Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!tut.cis.ohio-state.edu!bgsu-stu!klopfens
From: klopfens@bgsu-stu.UUCP (Bruce Klopfenstein)
Newsgroups: comp.society.futures
Subject: Assumption Drag in Forecasting
Keywords: forecasting
Message-ID: <5479@bgsu-stu.UUCP>
Date: 28 Feb 90 19:05:47 GMT
Organization: Bowling Green State University B.G., Oh.
Lines: 45


I have read with interest the postings so far on retrospective 
forecasting.  I have another area which I hope prompts some 
reactions.  William Ascher in his book on Forecasting examined 
past technological forecasts for a number of items including 
computer capabilities.  One of his findings that I found 
extremely interesting is that of "assumption drag," the reliance 
of previously stated assumptions in spite of empirical evidence 
to the contrary.

As a snow lover, I have witnessed this phenomenon in weather 
forecasting repeatedly.  At 3 PM, the National Weather Service 
issues a winter storm warning (not a watch) for that evening with 
heavy snow expected.  The snow does not develop.  By 9 PM, the 
updated forecast often will continue with the warning or at least 
a forecasting of significant amounts of snow.  The following 
morning there may be 2 inches of snow on the ground.  The same 
thing happened here a few days ago.  By 9 PM it was completely 
clear outside while the forecast had been for a 60% chance of 
accumulating snow.  The 3 AM forecast on that next day still said 
cloudy with snow flurries.  It was sunny with very few clouds.  
ALthough this is anecdotal evidence, I have witnessed it MANY 
times (my attention is heighted by forecasts of snow).  I wonder 
why the forecasters don't check the local conditions to see that 
they contradict the forecast itself.  It's as if so much effort 
and data went into the forecast, that the forecast is not changed 
even in the face of empirical evidence to the contrary.

I don't believe Ascher explains why assumption drag happens (I 
may be wrong) but rather that n it simply does happen.  Besides the 
explanation that so much work went into a forecast that the 
author is reticent to change it for that reason alone, the other 
(perhaps additional) explanation is that the short-term results 
from the environment could be the anomoly, and the forecast 
assumptions remain "valid" in the mind of the forecaster.

I wonder how many analogies there are in computer forecasting.  
(By the way, no need to explicate weather forecasters--I was just 
using that as an example to clarify assumption drag.)
-- 
Dr. Bruce C. Klopfenstein      |  klopfens@andy.bgsu.edu
Radio-TV-Film Department       |  klopfenstein@bgsuopie.bitnet
Bowling Green $tate University |  klopfens@bgsuvax.UUCP
Bowling Green, OH  43403       |  (419) 372-2138; 352-4818
                               |  fax (419) 372-2300