Sure, data is good to have, but what would past behavior tell you about the novel aspects of new behavior, and how would you be able to tell that from incomplete data on the past, anyway? For original behavior it would seem the usual tools don't help much, unless you do what I do and. turn models around to study what about individual behaviors they miss.
Phil Sent from my Verizon Wireless BlackBerry -----Original Message----- From: "Marcus G. Daniels" <[EMAIL PROTECTED]> Date: Mon, 12 Nov 2007 10:17:06 To:The Friday Morning Applied Complexity Coffee Group <friam@redfish.com> Cc:[EMAIL PROTECTED], [EMAIL PROTECTED], "caleb.thompson" <[EMAIL PROTECTED]> Subject: Re: [FRIAM] FRIAM and causality Nicholas Thompson wrote: > > To say that X is the cause of Y is to accuse X of Y. Given my > current belief that story-telling is at the base of EVERYTHING, I > think you convince somebody that X is the cause of Y just by telling > the most reasonable story in which it seems obvious that Y would not > have occurred had not X occurred. > I'm skeptical of the tradition that says we should have predictive models before measuring things in the world or interpreting data. Where does a hypothesis come from? I'd say it is little more than the prior expectations we have about how the world molded into a compact if/then type of story. And just because a model says to measure certain things (out a large universe of possible things to measure) doesn't mean the prescribed measurements are really independent samples, as there is some bias from a scientific culture. Given the advanced technology that exists for automated data collection, let's put aside the story telling (and the dogma that often underlies it) to see if the priors look very promising. For example, using machine learning techniques, infer models from partial data, and then predict the rest. Human experts are often wrong or in conflict, and not always a good source for setting prior expectations. Machines can help with that, by considering thousands or millions of possible explanations for phenomena based on a small number features found in a larger space of observables. When so found using a simple, statistically-sound metric, I really think the `experts' need to look at that result pretty hard. Marcus ============================================================ 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 ============================================================ 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
