Subject: Re: Successful FSBs
From: "Tim O'Reilly" <>
Date: Sat, 26 Oct 2002 11:37:40 -0700

On 10/26/02 2:12 AM, "Stephen J. Turnbull" <> wrote:

> There is a sort of Laffer curve for definitions.  If they are too
> strict, the applicable cases are too few, and only academic
> specialists are interested.  I'm certainly biased in that direction; I
> both try to correct for it and admit it explicitly on a regular
> basis.  But if they are too weak, then everything qualifies and S/N
> goes to pot as everybody jabbers about what interests them in terms
> they have more or less private definitions for.
> For example, Tim says he wants to include Amazon, which clearly has
> unnecessarily chosen the proprietary road on occasion (Amazon wins on
> title coverage, economies of mass purchase, and stock management
> including fast delivery; "one click" ordering just isn't that big a
> deal).

FWIW, I didn't say that Amazon was *now* an FSB.  I *did* say that any
definition of FSB that ipso facto excluded exploration of google, yahoo!,
amazon, and host of ISPs as FSBs was prejudicing the very question that the
thread was purporting to ask: what can would be entrepreneurs learn from
"successful FSBs."

The tragedy, I've argued, is that amazon, google, et al don't consider
themselves FSBs and act accordingly.  And for me, "accordingly" is
explicitly and judiciously engaging with open source projects they depend
on, NOT making everything they do open.  And of course, it's an even greater
tragedy that folks thinking about what makes a "successful free software
business" are arguing about who's pure enough to be considered rather than
trying to understand what really works and how.

It is my belief that the most successful business models for FSBs come from
the use of the software to deliver software services, not from its resale,
or from the resale of supporting goods.    The ISP market is the clearest
example of this model, in which companies charge a monthly fee for access to
services based on free software (Bind, Sendmail or equivalent, Apache, and
various elements of the TCP/IP stack).  But by extension the concept applies
to companies leveraging free software further up the service stack.  There's
a huge amount to be learned from studying such companies, including the
kinds of margins you can get from services built on commodity software (vs.
from proprietary software that isn't shared by competitors.)

I believe firmly that the economic message of free software is to develop
business models that assume relatively fungible commodity software, and to
get your money from services based on that software.  The challenge is to
understand whether (or when) you get your marketplace advantage from an
additional layer of proprietary software, and when you get it from things
external to the software itself.  (By analogy, in the hardware market, Dell
gets its advantage not from unique proprietary extensions to commodity
hardware, but from economies of scale, superior logistics, and the lower
cost structure of its direct sale business model.)

Even worse than when the companies that ought to be seen as part of the free
and open source ecosystem are disregarded is when the natural business model
of a piece of free software is signed over wholesale to a hostile monopoly.
I'm thinking of Verisign/Network Solutions.  Running domain name
registration services *is* the natural business model for BIND, but no one
in the FS business realized it until too late.  If FS advocates had been
thinking more about software as service, we would have avoided the whole
ICANN fiasco, and had a competitive domain name registration system based on
commodity free software.

Of course, it's also true that domain name registration wouldn't be as
profitable for a host of players as it is for a single monopoly player (even
now with regulated competitors nipping at the edges), but that's beside the

I keep hammering on this point because I think that software as service *is*
the future.  We're moving into a world in which software that runs on a
single isolated platform is going to be the exception rather than the rule.
Almost all software will have an online service component.  Microsoft
completely gets it.  What was Passport but an attempt to create identity
services?  MyServices had a whole host of forward-looking services.  But
Microsoft is having trouble getting traction, because they think like a
monopolist, and the market has seen that movie before, and doesn't like the
ending.  (But Microsoft will be back, hopefully with a more open version,
but it will likely still not be as open as many of us would like.)

To my mind, anyone who wants to think about "successful FSBs" should be
thinking about the range of services that will be part of the future
"internet operating system" and what businesses could be built on those
services.  Go study .Net and the MyServices vision, and ask yourself how
many of those components are mirroring things that are already out there as
free software.  While the free software pundits are arguing over who is an
angel allowed to dance on the head of this pin, we are fortunate that there
are individual developers building software that does look forward to the

For one example, take a look at a proprietary company like groove, which is
building various kinds of groupware services--not just software, but the
services that make that software work, such as management and
synchronization of an "XML cloud"--and ask yourselves whether similar
services could be built, say, on jabber, or even Apple's lightweight
rendezvous/zeroconf framework.

