Well, It is presented as an alternative to determinism; used to address the question: If an event has no cause then by what means does it arise? So far the only option I've ever come across is that it was random. It could just as well have not materialized as done so. And this is just as abhorrent to "free willers" as is determinism, at which point the discussion usually comes to an abrupt end or digresses into irrelevant issues.
IMO, "random" is not even an option as "cause" (it is, in essence, an effect, the unpredictable outcome). As far as I'm concerned, Carlin is correct in at least the assertion that the uncaused cause cannot happen in the framework of determinism. The reason, though, is because of our innate ability to as
sign (meaning, to the world). There will never not be "cause," not as long as we're around (to construct the determinstic framework).
The better alternative for an image of "free will" (one I've already mentioned) are the epistomological models that, rather than trying to revise the ontological model to squeeze uncaused causes into (or out of) causation, choose a different perspective from which to view the issue, the perspective of our own conscious participation in the universe. And really, isn't that what "free will" is all about? Us, doing things, wilfully. Asserting our being.
Here's me. There's you. Here's a moment in time, a moment when something in particular, something
significant, is happening to one or both of us. We participate in the event, even if only observing; we participate, in whatever actions and thoughts we do; together, "
we" must be a necessary part of the model of what is going on with the issue of "free will". By whatever flavour of free will one might define, our conscious participation is vital for "free will" to have happened. We are here, a part of the universe, and our minds are a part of the objective sequence of events taking place. To minimize the role of our participation in the sequence of events taking place in the universe (or worse, eliminate it) is to look at less than the whole picture (not saying you've done that, but it's a familiar approach) and to model less than a whole picture.
But I'm rambling now.