Subject: An OSS Experiment WAS: Sun, BSD, and GNU
From: Stephen Walli <>
Date: Mon, 31 May 1999 08:01:16 -0700

One of the things that is so hard in this space is experimentation.  To get
the data we would all love to have in some of these discussions, requires
analysis of what is working (to paraphrase earlier discussion, "how to OSS
programmers *eat*,") as well the ability to experiment.  Unfortunately,
experiments are hard to define and setup, and failure may cost one a
business employing people with mortgages to pay.

John Gilmore wrote:
jg> PS: Sun has about 3/4 the annual revenue of Microsoft -- roughly
jg> twelve billion dollars.  Sun could transition to GPL software, since
jg> GPL doesn't threaten most of Sun's revenue.  Microsoft could not; most
jg> of Microsoft's revenue is directly derived from duplicating software.
jg> This may be a good reason for Sun to move to GPL, since it would put
jg> pressure on Microsoft that Microsoft can't easily resist.  Note that
jg> Microsoft is actively buying into hardware companies (e.g. AT&T and
jg> cable TV companies), probably as a hedge against a completely
jg> software-based revenue stream.

To which Tim O'Reilly replied:
to> This is a cool idea.  I have never understood (e.g. with the
to> SCSL) why Sun cares that much about the kind of meager licensing
to> revenue they get from Java (and they hope, with Jini) when their
to> basic business is hardware.  But on the other hand, I haven't
to> analyzed their revenue, so I'm probably talking out of a cocked
to> hat.

It would be very interesting to see what would happen if a corporation like
IBM took a large software division in the PC world open source, like Lotus,
with it's huge suite of business applications.  IBM is ultimately a hardware
vendor with hardware revenues to fall back on.  IBM is also one of the
world's premier business service organizations (with service revenues to
fall back on), so building an open source "services and support" business
around such an OSS model is something they well understand.  It *is* IBM,
(and Lotus), two well known brands, so corporate IT consumers are not likely
to be "fearful" of buying.  This organization also already knows how to sell
into a corporate IT world, so the infrastructure is already in place to sell
to precisely the kind of corporate MIS management that is (i) going to
appreciate the economy of scale of buying fewer media/license kits versus
spending money on support and services, and (ii) have the pockets deep
enough to contemplate large support/service expenditures.  And it would take
the OSS model, well supported by a large recognized well-branded trusted
corporation up against an entrenched competitor to whom OSS is anathema, and
which probably can't mount the same level of service/support organization
any time soon. 

On the Lotus side, you have a well organized software development
organization already capable of supporting the product, evaluating any
proposed source modifications and integrating those that make sense in the
context of what's good for the Lotus product space.  There is also a well
organized product marketing organization that already supports a large user
base and partners base with information.  The publication of the information
of how the Lotus family of products works (the source) would allow a lot of
people the opportunity to create supporting product businesses. This
ultimately expands the IBM/Lotus marketplace one would expect.  (Assuming
conversion tools from Outlook-to-Notes exist in the Lotus world, I for one
would love to understand how to convert mail back OUT of the proprietary
Outlook format.) It is unlikely that anyone is going to compete against them
in a serious way by repackaging the Domino/R5/Notes world. (Who are you
going to buy Lotus Notes from to run your business of 20000 workstations --
IBM/Lotus or Walnut Creek?  A lot of small businesses will start with Walnut
Creek, but Lotus could quickly sell a lot of courseware to the small
business, and up sell the relationship over time.)

IBM/Lotus probably couldn't experiment with the GPL as the license --
rightly or not -- because the perception of the GPL would probably spook all
of the Lotus corp partners and third-party add-on developers, and any
successful business understands that one should never upset one's working
business relationships. (I am not requesting a flame war on which license is
best for society with this observation -- merely arguing it would likely be
a variation on the GPL-- not the GPL.)

Of course I just described, in a few paragraphs, an "experiment" that would
shake the stability of a lot of people's lives at a large corporation.  This
experiment would require a lot of thought and reassurance (not the least to
the sales force currently making money on licenses), and a deep
understanding of the Lotus revenue model, and how it fits to the IBM
corporate revenue model (which I certainly don't have).   It also takes the
OSS discussion out of the Linux vs NT, GPL vs BSD vs my license into the
realm of large corporate MIS revenues very quickly.