Free Republic
Browse · Search
News/Activism
Topics · Post Article

To: Tolik

Thomas Sowell's first job as an economist after graduating from Harvard was working for AT&T making economic predictions, using mathematical models. His boss showed him were the previous models (stacks of punch cards) where kept, so he could use them as a starting point. Young Thomas says "Oh, great, then we can see how well we've done!" His boss's reaction made it clear that he was *never* to bring up that idea again.

Actually, regression analysis allows one to place reliable bounds on the accuracy of a model, like when astronomers say that there is a 1 in 45,000 chance of a certain asteriod striking earth in 2036 (or whatever).

The IPCC seems to be paying homage to notion of error bounds by saying that there is a 90% chance of some {poorly defined unfavorable} event. What the author of this piece is talking about is what is known as validation. Historical data is perfectly good for validating a model, provided you can establish *all* the relevant input parameters and their associated uncertainties. I am personally unimpressed by anything I've read or seen about global warming to date. The alarmist scenarios are just too convenient for the politcal Left.


14 posted on 02/28/2007 9:19:06 AM PST by Lonesome in Massachussets (When I search out the massed wheeling circles of the stars, my feet no longer touch the earth)
[ Post Reply | Private Reply | To 1 | View Replies ]


To: Lonesome in Massachussets
Historical data is perfectly good for validating a model

But not if the model is inferred in whole or in part from that historical data. That can easily happen even without the modeler's knowledge. The only sure way to avoid that problem is to test a prediction about data as yet unknown.

22 posted on 02/28/2007 10:14:42 AM PST by edsheppa
[ Post Reply | Private Reply | To 14 | View Replies ]

Free Republic
Browse · Search
News/Activism
Topics · Post Article


FreeRepublic, LLC, PO BOX 9771, FRESNO, CA 93794
FreeRepublic.com is powered by software copyright 2000-2008 John Robinson