Anticipating the sunset

Unfortunately, Mr. Schwartz's time at Sun has been accompanied by a 50% decline in Sun's stock price. Whether he could possibly have done any better given the state of the company when he took over and state of the economy now is something one could debate, but we'll not do that here. More interesting, from the community's point of view, is the rumors that he could soon be looking for a new job.

It has often been said that if corporations were people, they would have the personality of a sociopathic teenager. Certainly companies can exhibit no end of the sort of moody, capricious, and even self-destructive behavior sometimes seen in adolescents - then they come back and ask for more money. An abrupt change at Sun could well bring in a CEO determined to show that his predecessor's policies were fundamentally wrong and were primarily responsible for Sun's problems. And that could bring some interesting changes.

Imagine a Sun which decided that it could no longer afford to share its Valuable Intellectual Property with the world. Perhaps Solaris, OpenOffice, Java, etc. would be relicensed under the new, Sun Proprietary Overtly Indecent License (SPOIL), with no more free releases. Hungry lawyers could start prowling for cases where Solaris code has been mixed into projects with incompatible licenses. StarOffice might go OOXML-only. MySQL could shift to a new, undocumented on-disk format with users' data subject to Sun-controlled DRM on every table. The new Java license would forbid the publication of not just benchmark results, but also of criticism of features of the language.

Clearly, some of these scenarios are rather far afield - though they are fun to make up. But, if we have learned anything from the SCO story, it must be that a company which presents itself as a solid part of the community can, in short order, turn around and go against us. Even if Sun does not degenerate to the point of starting legal attacks against free software, it could certainly put an end to the many contributions that it is making now.

Whenever one deals in company-owned free software, one should consider what happens if that company goes away. Projects with distributed copyright ownership are mostly immune to this kind of problem; there is no single company which could create huge problems for the Linux kernel by withdrawing its participation, for example. (Along these lines, it's worth noting that Evolution recently stopped requiring copyright assignments from its developers). But, in situations where a single company owns the copyrights and dominates development, a change of heart could make a real difference to downstream users. It all depends on what sort of community has developed around the code.

If future versions of Solaris were to be proprietary-only, the current releases would still be out there. But the Solaris development community outside of Sun is tiny, so chances are good that such a move would kill OpenSolaris as a free software project - to the extent that it is one now. Anybody wishing to continue to use Solaris would probably have to move to the proprietary version. OpenOffice.org would likely survive, though the external development community - never encouraged that much by Sun - would have to organize itself and, perhaps, choose a new name. Java is entirely subject to Sun's policies regarding conformance tests and such; it could easily revert to its status from a few years ago. And so on. The point is that a change of heart at Sun could easily make us appreciate the company's relatively friendly attitude now, and could create difficulties for distributors and users of Sun-sponsored projects.

There are plenty of other single-owner projects out there, of course. Many of them are entirely dependent on the continued good will (and viability) of their sponsoring companies. Others are less so. Copyrights on code released by the GNU project are generally owned by the Free Software Foundation. But, if Richard Stallman were to hit his head in an unfortunate contra dancing accident and decide that, henceforth, FSF-owned code would only be released under the binary-only GPLv4, those projects would not suffer much. Instead, the development community behind that code - strongly influenced but not controlled by the FSF - would quickly move to a new home and continue its work. For a practical example, see the creation of X.org in the wake of the relicensing of XFree86.

With any luck at all, the silly scenarios outlined above will not come to pass. But there is value in pondering how things could go. Such thought quickly leads to the conclusion that a vibrant development community is not just good because it leads to faster progress and more cool features. That community is the source for the long-term support for the code, support which is not subject to one company's quarterly results.

I don't see this article as very useful or constructive, honestly.  Regardless of the details
of conformance tests or whatever (and the MySQL example was just silly), the code is Free
Software now, and more than anything else, the code matters.  Talk is cheap.

We can worry about scenarios like this if they happen; but speculating about them beforehand
just creates negative energy.

I too am kind of down on the speculation of regime change on Sun, though I'm not sure I can
articulate why.  However, I think the point made is entirely valid.  Code availability
matters, but code community matters also.  

There are different levels of concern, of course.  For projects like Postgresql where
development began outside a commercial environment, and have continued into a variety of
sponsored environments, the lack of a large community of participation is not nearly as
important as the demonstrated ability of the project to adapt to sponsorship change.  Projects
like Solaris which have only ever lived inside the auspices of a single organization are of
more concern over view changes in the owning organization.

It's a meaningful point to consider when looking at risk-management of software use in the
open source world.

postgresql almost died when Greatbridge went under (A .com that hired most of the developers).
the core developers point out that it's not a coincidence that they all work for different
companies now.

they are enforcing the diversification themselves where a large project (say the kernel) gains
the diversification from the sheer number of people who work on it, but even there, for many
years Linus avoided working for any of the Linux companies, specifically to avoid the
appearance that any one company would have undo influence on his work

I've seen several other projects flounder when they core developer(s) have real-life things
cause them grief. the point that we need to watch out for this sort of thing is very true.

