-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Phil Henshaw on 12/16/2007 10:24 AM: > from your 12/10 post "In the end, most of the "shapes in the fog" _can_ > be identified, named, and described. But, some of them resist. " > > I'd say it backward, sort of, that "there are things beyond our > knowledge that are worth exploring", or "That it seems worth the trouble > and cast mistrust on one's own operating assumptions in order to puzzle > over 'shapes in the fog' since some lead to wonderful discoveries and > they all test and hone our method of finding them." I would technically > disagree with "most ... _can_ be identified, named and described". To > even count them I think we need to have already IND'd them. The key > moment in that process is when we 'get our hooks on' something, when our > grasp of something new has reached a point where we see that more effort > will produce a positive return. For me that's often coincides with the > time when I can identify the tell tale signs of a natural system > beginning or ending.
Well, I'm not going to vehemently disagree with you about the IND before counting; but, I am going to point out your reliance on sequentiality because it relates below as well. First, though, I want to defend my statement that the "shapes in the fog" (SOF) _can_ be IND&C. The primary thread, here, has been that all models are always false. And that means that it is not only possible but common and reasonable to identify, name, describe, and count things of which we're partly (or mostly) ignorant. So, you may be right if you said that we may not be able to IND&C all the SOF perfectly accurately. But, we _can_ IND&C all the SOF inaccurately and imperfectly. But, more importantly, the IND&C is not a sequential process. We don't IND a shape and _then_ count it. We're always doing some subset of the 4 on some subset of the SOF concurrently and at varying degrees of accuracy. I know I'm preaching to the choir, here. But, I wanted to make that form of methodological non-linearity clear. > taking your four modeling principles, > R1: co-simulation of 3 models: > M1: a synthetic model, > M2: a reference/pedigreed model, and > M3: a data model > R2: inter-aspect influences are discrete > R3: models are designed to be transpersonal > R4: well-defined similarity measure(s) > Well, to start out, these are not sequential but concurrent principles. > I see outward search in R1, but more traditional problem solving > toward a deterministic result in R2 R3 & R4, i.e. first 'searching out > of the box' followed by 'working in the box'. I think it's good to > explicitly focus on a process of alternating 'search' and 'work' tasks, > continually asking, am I asking the right questions, etc. > > For an example, you might extract a network of nodes and interactions > from a complex system. That projects (reduces) the natural physical > object onto a certain plane of definable relationships suitable for > analysis. Then make your 3 types of models. Once you do the analysis > you might try seeing how it fits back in the subject, what's happening > in the physical system to make the identified features of the network > possible, and what's changing in the physical system that makes the > features impossible, etc. I object to the sequentiality. These principles don't really work if you apply 1 _then_ apply the other. They should all apply, in varying degrees under varying circumstances, concurrently. So, it is weak to adhere to R2-R4 and _then_ coerce your methods so that they adhere to R1. _Weak_ but not misguided. But your response above makes me think I've done a bad job outlining these principles. R1-R4 are not constructive. You seem to be using them as if they were constructive methods rather than selective principles. It doesn't really matter where the models come from, how they were constructed, or whether they even match R1-R4 in the first place. But, a good multi-modeling selection methodology will use all of R1-R4 (if not more, as Marcus haphazardly points out). It helps to use the word "apply" rather than the word "use"... as if you were applying a predicate to pre-existing models. - -- glen e. p. ropella, 971-219-3846, http://tempusdictum.com There is nothing so bad that politics cannot make it worse. -- Thomas Sowell -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFHdXlMZeB+vOTnLkoRAtl+AJ9Yqq0DpqXsWzm2n/IC7mn11yZ6AwCg31tY uwn/wuElukxEb1uf6xUJ09k= =go1M -----END PGP SIGNATURE----- ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org