Enclosed is the article ``The Computerless Computer Company'' by
Andrew Rappaport and Shmuel Halevi from The Harvard Business Review,
July-August 1991.  This article is relevant to the Free Software
Business Model discussion because the central idea of the AR/SH
Computerless Computer Company thesis is that at this point in history
there is a glut of know-how in the manufacturing domain.  Thus the
only way to add value, and thus the only way to make a profit, is to
operate where knowledge is the key to success - through the strategic
use of software.

The HBR article is rather long, so I have also included the (much
shorter) San Fransisco Chronicle review of the article by Don Clark
before the A.R./S.H. article.  First though, a bit about the authors,
quoting from the biographies provided by HBR:

``Andrew S. Rappaport and Shmuel Halevi are, respectively, president
and vice president of The Technology Research Group in Boston. The
firm, founded in 1984, advises semiconductor, computer, and software
companies in the United States and Europe on business strategy,
marketing, and product development.''

The article presents the authors' reasons for the current computer
industry shakeout.  The thesis presented is that there is a world glut
of really good manufacturing capability (90th percentile capability
they call it) and that trends indicate that there will be more of it
in the years ahead. To them, this means that manufacturing as a
process no longer provides decisive market advantage, in fact, it
provides an extreme disadvantage.

Certain companies have taken advantage of the excess manufacturing
capability and concentrated on using that excess capability to help
them provide value for their customers.  The concept is to provide
value by using other people's manufacturing capabilities, R&D
expenditures and such to leverage your own product. Their prototype
here is Microsoft.

Other companies on the other hand have tried to duplicate
manufacturing capabilities which already exist elsewhere and have
provided poor returns for their investors and held tenuous or
declining market positions.  Their prototype here is Apple (because of
the lead times required for articles in a journal such as HBR, the
submission was clearly written before the hoopla surrounding the
Apple/IBM/Motorola deal).

They point out that today's silicon manufacturing processes provide
more capability than can be consumed in present-day products.  Thus,
there is an excess of electronics (silicon) production capacity and
hence all the concentration on quality, speed, openness, compatibility
and other microdifferentiation factors which are the rage today.

In the author's view, the most successful companies today and tomorrow
will not be using top of the line manufacturing processes, rather they
will be using second-string production capacity from external (off
shore) manufacturing sites.  Their products though will be configured
in such a way to provide nonduplicatable (i.e. proprietary) value for
the customer.

This is not that novel a concept when considered historically: find
something people desperately need and which others can't make and
sell it for high prices.  That idea seems to have been lost in the
last few years of concentrating on open systems, standards, and all
that.  Such a situation should be a signal, obvious to somebody, that
if everybody is doing the same thing, the only differentiating factor
will be quality, or openness or compatibility.  These attributes
hardly fit the ``desperate need'' or ``others can't make'' criterion.

Rappaport and Halevi outline three major rules for competition in the
current and future era of excess manufacturing capability:

- ``Compete on utility, not on power''

They state that, a hardware strategy based on ``open systems'' is a
prescription for corporate suicide.  Companies that live by low entry
barriers also die by them.  Products must use current standards, but
must be configured to provide value in arenas governed by proprietary
products; leverage standards to attain higher heights, don't compete
on the execution of the standard alone.

- ``Monopolize the true sources of added value; create vigorous
competition for enabling components''

Their prototype for this rule is Sun's development and subsequent
licensing of the Sparc architecture.  The idea is to create something
of real value, such as the Sparc architecture, but ensure that there
is a true commodity market for the components which go into making
that product.  An anti-example is Apple's choice of manufacturing
their own proprietary hardware and software instead of outsourcing the
hardware and licensing out the software.  Unmentioned but fitting the
pattern is Jobs' NeXT Inc. which still makes their own hardware.

- ``Maximize the sophistication of the value you deliver; minimize the
sophistication of the technology you consume''

Their example here is Silicon Graphics which beat out Evans &
Sutherland in the high-performance graphics market using standard
semiconductor technologies coupled with standard software interfaces
and put in a workstation configuration.  This rule governs the authors
push for more software-level products which provide value on top of
standardized hardware.

In their words, the computer companies that prosper into the next
century will be those that focus on inventing new markets rather than
on building new products.  Manufacturing does not contribute to
margins, but instead is a ``reward'' for doing everything else
(customer support, distribution, integration) right.

Its an interesting thesis but one must remember that many of the
companies mentioned in in this article have since fallen on hard times
(remember, this was published in July 1991).  They have had trouble
specifically because of this ``computerless'' mode of operation.  For
example, Chips & Technologies announced its its first-ever loss in
Spring of 1992 precisely because it was forced to buy manufacturing
capability at premium rates for the previous quarter (the glut in
manufacturing went away for a few months and the producers charged
what the market would bear).

One company which has not fallen on hard times since the article was
written is Intel.  The authors are only able to dismiss Intel from
their theory with the comment ``Intel is the exception that proves 
the rule.''  So, the interesting question here is, has Intel prospered
_because_ of its vertical integration in the past five years, or
_in_spite_ of it?   The ongoing clone wars make it clear that history
has not given us the answer here just yet.

The central question then for the Free Software Business Model, or for
any software-based business model for that matter, is how to take
advantage of this trend.  If it is in fact a real trend; there are
some serious questions about that too as I've mentioned.

So, to my mind then the question of the Free Software Business Model
boils down to several rather deep questions:

1. Is the computerless computer company a viable business paradigm?

   Possibly and possibly not; an in-depth answer here would require
   an analysis of international trade issues, technology trends, and
   a bit of prognostication about the activities of Japan's
   semiconductor industry, MITI and our Commerce Department, Congress
   and the White House over the coming years.

2. Is it still possible to sell value that others can't provide while
   giving clear and explicit access to that product?

   The answer would seem to be yes so long as one chooses the market
   (the buyers) carefully.  Clearly certain consumers don't care
   whether they have access to the source code or not.  Others do.
   Market definition and motivation here would seem to be the central
   problem.

3. If one can ``sell'' software while giving it away then, what are
   the key choke points (if you will) in the current application
   domains where extremely sophisticated software skills are required
   to provide the value?

   Installation and development of corporate networks would be one
   area (enterprise integration).  Embedded systems development
   evironments would seem to be another area where the software is the
   value-add while the hardware is commodity.  Identifying further
   application areas where commodity technologies which are governed by
   published standards are in strong demand would seem to be the key here.




					W.


[ See asylum.sf.ca.us:pub/fsb/computerless.computer.companies for review ]