home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Simtel MSDOS 1992 December
/
simtel1292_SIMTEL_1292_Walnut_Creek.iso
/
msdos
/
hypertxt
/
hyper.arc
/
FILE72
< prev
next >
Wrap
Text File
|
1988-02-01
|
4KB
|
90 lines
Spreadsheets -- The good and the bad!
====================================
Given my background -- an undergraduate degree in math and a graduate degree in
a field that emphasized Operations Research -- you'd think I'd jump for joy at
the mention of spreadsheets. See example spreadsheet <FILE16>
Well, I did for a while. But frankly, I haven't used a spreadsheet in five
years, and I don't plan to again for the rest of my life. Is it spreadsheet
phobia -- or is there life after Lotus, Excel, and Quatro?
Here's why I don't use spreadsheets. They're the wrong tool, are poorly
designed, and I'd rather think. Let's look at each.
Spreadsheets are the wrong tools
================================
In "THRIVING ON CHAOS," the central theme is that the U.S. is headed for
disaster for two reasons -- obsession with bigness and automation.
In contrast, the Japanese with constant emphasis on small production runs,
constant improvement, and value-added labor can rapidly adapt, modify, and
differentiate products in niche markets. For example, how do you run a
spreadsheet in your market for the next six months if:
┌───────────────────────────────────────────────────────────────────────────┐
│ A major new Korean company enters your market │
│ Five Japanese companies are each aggressively improving their product │
│ Seven US companies sell their unprofitable divisions to hot-shot managers│
│ The exchange rate affecting your supplies varies by 50% │
│ The energy costs fluctuate by 25% │
└───────────────────────────────────────────────────────────────────────────┘
The conclusion -- mathematical modeling and strategic planning are the wrong ways
to survive such markets. Instead, meticulous attention to customers and
rapid incremental product improvements are the keys to the future.
Poor Design
===========
With the electronic spreadsheet now ten years old, by now they should
handle simple things like -- significant digits and standard deviation.
For example, when you multiply numbers of various accuracy (123, 4x10 to
the 3rd power) the answer is accurate only to the least significant digit
(in this case one). Yet, spreadsheets mix numbers of various accuracies
without regard to the base rules of arithmetic.
To be a bit more critical, spreadsheets don't handle numbers with
associated standard deviations (the normal way to measure accuracy). Why?
The answer is obvious. After a few arithmetic operations that include
propagating the measures of accuracy, most users are surprised to discover
standard deviations are larger than their answers. Colloquially,
multiplying mush produces exponential growth in mush -- and no one wants
accurate useless answers. That's the reason software for producing
inaccurate but precise numbers appeals to certain markets. If people want
numbers to believe in, spreadsheets are certainly a religion.
However, that raises the question of spreadsheet productivity. Is the new
generation of faster spreadsheets just for those who measure success by the
size of the crank handle on their adding machine? If the newer
spreadsheets are really four times faster, do you need 1/4 of the producers
or 4 times as many output readers. Do greenbar sheets of numbers expand to
fill the space (or pocketbooks) available? You answer it.
I'd rather think!
=================
The heart of most number manipulation is only verification. Seldom
does a spreadsheet creator shout "Eureka" and less seldom do such insights
ever relate to reality. That's the central problem in mathematical
modeling...the environment usually changes faster than the model.
Mathematical models are naturally biased to be more wrong than right.
I find many more ideas from using MaxThink and HOUDINI than from
spreadsheets. From my viewpoint, finding insights and discoveries are more
satisfying to me and maybe more useful to others than running a high-priced
adding machine.
REFERENCES :------------------------------
Spreadsheet thinking <FILE72>
Spreadsheet problems <FILE27>
Neil Larson 1/14/88 FILE72
44 Rincon Rd., Kensington, CA 94707
Copyright MaxThink 1988 -- Call 415-428-0104 for permission to reprint