I mean easy to discern. Jim Bromer
Why nod - just start with simple projections of conjectured relations. On Tue, Jun 24, 2014 at 6:51 PM, Jim Bromer <[email protected]> wrote: > Failure Analysis presupposes a fine discrimination of what constitutes a > failure. The ability to reliably create a good theory in response to a > failure requires intelligence as a pre-requisite. I studied a lot different > kinds of ideas like this 20 or 30 years ago. But, as far as I am concerned > the real question now is how can we reliably create a better theory from > earlier theories. Part of the problem is that we may, in many cases, not > have a good criteria to determine how good the theory is and a good theory > may fail because a minor part of it is not reliable. > The kinds of discussions that we have in these groups are demonstrations > of this problem but they are usually complicated theories that are largely > based on basic assumptions that we can develop well. (Notice that some > people have assumptions that seem obviously wrong and there can be other > assumptions that people disagree with even though they cannot substantiate > their point of view. For example, some people think my theory that a > polynomial time solution to logical satisfiability is quite possible seems > like blatant nonsense to some people, while I think their certainty about > something which is famously unproven is based on some pretty insubstantial > premises. > > So what do I do to try to create a better theory. I keep trying different > things and if the results do not produce any improvements then I have to > adjust my expectations of the feasibility of finding a solution. On the > other hand, if I find something that works in a variety of simple cases > then I can keep trying more complicated cases to see if it works within the > significant characteristics of the problem. (Like looking at the increase > in complexity for the worse cases as the number of variables increase, and > then examining a range of problems in well-known formats like 3-SAT.) > > But failure in Logical Satisfiability is easy to discriminate as long as > you are careful. > > Jim Bromer > > Why nod - just start with simple projections of conjectured relations. > > > On Tue, Jun 24, 2014 at 2:53 AM, Piaget Modeler via AGI <[email protected]> > wrote: > >> I'm just beginning this enquiry, >> >> Found this: http://research.me.udel.edu/~jglancey/FailureAnalysis.pdf >> >> as well as a pointer to Anticipatory Failure Determination (AFD) >> >> Cheers so far... >> >> ~PM >> >> ------------------------------ >> From: [email protected] >> To: [email protected] >> Subject: [agi] Failure Predictors >> Date: Mon, 23 Jun 2014 23:13:51 -0700 >> >> >> Wonder if anyone is still alive on this thread besides Jim B. >> >> I guess it's the summer. >> >> >> When we experience an expectation failure, somehow I think we form a >> theory of the failure, and seek to elaborate that theory. >> >> I don't have the right keywords but is there any literature or research >> on creating failure predictors, or forming failure hypotheses? >> >> Kindly advise. >> >> ~PM >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >> <https://www.listbox.com/member/archive/rss/303/19999924-4a978ccc> | >> Modify <https://www.listbox.com/member/?&;> Your Subscription >> <http://www.listbox.com> >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >> <https://www.listbox.com/member/archive/rss/303/24379807-f5817f28> | >> Modify >> <https://www.listbox.com/member/?&;> >> Your Subscription <http://www.listbox.com> >> > > ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