> Regardless of the details of conformance tests or whatever (and the MySQL example was just
silly), the code is Free Software now, and more than anything else, the code matters.

Sure, the code is paramount, but I think the article makes a good point about the community
around the code being almost equally important. If OpenOffice is 'taken away from us', then
sure, we can continue to develop it - assuming we can find among us motivated individuals. My
concern is that it will be far easier to find people eager to start yet another office suite
rather than continue work on OO.

The FOSS community is pretty dependent upon some FOSS-but-commercially-developed projects,
like OpenOffice and Qt. It makes sense to speculate about what might happen if that
development should cease or something close to that.

That said, I do feel some unease in speculating about such things, in that it might offend the
relevant corporate entity, here, Sun. Sun deserves our thanks, not our offenses, although
still, we must be prepared for the worst.

For what it's worth, I found the article thought provoking and the examples entertaining.  I
also enjoyed reading walters' comment though: negative energy directed at an article on the
grounds that it contains too much negative energy.  Brilliant use of irony!

There was one scenario you failed to mention Jonathan: That a new CEO more committed to Open
Source replace Mr Schwartz.  

Interesting take. I'd never have thought the free software community would have received
anything of value from the SCO fiasco, but I guess that a concrete bad example is useful.
Perhaps all the wariness over Bitkeeper was a result of SCO's actions, and the outcome
(Bitkeeper's withdrawal of free use) another concrete example for the points in this article.

Sun really is a large contributor to the free software community now, but it is still just one
CEO away from being a pain in the posterior. 

Perhaps, says the pessimistic user of free software, it would be wise to assume that change is
inevitable, and that keeping tabs on vulnerable projects would be a good idea. Not just those
from Sun, but those from IBM, Intel and other firms responsible to their stockholders.

Well, Linux was scoured for years by a small army of lawyers all salivating over the
possibility of finding "stolen" property... and none was found.  What other operating systems
can make that claim?

Note that the FSF copyright assignment contracts would not allow a binary only release.  If
they breached the contract in such a way, then they wouldn't actually own the copyright on the
code.

I wish more projects that required copyright assignment built similar provisions into their
copyright assignment contracts so you didn't have to put as much trust in them.

Wow, the comments in that link are interesting:

http://blogs.barrons.com/techtraderdaily/2008/07/09/sun-m...

There are a lot of comments by Sun employees, and a lot of comments by investors or potential
investors.  Many of the latter are strongly critical of the open source strategy (which they
generally refer to as the "giving things away strategy" or the "free strategy").

Oh yes, this is going to be interesting...

It would be nice if you could give some pointers (strings to search for) to interesting
comments from potential investors.

I read a few dozens comments from employees and got tired tired of it. Almost all of them are
basically just saying 1. "pony tail" (Jon Schwartz) must leave 2. the board is inept 3. the
whole management chain is rotten 4. this company is going belly up so fast it's going to get a
speeding fine.

I find it tiresome to read this much concentrated anger and bitterness.

I saw a lot of that from employees and ex-employees, too.  Also I noticed that many of them
praised their fellow "front-line" employees (by which I take it that they mostly mean
engineers) for being good and hard-working.

Anyway, here are some search strings for comments from the business/investment crowd:


"free"

(Only a few of the comments that contain the string "free" are not about Sun's Open
Source/Free Software strategy, and only a few of the comments that are about that strategy are
not from investor types excoriating that strategy.)

"open source"

:-)

I have read a sample of the comments on the blog, it seems to me that a small percentage
focuses on open source profitability, and 80% or more are on bad management procedures, lot of
internal bureacracy, CEO micro-managing (including 75 emails to discuss the best mug to
celebrate the MySQL deal) yes-man culture, not focusing on some technology but doing almost
anything, too many employees, and so on. 

Maybe these two different aspects are deeply connected: to be profitable with open source,
your business and development procedures must be of stellar quality (and, au contraire, to be
profitable with closed source could give you 1-2 years advantage over competitors). 

I agree with you, it will be interesting to see if they throw out the baby (open source) with
the bath water (bad management).

In terms of 'what might happen' it is worth remembering Zimbra and CUPS.

I may be wrong but I think both of these projects require the assignment of copyrights back to
a central entity.

Zimbra, which is a high profile project in a very important segment - i.e. competing with MS
Exchange, nearly became property of Microsoft when they attempted to buy Yahoo.

CUPS (which definitely required code assignment) was purchased by Apple (allegedly to prevent
it moving to GPL3). Apple could close-source the project any time they wanted.

Having seen these two events my feeling is that copyright assignment is to be avoided. If I
was selecting a package for my organisation I would consider copyright assignment to be a
negative feature.