Subject: Re: Tim's paradigm shift
From: Ben Tilly <>
Date: Sat, 10 Jul 2004 10:22:22 -0700

On Fri, 9 Jul 2004 17:42:33 -0700, Tim O'Reilly <> wrote:
> On Jul 8, 2004, at 7:23 PM, Brian Behlendorf wrote:
> > The right to fork is, as I see it, the fundamental right inherent in
> > any truly open source license.  It's the right to clone a creation
> > *out* of the hands of its creator and continue its evolution *despite*
> > where the original author(s) intend to take it.  It's how nearly every
> > major Open Source project starts, it's what keeps the investment risk
> > around collaboration low, and it's what keeps leaders of projects
> > humble.  IBM doesn't have to trust Linus with the kernel - it knows it
> > could fork at any time should their priorities and those of Linus
> > differ.
> ... discussion of CDDB and FreeDB omitted.
> But briefly, in response, my sense is that FreeDB hasn't really taken
> off, and the reason is that the cost of duplicating CDDB is greater
> than the "tax" they impose on the system as a single source.  And that
> will happen in many areas.  But note that "the right to fork" means
> claiming the right to the database, and that's a very different
> challenge than the right to source code, with different legal regimes
> coming into play.

It is so different a regime that I feel that pushing for such a right is
counterproductive.  As soon as you start talking about databases
and an internet business you get to user information.  Most of your
users don't want that data to be freely forked because they think
(probably rightly) that their personal information will wind up in the
hands of a spammer.

Furthermore even if you did establish the right to fork at this level,
you'd just move the boundary.  For instance in many valuable
services there is a web of legal contracts that you need to provide
that service.  Shall we fork those as well?

> I agree that this is a real risk.  But I would posit that "single
> source" at particular companies isn't the problem.  If standards are
> open, then there is always the possibility of competition.  And even in
> the absence of standards, it may be relatively easy to do -- note how
> quickly Kazaa rose up after Napster was shut down.  The obstacle raised
> by a big collaborative database is operational rather than intrinsic,
> and P2P can address some of the operational issues more easily than a
> centralized db.  But those obstacles can still be significant.  The
> question I have to ask, though, from a business point of view, is
> whether this is really BAD.

I disagree that the obstacle raised by collaborative data is
entirely operational.  In many services, the service involves
interaction.  Even if someone else has better technology and
equivalent data, the network effects still provide a centralizing

A poorly known but amusing example of this is Instinet.  They
are an online trading service that started back in the days of
the PDP-10.  In fact the platform was written in PDP-10
assembly.  And (at least as of 2000) never got rewritten.
When their PDP-10s finally were breaking down and could not
be fixed or replaced in the late 90's they were replaced by
*emulated* PDP-10s running on Linux!  After you negotiate
the Internet-Decnet bridge, to the emulated slow machines,
make a trade, and then go back, a trade takes about 1s to
execute.  Their costs are far higher than anyone else, as are
their prices.  But because of popularity, you always can make
a trade there so they continue to dominate.  (That might be
changing now.)

Their competitors have better technology, lower costs, and
equivalent data.  Mindshare and customers are not small
obstacles to overcome though.

> I continue to feel that the intensity of the feeling around the right
> to fork and the danger of lock in has been exacerbated beyond the
> natural level by the actions of a single monopolist.  In most cases,
> companies are less able to extract such a severe tax from their users,
> and it's often more onerous to build a competing solution than it is to
> accept the tax.  This isn't all bad, because it results in people
> figuring out how to cooperate rather than trying to compete on every
> front.

I disagree with the claim that the intensity has gone beyond
a natural level.

My view is that people and companies balance out opposing
incentives fairly well.  Given lock-in, the company with lock-in
will naturally gravitate towards pusing customers to balance
the entire value of said lock-in against some combination of
poor products, poor service and high prices.  After all, until
you push that limit, your customers can't provide you any
real incentives not to let things drift in those directions, and
there are always internal incentives to behave those ways.

Therefore the degree of eventual abuse and the degree of
lock-in are strongly correlated.  This prediction is, incidentally,
the natural outcome of the advice given in _Information
Rules_ for how to run a business which has achieved lock-in.

Microsoft today, and IBM before it, abused the IT world more
than others mainly because they could.  Anyone else who
gets into the same position likely will do the same because
they'll have the same incentives.

> At the end of the day, the open source world doesn't look that
> different from the commercial software world in the overall balance of
> competition between projects.  "Forking" happens in the commercial
> world as well; the forker just doesn't have quite as big a leg up.  But
> take Geronimo, for instance, vs JBoss.  I'm not sure that in practice
> that was all that different from the situation when the various NCSA
> people went off to work for Netscape, Microsoft or Spyglass, and
> everyone was recreating the same wheel for a while.

This I agree with - and disagree with at the same time.  I
agree that the overall dynamics look similar.  However I also
maintain that the stresses that lead up to "forking" in a
proprietary world are much greater than in the open source
world.  So the similarity misses a significant experiential

However I also believe that there will also be a mix between
things that can be provided in an open manner and ones that
require proprietary incentives to provide.  As long as the
proprietary actors do not manage to capture and manipulate
governments into giving themselves an unfair advantage, I
think that this dynamic mix is a good thing.