Subject: Re: Tim's paradigm shift
From: "Benjamin J. Tilly " <ben_tilly@operamail.com>
Date: Tue, 29 Jun 2004 12:36:22 +0800

Let's try this again, this time hitting "Reply-All", shall we?

Russell Nelson <nelson@crynwr.com>
 wrote:
> We should be talking about this, eh?

But why?  Isn't it about this open sores stuff, not free
software?

:-P

> From: Tim O'Reilly <tim@oreilly.com>

> Date: June 26, 2004 12:57:30 PM EDT
> To: Dave Farber <dave@farber.net>

> Subject: The open source paradigm shift
> 
> Dave, this may be of interest to your readers.
> 
> I finally got around to writing up the talk I've been
> giving for the past year or so, at locations as divergent
> as Red Hat and Microsoft, about the way that the
> commoditization of software is driving value up the stack
> to web apps, and how these apps leverage network effects
> (and in particular what I've been calling "the
> architecture of participation") as their key tool in
> gaining competitive advantage.   I also talk about how
> the "Intel Inside" positioning is being rediscovered in
> various software niches, and how commodity software means
> that assembly of custom Linux distributions (rather than
> added value) may be a key competency.

Sounds like good stuff.

> http://tim.oreilly.com/opensource/paradigmshift_0504.html


s/good/long/ # VERY long

OK, I read it.  Reactions.

First a pet peeve.  I _hate_ having people talk about paradigm
shifts while missing Kuhn's original point.  Tim is better than
most - he obviously understand's Kuhn - but he did nothing to
address the common misconceptions.  So, just to satisfy my pet
peeve, here's the synopsis:

  Any paradigm worth discarding lightly was not worth owning in
  the first place.  In fact the appearance of progress only comes
  in a field when some paradigm is so compelling that everyone
  agrees on it.  In the absence of a shared paradigm, effort is
  spent in trying to convince each other, and no observation or
  fact can be seen as progress because no shared context exists
  to frame it as such.

  Of course any paradigm is limited.  In retrospect, the periods
  that we identify with having the greatest growth and insight
  are those where an old consensus broke down and a new one was
  created.  Being early to understand the ideas that will become
  the new consensus points you in obvious directions that are so
  far unexplored, generally to your great benefit.

  This makes a paradigm shift look good.  But it is only good if
  the old paradigm really _has_ broken down, and is only good if
  what you think will become the new paradigm really _is_ going
  to do that.  For every person who finds a new paradigm in the
  birthing, hundreds more only think that they have.  Furthermore
  when a field is constantly undergoing "paradigm shifts",
the net
  effect is chaos.  No idea can be framed in a lasting context, no
  fact can be picked out as significant, and no agreement can be
  found to simplify advanced discussion.

OK, that is enough Kuhn to be able to safely ignore over 90% of the
references that you see to "paradigm shifts".  Maybe over
99%.  Just
look for bafflegab about how you should jump onto this coming
paradigm shift because paradigm shifts are great things.

Of course that doesn't apply to Tim.  Tim is not inclined to
shallow analysis.  And the last couple of decades he has consistently
demonstrated his ability to catch early trends before they are
apparent to other people.  Furthermore he is trying to document a
shift that is there, not one that he thinks people need to accept.

That said, perhaps I'm too close to some Tim's past discussions, but I
don't see this as a new paradigm.  It resembles closely what Tim has
been talking about for a while.  The dynamics that open source talks
about can happen without actually getting the guarantees of the OSD.
The trend is for monopoly to become commoditized, moving business
opportunities to complementary areas of business.  Often the process
creates those areas of business.  Currently we are seeing a process of
commoditization of the tools for building networked businesses.  The
open source prescription does not work for those because the useful
service requires data and networks that are not readily replicated.
Networked businesses that successfully create positive network
feedback achieve another form of lock-in.

This all sounds very familiar.

Now what do I wish that Tim had said that he didn't, or had said
differently?

Well I wish that he brought up _Information Rules_.  The discussions
there on standardization and network effects would provide a lot of
useful background for audience members who want to turn theory into
practice.  Furthermore, internet opportunities notwithstanding, every
business has cases where they face API lock-in as well, and it is
good to understand how those dynamics work.

The stuff about customizability was something that I guess I'm not
too used to hearing Tim talk about.  It didn't strike me like a
thunderbolt though - probably because I've pretty much ONLY worked
for companies that fit his description.  The software is the method
of delivering something that isn't software, and without maintainance
that something doesn't retain value.  In my last job it was up to
date models of bonds.  In my current job it is information on where to
find an apartment.  I guess that in the past a large fraction of the
applications which this applied to were in house applications that
were not sold, and therefore weren't visible business opportunities.
Now they are visible.

That's a shift - and certainly a shift in terms of what it means for
software to be free of proprietary lock-ins - but I still don't see
it as great news.

As for significant omissions, in any discussion of network effects
and the internet, I'm sure that Andrew Odlyzko is bound to have said
something worth pointing to.  Browse http://www.dtc.umn.edu/~odlyzko/
and find it. ;-)

Cheers,
Ben