some logically prior cause B responsible for the production of the effect in A. Now, either B is responsible for the production of the effect in A in a *proximate* way, or in an *ultimate* way. In the former case, B turns out to be no different than A. For although B is the cause of the effect in A, the same effect present in A is also present in B – that is, B is subject to change as much as A is, and B has come into existence just as much as A has. B’s power to cause the effect in A turns out in the end to be wholly derivative. In the latter case, however, B is the cause of the effect in A while in no way sharing the same effect as A. B’s power to cause the effect in A is *absolute*, that is, *non-derivative*. No further cause is need – nor can be given – to explain the effect in A. We have reached the end of our explanation.

But suppose that B is responsible for the production of the effect in A only in a proximate sense. This means that there must be some logically prior cause C responsible for the effect in B; and suppose that C stands in explanation of the effect in B also as a proximate cause. Here, too, there must be some logically prior cause D responsible for the effect in C; and so on and on...to infinity. [1]

It should be clear that, although we have, in one sense, *adequately* explained what accounts, not only for the effect in A, but also for the effect in B, C, and so on – namely, in terms of immediate (or proximate) causes – there is a more fundamental sense in which we haven’t explained much at all. In other words, we may have explained why the effect in question belongs to this-or-that particular member; but we have not explained why the effect, *as such*, exists. At every step we have posited a cause for the effect in A that is *itself* subject to the same effect, and hence in need of a cause just as much as A is. It would be no good simply to say that A undergoes change because of something that itself undergoes change; or that A’s coming-to-be is explained by something which itself comes-to-be. And it would be insult to injury to further insist that it suffices to carry out this mode of explanation into infinity.

To see this point more clearly, let's lump together the whole nexus of cause/effect composites (A, B, C…) into a single collection (call it X). Since every member of X stands in need of a cause, we can say that every member of X is, in a sense, an effect. But then X is an effect also, since "if you take a group of things that are all effects, the collection itself will be an effect." [2] Hence X requires a cause. Now the cause of X cannot itself be an effect. For if it were it would be a *member* of X, and would thus be the equivalent of saying that X causes itself, which is impossible [3]. Hence the cause of X must be a cause that is not itself an effect.

--------

As you can see, I take "effect" to be synonymous with both change and efficient causation. If the argument succeeds, then it gets us to God qua Unchanged Changer, qua Uncaused Cause, and perhaps qua Necessary Being, since the cause of the collection X, by virtue of being uncaused, must be necessary.

[1] The problem of infinite regresses need not even arise, since it's being assumed from the outset that our causal chain stretches into infinity. I think this is nice.

[2] No doubt a critic will insist that I am committing a fallacy of composition. Here I would simply ask to see an invalid inference from parts to whole based on the principle: "if you take a group of things that are all effects, the collection itself will be an effect."

[3] This move is crucial to the argument, and therefore so is arguing, not from member A to its ultimate cause, but *from X to its ultimate cause*.