The Report of the California Performance Review is now available on-line. It contains four volumes and over 1,200 recommendations on 279 different issues, all with a purpose "to reform and revitalize California's state government."

Intellectual property comes up twice, both times in the category "Statewide Operations," and both times in connection with software..

Recommendation SO 06 is "Create a Code Library to Leverage State Intellectual Property Rights." It notes that the software code written by state departments is valuable and should be protected through patent and copyright. Furthermore, the state should make money by licensing the code to others, and should keep a wary eye out to detect and pursue any infringements.

Then there is SO 10, "Explore Open Source Alternatives," which notes that "open source software can be a more cost-effective alternative" and recommends that "Departments should . . . implement open source alternatives where feasible."

In theory, the two recommendations can be reconciled because the term "where feasible" can caulk chasms of inconsistency. But in fact the fine print of SO 10 accepts as truths that open source has lower total cost of ownership than proprietary products, is more secure, and is less vulnerable to attack, which makes the conclusion about its "feasibility" pretty foregone.

Before buying off on it, the California decision makers need to probe more deeply.

To begin, "open source software" is not a meaningful category. There are dozens of different licenses, and they have significantly different consequences. For example, if a state agency adds code to an open source program licensed under the GPL, or adds GPL'ed code to a state-written program, then the state forfeits its IP rights to that code, in apparent contravention to SO 06. The BSD license, in contrast, is more forgiving if code is added to a proprietary program, but attaching it to a newly-written product still cedes any possibility of making money from the effort.

"Open source" is not a meaningful category in another sense: one must distinguish between the open-source operating system Linux and open source applications. IBM, HP, Sun, Dell, and other companies are pro-Linux without favoring open source for applications. Sun keeps hold of Java. IBM makes billions from selling its proprietary aps, and intends to keep it that way. It wants to turn the operating system into a commodity, but not the aps.

In essence, Linux has become a way for tech companies to cooperate on a standard Unix-based operating system and bundle it with hardware, not a new approach to productive enterprise. Thus its use raises standard choices between buying something -- in this case, computer services -- as a bundle (hardware/software/services) and buying it in disaggregated form. In some cases, the bundle, including Linux, may be the best solution, but many factors are in play, and they are all hotly disputed, including questions of total cost of ownership, security, and vulnerability, maintenance over time, guarantees against infringement of others' intellectual property, and compatibility with future innovation as hardware develops.

For more on these issues -- probably more than most people want to know -- consult PFF's paper on The Enigma of Open Source Software (March 2004), plus the material cited in an earlier WebJournal entry.