But it may be said:

An infinite regress of non-existent things is not like an infinite regress of existent things. Rather, the former is possible, unlike the latter, and the coming about of the origination of temporally originated events depends upon an infinite regress of non-existent things.

The reply is:

If an infinite regress of non-existent things is not something real, it has no reality. The possibility of temporally originated events arising would then depend upon something that has no reality. This is false. If, however, their regress is something real, then it has been established that an infinite regress of real things is admissible and that it is beginningless, even though each of those regresses is not beginningless. This contradicts what they have said concerning the impossibility of an infinite regress of temporally originated events.

They therefore face two alternatives: either they affirm preference without a determining factor, or they affirm the admissibility of an infinite regress.

This is precisely the dilemma they face when they say that temporally originated events must have a beginning. Just as this claim commits them either to preference without a determining factor or to an infinite regress, so too does their claim (that the possibility of these events must have a beginning) commit them to one or the other.

Preference without a complete determining factor is impossible. They agree that preference without an agent that determines it is impossible, but they do not require that what determines the preference be complete. Rather, they say:

Complete preference arises without preponderance arising, and preponderance arises without a complete determining factor. This is based on the claim that an agent endowed with power gives preference to one of two things within his power without a determining factor, since the beginningless will gives preference to one of two equivalent alternatives without a determining factor. And affirming the admissibility of an infinite regress