> > No: to know, you must first believe. What's
> > more, you're neglecting the objects of these
> > verbs, which breaks it out of the recursion.
> > Recursion is not bad: recursive programming is
> > the best way to solve certain problems.
>
> But in order to break out of a recursive loop, you must leave the loop when you've satisfied some condition.

I didn't express myself very well on that point. You don't actually "break out" of the recursive loop, because you never get to the point where you know everything, just as, with science, the cycle of learning never stops there either. Every question you answer, in ANY domain, tends to turn up more questions than it answers, and this is no different. It MUST be so, for if we cannot completely understand our universe, how could we completely understand its creator? But we can advance our knowledge nonetheless, and the reason the cyclic nature of it is ok is that each iteration is *different*. You're not trying to establish the same thing on the current iteration as you are on the previous. A flaw would only show up if you were.

A good analogy, perhaps, is a dictionary. A dictionary defines words in terms of...other words, whose meaning are determined in the dictionary. If you look up one word, you have to look up several more, and eventually you get back to the words you had to look up in the first place. Well, if someone handed you a Swahili dictionary, it's doubtful that you'd be able to learn the language, much less make a convincing argument to someone else about your conclusions, but if Someone gives those diligently seeking to learn the language the first handful of the words, you could look up definitions that use those words and learn others, and use them to learn others, and use them to learn others, and use them to learn others, and use them to learn others, and if it weren't obvious this is where the cyclic part comes in, and yet, even assuming an infinitely large dictionary, there is no logical fallacy in this kind of learning unless the initial words were wrong, and yet, if the initial words were wrong, how could this expansive and intricate dictionary make continue to make sense with every iteration? Eventually the statistical probability that the initial words were incorrect become so minutely small that, if one encounters a seeming inconsistency in the dictionary, the error has a far greater chance of being in misunderstanding of a definition in a recent iteration rather than an error in the whole kit and kaboodle.