The End of Open Source? (rev 1.2)

It seems some folks are building a cathedral smack-dab in the middle of the bazaar.

Two unrelated news stories in the past week caught my attention:

• Oracle acquired Innobase OY, makers of the InnoDB storage engine
• Tenable Security announced Nessus 3.0 will not be a GPL release, and source code will not be freely available

Renaud Deraison, the principle author of the Nessus security scanner, cited a particular form of freeloading — in which GPL-licensed code is embedded in turnkey hardware, thus exploiting a loophole in the GPL — as the main competitive threat that prompted the policy change.

This, to be sure, is an isolated event; unfortunate, but not of substantial consequence to the open-source community as as whole.

Oracle's acquisition of Innobase, however, is a bombshell — particularly to the LAMP subset of the open-source world.

While MySQL is designed to support any number of data storage engines (including MySQL's own MyISAM file-based storage system), in MySQL 4.x and the new 5.0 most of the critical new features are supported only by InnoDB — including most of the functionality crucial for enterprise-class applications, such as relational integrity, stored procedures, and transaction-processing capabilities.

Oracle's acquisition of Innobase effectively gives the database-software giant strategic control of the near-term future of MySQL. At the very least, it casts a long and dark shadow of doubt on the lifespan of applications built on MySQL 5.0 that require both the InnoDB feature set and unencumbered licensing terms à la GPL.

A strong technical argument can be made that InnoDB is not a crucial piece of the MySQL 5.0 architecture in the long run. Indeed, Oracle's move may prompt MySQL AB to devote the necessary resources to developing an in-house storage subsystem that provides similar features. A number of "on second thought" pieces such as this one from InfoWorld seem to be backing-off from the Chicken-Little tone of the first reports about the Oracle-Oy deal. Also, Oracle's move to have a say in MySQL suggests that Oracle leadership itself now sees MySQL as a genuine potential threat -- good news for MySQL, in a way.

But the real damage here isn't to the technical coherence of MySQL's product line. The true impact of Oracle's move will be seen in the thousands of IT managers who were just barely starting to warm up to open source, and who will now feel (at least for a time) that a decision to license Oracle or Microsoft SQL Server is less likely to put their jobs at risk. This is all the more regrettable as the open-source movement has been in a dire need of a credible database-management component if it is to go beyond the modest beachheads it has established in corporate adoption of open-source applications (PostgreSQL has had most of the crucial features for quite some time now, but not the requisite performance).

Taken together, the Nessus and MySQL events strongly suggest that the tragedy of the commons, regrettably but inevitably, has reached the shores of open source.

My interpretation of these news items, to be sure, is considerably tainted by my current reading: Lawrence Lessig's The Future of Ideas: The Fate of The Commons In a Connected World. Nearly every review of this book includes the word "sobering"; it is indeed. In it, Lessig thoroughly documents the trends undermining the "Innovation Commons" of the Internet.

While I am not quite as pessimistic as Lessig, I see an unmistakeable turning of the tide. But it is not yet another ascendancy of private property over public; nor is this another battle in the ongoing war between the Old and the New.

These recent events suggest that the underlying problem may be cultural. Tenable Security and MySQL AB are both for-profit businesses trying to leverage the perceived-value, ownership-cost and other benefits of the open-source model. Why are these business models so fragile? Is it just because they are new, and we've not accrued enough precedent to know exactly how to tweak the formula?

Or could it be that the open-source movement remains, at its core, a digital incarnation of a civics movement, steeped in volunteerism, community responsibility and the basic desire to do Good?

Stakeholders in open source face a choice that is, like all classic conundrums, easy to say and hard to do: either wholeheartedly abandon the "civic sentimentality" and other vestigial traces of the original do-gooder ethos behind the GPL, embracing the licensing model strictly for its tangible benefits; or fully accept the open-source codebase as a commons, the use of which requires a substantial exercise of social responsibity in the licensee's or beneficiary's business.

The fully-selfish route risks alienating the legions of do-gooders who, like it or not, are the ones creating and maintaining the open-source codebase. The latter course, which I have chosen for myself, means revising profit expectations drastically downward. But it is the only sustainable choice.

All of us invested in open source must be wary of killing the geese that lay golden eggs. It is inherent to the nature of common resources that their exploitation will never yield the kinds of profits and multiples of capital that can be achieved by exploiting controlled, fully-owned resources. But the alternative is just a dead goose.