Re: Rules of Thumb

[Kromkowski@aol.com]

>>In a series of lectures to Ford statisticians, Deming once showed an
example of where eyeballing was the best early indicator of a difference
between three processes. I don't think he was against eyeballing, per
se; he was against eyeballing when there were better tests available.
And I think he was certainly FOR gaining enough profound knowledge to
know the difference.<<

I believe I am familiar with this part of the lectures.  They were posted on 
DEN.  

1) The point in the lecture was a hypothetical situation.  Nonetheless, as I 
recall, the was an assumption, something like, "we don't feel good about this 
rivet connection or something similar".  In other words, there was an assumed 
consensus that something better was required.  The messy details of how we 
got to this assumption were left out.

2)  Deming specifically writes about the need for rules IN ADVANCE to be used 
for the future.  The problem with "rules of thumb" and "eyeballing"  (which 
are really one in the same) are that they are ad hoc, and rationalized post 
hoc, by ignoring all the times it didn't work out right.  See OOTC, 319-320.  
If one needs to use professional judgment, use it courageously; but don't try 
to cya by the ad hoc use of data.  (This is what gets me the most -- putting 
on the cloak of science.)  It's this use of "rules of thumb" that creates 
mistrust of science and mistrust of the use of data which leads to using no 
data; i.e., the erroneous encouragement of the other side of action on the 
last data point.  The point is to use theory.  Explain everything in the past 
with the goal of predicting everything in the future.  Rules of thumb don't 
do that.

3) I submit that Deming was against eyeballing per se, although he 
acknowledges his own frailty!  He writes:

    "It is a hazard to use judgment to distinguish between special causes and 
common causes.  Judgment has been wrong every time so far; see Examples 1 and 
2 on pages 356 and 357.  The naked eye looking at figures __is not a safe 
guide__, though I am certainly guilty of using the eyeball method in 
__extreme__ circumstances."  OOTC at 320.  (Emphasis supplied.)

He previously had noted:

    "Good management and good supervision __require__ knowledge of the 
calculations that will separate two kinds of cause."  OOTC at 316. (Emphasis 
supplied.)  

See also, OOTC at 278, under heading "Practice is more exacting than pure 
science; more exacting than teaching."

As to this statement: "he was against eyeballing when there were better tests 
available.  And I think he was certainly FOR gaining enough profound 
knowledge to know the difference."

You and I are nearly uniformly coming from the same page on most issues on 
the DEN, but I think this kind of thing leads us down the road of gnosticism. 
 How is it that any one would know whether a professional or manager or 
leader by authority of office has enough of this "profound knowledge" 
(whatever that might actually be)?  Of course, one would know by having 
"profound knowledge".  This is the slippery slope that leads ever so slowing 
and imperceptibly to bad things.  
--
John David Kromkowski
Attorney at Law
6600 York Road - Suite 108
Baltimore, Maryland   21212-2028
Kromkowski@aol.comTelephone:  (410) 377-6248
Facsimile:  (410) 372-0624
======================================================================