Re: GPL vs. sweatshops (was Re: Bias)

On Tue, 11 Jun 2002, Roland wrote:

> Yes, but the question is, if the improvements are good enough. I think most 
> of free software developers would still prefer to use GCC. Even most 
> proprietary softwares would probably still use GCC. This is because better 
> compilers often only generate faster code, and speed is not that much 
> important for most softwares.

This is the key, I think.  In general, the free, public tree -- if vital 
and active enough -- is likely to be the choice of preference in general.

> On the other hand, companies improving the GCC would not return their 
> strategic improvements to the GCC public tree. So who would benefit from 
> it? Only 2 groups:
> 
> 1. The company.
> 2. Clients who are able to pay for the improved compiler.

In the short term, that's probably true.  Of course, if those clients 
weren't willing to pay, the company wouldn't risk the development cost, so 
they couldn't benefit.

> On the other hand the GPL demands all improvements to be returned for the 
> benefit of all. In that sense the GPL is more democratic because EVERYONE 
> benefits and not just a small priviledged minority. I think the GPL is a 
> good way to democratize technologie.

Democratic?  Isn't GPL philosophy more Socialist or Communist in nature?

NOBODY benefits from the proprietary company's efforts if they refuse to 
use the GPL codebase due to its demands.  That's the catch -- were the 
company to use the GPL code, they'd have to return all their improvements, 
but because of that very effect, they're bound to avoid it like the plague.

The other thing to consider is that the value of proprietary enhancements 
tends to diminish over time, while the maintenance hassle of maintaining a 
separate tree remains.  This may tend to encourage proprietary companies to 
eventually contribute most code back to BSD projects sooner or later, if 
it's not of strategic significance.  Perhaps this effect may amoreliate the 
apparent one-way sharing of the BSD license, in many cases.

As long as the public BSD project remains active, it's advantageous to 
merge back as much code as possible into the public tree to reduce 
maintenance costs, keeping only the strategic bits proprietary.  Still, 
they don't have to, so some will likely take advantage...

I can see value in the GPL or other strong copyleft as a bargaining chip, 
if the core developers of a public project want a revenue stream from the 
proprietary interests wanting to leverage that code.  This is the business 
model adopted by Sleepycat, and it's been VERY successful for them, and
very advantageous to proprietary vendors AND the free software world at the 
same time.  This is one of the few proven open-source models, but it works 
best (perhaps only?) with code that needs to be built upon to be useful, 
such as Berkeley DB or other library-type code.

In fact, way back in February, when this list was first created, I was the 
one who suggested that Wine consider following the Sleepcat model, since it 
had been so successful before.  Everyone decided against pursuing the idea, 
so I gave it up.  I thought it would have given the Wine project a chance 
to draw significant funding, but so be it.

Deven