1Re: What do users REALLY want?0

["Leslie A. Walko" <attila@FLASH.BELLCORE.COM>]

    Date: Thu, 21 Dec 89 16:20 EST
    From: Brad Miller <miller@CS.ROCHESTER.EDU>

1	Date: Thu, 21 Dec 89 12:10 EST
	From: attila@flash.bellcore.com (Leslie A. Walko)


	As far as the difficulty of porting is concerned, Joshua would be easy.

    I may well be confused, but isn't writing code for Joshua equivalent to
    writing code for, e.g. flavors (I'm thinking of define-predicate-method). In
    that case, don't you effectively have to port all of Genera to have a useful
    Joshua (i.e. so you can compile, load, etc. all your modifications and
    functions that need to be on the defined hooks)?

    Or is the idea to provide Joshua on some other development platform?

0Yes, what I had in mind was to provide Joshua on top of another
platform.  It seems that Symbolics has a tool around that allowed
rewriting of Flavors into New Flavors syntax; they suggested that the
same could be done to move New Flavors to CLOS.  The developers I spoke
with also suggested that the machine dependent performance features of
Joshua are minimal.  All in all, they implied that the overall port to
Unix would be in the more than one month but less than one man-year.
This suggests that the cost is trivial, while the benefit potential is
great.

MY PRINCIPAL CONCERN IS NOT ABOUT TECHNICALITIES HOWEVER.  The lispm
community seems to be splendid at technicalities, but a little high
handed about business rationale.  I was really more interested in
generating some discussion on how the user community feels about a
change in Symbolics corporate strategy.  To put in more succinctly, a
strategy comprised of
	1. pushing hardware out the door as fast as possible
	   via dramatic price cuts (selling close to *real* unit cost)
	2. marketing and pricing development environment separate from
	   hardware
	3. porting layered software to Unix platforms.

The reason why I think #1 is good is because it keeps the existing user
base and could [once more] GENERATE AN *INCREASE* IN THE USER BASE.
Adopting #2 is good because it bring Symbolics in line with the market,
and ALLOWS DROPPING THE PRICE ON THE HARDWARE.  It also allows better
price/production controls and more flexible marketing strategies.
Lastly, #3 is good because it penetrates new markets where Symbolics
could not compete before because its OS is not SOE (Standard Operating
Environment) or due to its high cost; it will also increase software
UNIT VOLUME generating more revenue and feed back to superior product
development; it also builds a safety net for Symbolics in case the
hardware market becomes a loosing business again.

It is especially important to point out, that to survive as a viable
business it is necessary to be able to IDENTIFY PROFIT AND LOSS centers.
If the product is bundled, the hardware operations cannot be separated
from the software operations and cannot be managed by anything other
than guesswork.

So all, yea amateur economists, speak up!

Leslie Walko	attila@bellcore.com