Re: Proposal for LedgerSMB 1.5: Refactor and move some utility functionality to CPAN

On Fri, May 24, 2013 at 8:49 AM, Chris Bennett <..hidden..> wrote:

I always prefer BSD or even more permissive licensing.
If someone can then sell a proprietary version, they might be willing to
submit a module with the same kind of licensing. No guarantee of that,
of course. But why not?

I think there are larger issues actually than whether someone can release a proprietary version. Permissive licensing largely means that we do not assert control over the work of others, and we trust them to make decent and ethical decisions in the contexts in which they find themselves, and that there is established a commons which everyone may access. Those who release proprietary versions get to show the community what features may be needed, and very typically they will either be under pressure to give back what they have done or go their own way, often in later competition with the community. If we assert that software freedom is economically valuable, then free software will win out all things being equal. I assume that those who build their businesses on free/open source software see the economic benefits of this approach and prefer not to pay people for the privilege of restricting what we can do with our software....

There is something else that does happen. Companies close. People
retire. A contract for their software may end. Sometimes people die.
This actually happened to me during one project.
All of their proprietary software may suddenly become available with a
BSD license.

I suspect proprietary software is unlikely to go this route. The closest example I can think of is OpenIndiana (a fork of OpenSolaris). When companies go under, they usually like to try to sell interests in things to other companies. However in the end, it doesn't really matter. FreeBSD is still today much more prevalent than OpenIndiana is, and it has stronger development behind it. PostgreSQL is becoming the default RDBMS people associate with open source. Programs that spend years developing their commons typically outcompete everyone else eventually.

BSD licenses work well because they create common ground between companies, each of which may have a proprietary version. Everyone has an incentive to expand the commons in every direction other than their niche, so for mature programs today, I have trouble seeing them as viable unless they are open source or niche markets. Even what starts out as a niche market can become a full market for a BSD-licensed project.

Consider Green Plum which started off as a good corporate member of the PostgreSQL community but no longer contributes anything back. They offer, as a proprietary product, basically a massively multi-parallel version of PostgreSQL. However because they don't contribute back they are stuck on an older version of the code, and miss out on new features. Additionally, now Nippon Telephone and Telegraph, along with some others, have put together a BSD-licensed massively multiparallel version of PostgreSQL called Postgres-XC. Green Plum is now stuck working on an old version of the code and in competition with open source solutions... not a good place to be. Green plum may never be open source, but Postgres-XC is a viable, open alternative.

I don't really have a problem with proprietary solutions on the margins. If a market is viable, the community will move in, and there may be some things that open source business models can't yet reach.

Best Wishes,

Chris Travers