Or take a look at Ping Identity Services ( and other
folks trying to build digital identity alternatives to Passport.  They
aren't "free software" per se, but they are fellow-travelers who ought to be
engaged in the discussion.
> I agree with him that Amazon wins if the best platforms are all free,
> but this is a matter of pissing in everybody else's soup.  The point
> is to make sure that nobody steals a march on you by introducing a
> much better platform than yours, not that Amazon has a core competence
> in use of platforms, much less development.  But Amazon is basically
> an ASP, and they want to keep the software concerning their core
> competences private.  To the extent that that integrates with OSS
> they've pulled in, they are _not_ going to contribute back to the
> community, and in fact they probably don't want their developers to
> participate in the community.

That's actually not true.  The developers there are very open to working
with outsiders.  I've been working that angle for some time, and making
connections for them.  But I won't argue your basic point.
> Like Tim himself, Amazon is in the business of relying on IP---it
> distributes it.  Unlike Tim (AFAIK), proprietary IP is also a core
> competence (I would assume, and their IP strategy indicates that they
> think so).  This is an FSB?  I think not.

I agree that Amazon is in the business of relying on proprietary IP, but so
is Collab.Net.  I'm sure that neither qualifies as an FSB by the narrow
definitions espoused here.  But there should be a definition that
understands more nuance than has been shown in this discussion.  For
example, even a company like Red Hat, which most people here would (I think)
qualify as an FSB, relies on proprietary IP.  They get a significant
fraction of their revenue from training.  Are their training materials all
under the GPL, and circulating on the net, able to be used by competing "Red
Hat" trainers?  I think not.  In the end, I think you'll find very few
"pure" FSBs.

I would suggest a set of categories, perhaps something like this:

Company relies on free software for the majority of its revenue and espouses
free software ideology and thinks of itself as part of the F/OSS community.

Examples:  Red Hat, LinuxCare, VA Linux (the latter two included for
historical reasons)

Company relies on free software for the majority of its revenue but
disregards free software ideology.

Examples:  Uunet and other ISPs

Company relies on free software for the majority of its revenue but is
actually hostile to free software ideology.

Examples:  Verisign/Network Solutions

Company relies on free software and additional proprietary software or other
IP, and thinks of itself as part of the F/OSS community even though not all
its software is free.

Examples:  Collab.Net, Sleepycat, Aladdin, O'Reilly

Company relies heavily on free software and additional proprietary software,
but disregards free software ideology and doesn't think of itself as part of
the F/OSS community.

Examples:  Google, Amazon, TiVo

Company leverages free software strategically because it benefits from
commodity software, and proactively engages with the F/OSS community even
though it gets most of its revenue from other sources.

Examples:  IBM, HP, (Apple), (Sun)

Company leverages free software strategically, tries to learn lessons from
the free/open source software model, but depends on proprietary software or
other proprietary IP for the heart of its business model.

Examples:  Microsoft, (Sun)

Company uses proprietary software model to deliver services, but might be
able to leverage open source model to improve its business.

Example:  AOL (Mapquest and AIM (not to mention the core AOL service) are
both Microsoft targets, and I predict that both will fall because of AOL's
go-it-alone strategy, whereas if they were to embrace at least open
standards, and perhaps F/OSS, they could stay ahead of the game.)

By studying the strategies, successes, and failures of companies in each of
these categories, it might be possible to develop some useful advice for
would-be free software entrepreneurs.  And more importantly from my point of
view, we might get more allies and supporters in the fight to keep the next
generation of computing open.

That's my real agenda.  I'm convinced that free and open source software
have built an amazing open computing platform (i.e. the internet) but that
many free and open source advocates haven't thought through where that
platform goes next, and the consequences of failing to get people who are
building components of that platform to think about their indebtedness to
the free/open source software ecology.

Tim O'Reilly @ O'Reilly & Associates, Inc.
1005 Gravenstein Highway North, Sebastopol, CA 95472