Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!decwrl!labrea!siegman@sierra.Stanford.EDU
From: siegman@sierra.Stanford.EDU (Anthony E. Siegman)
Newsgroups: comp.sys.mac
Subject: TK/Solver! as a Technical Tool: What Happened To It? (long)
Message-ID: <81@sierra.stanford.edu>
Date: 25 Mar 89 21:48:36 GMT
Sender: siegman@sierra.STANFORD.EDU (Anthony E. Siegman)
Reply-To: siegman@sierra.UUCP (Anthony E. Siegman)
Distribution: comp.sys.mac, comp.edu
Organization: Stanford University
Lines: 65


In Germany in 1984 a colleague showed me a version of "TK/Solver!"
(sp?) for the Macintosh which ran on a 512K Fat Mac, and impressed
me very much.  You typed all the equations for a given problem into
an Equation Window.  These did not have to be simple assignment
statements, or be entered in any special order; you could type
equations like

     sin(2*pi*x+theta) + c = a*x^2 + b*y^2

with functions and variables on both sides of the = sign.

As you typed equations into the Equation Window, all the variables
you used were automatically collected in a Variables Window, where
you could either assign values to them, if they were intended to be
inputs, or leave them unassigned if they were intended to be
outputs.  There was full editing in both windows, and a simple
procedure to step any of the input variables through a sequence of
values or a range with controlled limits and step size.

When you had everything entered, you clicked a "SOLVE" button and
the program implicitly solved the complete set of equations using
the specified inputs to find all the unknown outputs, assuming you
had enough equations to determine a solution.  No programming or
worrying about how to do the iterations was involved.  If the input
contained a sequence of values, the output could be a table of
outputs versus inputs, or an auto-scaled plot of any variable
versus any other variable.  The program had all the usual
capabilities for Saving and Opening sets of equations, exporting
results, printing results, and so on, and the whole system seemed
to work beautifully.

It seemed to me this program was obviously going to be the analog
for the scientific and engineering world of the spreadsheet for the
business and financial world -- the VisiCalc for the slide rule
set.  I would have sworn that it would spread like wildfire -- I
had visions of students doing problem assignments with zero
brainpower by just typing in all the equations in the textbook and
values for all the known variables, and letting SOLVE find all the
unknowns.

Instead, this version of TK/Solver! was never widely distributed
and now seems to have disappeared completely (I gather there were
commercial difficulties with the software firm involved).  More
surprisingly, no major competitors have appeared, even though this
would seem to have been a natural, and even though versions of
TK/Solver!  exist for other machines, e.g. for hp desktop machines.
The only similar program I know of for the Mac is Borland's
"Eureka", which seemed to me to be a poorer-quality version of what
I remember seeing in Germany, and which does not seem to have had
much success either.

Of course Mathematica and other considerably more complex programs
are becoming available now, though Mathematica is very much more
expensive and requires a fully loaded Mac II.  But why did
TK/Solver! or some similar program for the Mac never take off the
way spreadsheets did?  Are there just a lot more business and
financial spreadsheet customers than scientific and engineering
customers out there? Or was it just a fluke that this idea did not
get picked up by others and spread more widely?  Even with
Mathematica today, I'll still like to have a small, simple,
inexpensive TK/Solver! descendent to run on my Mac Plus.  What
happened?

--------