[For myself: Copied from the 7/28/2003 Science [43637] posting.]

Incidentally, when Feynman talks about "laws" (e.g., he says something like "They do not find any laws" talking about social sciences), he is talking about laws that relates two or more "numbers" (i.e., the numerical values of certain things, time duration, distance (the difference in two positions, for example), weight, etc.) They DO compute a consequence about numbers from "laws" about certain numbers (i.e., values of certain things). We do not compute a number by our hypotheses; we "only" deduce impossibility and the lack thereof from our hypotheses. And while the physicist checks to make sure that they do deduce the consequence correctly, we must check to make sure, as best we can, that we do deduce our consequences from our hypotheses. It is by doing so and by making sure, as much as possible, not to use hypotheses that have been shown to be invalid or those whose validity has not been checked before, that we can ensure the significance of the result of our Experiments and learn from errors (i.e., from the failure to obtain experimental results as predicted).