Simple, the process of formulating and testing hypothesis can be applied to itself (self-reference). Reasoning does not need to be hierarchal. Imagine a network of facts: A, B, C, D, E, F say. A can support B, B can support A, C can support A... (each supports the others).Can you give me an example of a self-referential process which tests hypotheses, i.e., which tests its own validity as a test? Moreover, how do you know that A supports B, say? Isn't that a hypothesis itself?
The critera for judging which hypotheses are 'better' are themselves hypotheses open to improvement. Okay, but can you give me an example of this? Further, is the meta-hypothesis you've just invoked now the subject of an even larger hypothesis for it to have any weight? Sounds like an infinite regress to me.
All that is needed for reasoning is a relative comparison of competing hypotheses.Don't you need some kind of standard to compare things if you want your comparison to have any meaning? We can't have science running around stating things like "hypothesis A is better than B is better than C is better than A," as above, using each as the standard for the other. What does such a statement mean?
Think of capitalism as an analogy. Imagine multiple competing businesses. All that is needed for a business to win the market is that it be better than all the others. It does not need to be perfect, just better than the others. It's the same with reasoning. Science does not work by trying to find the absolute truth. It works by weighing up competing hypotheses and choosing the one that is best relative to the others.Here, one has a unit of currency to determine to determine without any self-referential scheme which business has the most money. Assuming no one is tied, you can always determine who has the best business.
In any case, I'm not sure I'm getting the gist of your response-- some examples of these self-testing tests and standardless comparisons would help.