Subject: Re: Releasing under OS, what License?
From: DJ Delorie <dj@delorie.com>
Date: Mon, 25 Jun 2001 01:31:20 -0400


> Yes, they do.  The group of copyright holders has the advantage that
> they can release improvements under proprietary terms.

See my other reply.  Plus, if they plan on doing this, they cannot
take advantage of what others build upon their GPL'd work.  That helps
the "others".  In the case of Cygwin, for example, there are add-ons
out there that RH can't touch, because they're pure GPL'd and RH wants
to be able to provide a non-GPL cygwin to customers.

> The LGPL makes this more nearly symmetric.

Why?

> From now on "Red Hat is the biggest employer of glibc developers" is
> an "accident" of history that allows them to maintain their position
> in the future at lower cost than would-be entrants need to pay to
> catch up.

That's not an accident of history, that's the status quo.  If those
developers leave RH and start a new company, the fact that RH was once
the leader is of no help - they still can't support glibc as well as
the new company.  RH has an advantage because RH has the people, not
because RH had the people.

The "accident of history" with RH (and Cygnus, I guess) is that they
were the first to market with a sustainable business plan (which is
the case for any successful business, I suppose).  That positioned
them to be the market leader long enough to grow to the point that
competing *directly* with them is no longer a viable option for most
startups who don't have the bankroll to afford that.  So, the startups
"compete" by building on RH's technology.

> "Fair or not," as you say, is a matter of view point.

Yup.

> Note that mere size confers an advantage under GPL.

Yup, but, IMHO, that's "fair".  A large company is paying people to
provide a service to the user, and based on the value of that service
they fail or succeed.  If a new company can provide better service,
they'll succeed instead (perhaps).

> Fending off entrants is a similarly straightforward exercise,
> requiring only the ability to merge sources, for the large
> incumbent.  Is it really good for society (and the industry and
> especially the movement) to encourage bigness for its own sake?
> (Real question.  Obviously, I don't think so.  But I don't know.)

I think in this type of business, bigness has certain advantages, when
the customer is itself big.  I don't think a little company can expect
to get a support contract with a big company without a *lot* of
reputation to back it up with.  Plus until open source is ubiquitous,
getting these large customers benefits us all, because it makes our
potential customer base larger.

Plus, my experience at Cygnus leads me to believe that a big company
won't always crush a little competitor, because a big company may not
be able to profit off small customers (too much overhead).

And lastly, the ability to merge is only half the problem.  The other
half is getting all the merged in bits to work, which in my experience
is *not* trivial.  Ironing out the problems helps the "little guys"
too.