(Tom Carey Interview Continued)
MozillaQuest Magazine: Related to the previous question, it seems the biggest problem for both IBM and the GNU/Linux community might be the inclusion of JFS (Journaling File System), NUMA (Non-uniform Memory Access) software, RCU (Read, Copy, and Update), and other such packages into the Linux kernel. It seems, all things considered, that IBM did contribute these packages to the Linux kernel developers. It also appears that these are not part of Unix, per se, but rather are IBM extensions to Unix that might not be covered by SCO's Unix copyrights, to whatever extent SCO owns Unix copyrights.
MozillaQuest Magazine: SCO now owns the Unix license agreements under which IBM uses Unix and develops, distributes, and markets its AIX Unix variant. SCO's Amended Complaint alleges that under IBM's Unix license, JFS, NUMA software, RCU, and other such packages are derived works and therefore subject to the Unix license under which IBM uses Unix. Therefore to the extent that IBM did contribute these packages and technologies to the Linux kernel developers, IBM violated the License agreements and is subject to breach of contract and possibly to Unix license revocation.
MozillaQuest Magazine: On the other hand, JFS, NUMA software, RCU, and other such packages might not meet the definition of derived works under the U.S. Copyright Act (17 U.S.C. 101, et. seq. Please see Derivative Work in the sidebar.)
MozillaQuest Magazine: If JFS, NUMA software, RCU, and other such packages do not meet the definition of derivative works under the U.S. Copyright Act (17 U.S.C. 101, et. seq.), does that mean that SCO would not have any copyright infringement claims against Linux kernel and GNU/Linux operating system developers, Linux distribution providers, Linux users, and so forth?
Thomas C. Carey: I doubt that JFS, NUMA, RCU and such other packages are derivative works of UNIX. They are not a recast version of UNIX; they are programs that have use when linked to UNIX. They are more appropriately classified as separate works. The question is, who owns the copyright in those works? The answer is likely to be found in the contracts between IBM and AT&T, and IBM and SCO. [Please see Contract and End-User Liability Note in the sidebar]
--- MozillaQuest Magazine: Does SCO recently admitting that IBM owns the copyrights also serve as an implied admission that IBM did develop that code within the framework established by the contracts?
--- Thomas C. Carey: Admitting that IBM owns the copyright in JFS, etc. weakens SCO's copyright position, substantially. (Emphasis Added.)
- MozillaQuest Magazine: Is this as to the U.S. Copyright Act definition of derivative work or as to the Unix license definition of derivative work? Could these be two very different things that need to be distinguished carefully so the readers will understand what could be a very complex and important distinction?
- Thomas C. Carey: The IBM-AT&T license agreements do not define the term "derivative work". Consequently, the term almost certainly has the same meaning for purposes of the contract as it has for purposes of the Copyright Act.
-- MozillaQuest Magazine: So then:
(a) you do not believe that these extensions to Unix developed by IBM and Sequent, JFS, NUMA software, RCU, and so forth meet the Copyright Act definition of "derivative work" and
(b) therefore IBM's Unix license does not prevent IBM from contributing them to the GNU/Linux community?
-- Thomas C. Carey: Whether they constitute derivative works probably depends upon the manner in which they were created. If IBM adhered to the standards described earlier (i.e., the programmers did not refer to the source code or the manuals), then they probably do not constitute derivative works. If the applicable agreement is the 1985 agreement, then the 1985 Side Letter is pertinent. That Side Letter permits IBM to retain title to its modifications. And, as described earlier, it even permits reverse engineering, subject to some procedural limitations. Thus, if IBM was careful, the Side Letter should cover IBM's porting of its own work to Linux. It seems likely that there are separate agreements covering Project Monterey, agreements that have not yet seen the light of day. If so, they will be very important in determining who owns which fruits of that effort. [Emphasis Added.]
-- MozillaQuest Magazine: Doesn't the way the language in the Unix licenses was structured include an operational/contractual definition of "derivative work" embedded in the license(s) that in effect makes any Unix/AIX development part of the "Software Product"?
-- Thomas C. Carey: I do not believe that the IBM-AT&T or IBM SCO agreements define "derivative". If the 1985 agreement controls, this is just semantics. The way the 1985 Side Letter reads, if IBM coded an entire module without referring to UNIX source code or manuals, it belongs to IBM, whether it is derivative or not. Only AT&T coded modules belong to AT&T. If a module contains a mixture of AT&T and IBM code, each company owns the code it wrote.
Thomas C. Carey: For example, section 2.01 of the 1985 contract between IBM and AT&T states that IBM is free to modify and make derivative works of the product, but the modifications and derivative works are to be considered part of AT&T's "Software Product". However, section 2.01 was modified by virtue of the 1985 Side Letter (Exhibit C to the complaint), in which AT&T said that modifications and derivative works created by IBM would belong to IBM.
- MozillaQuest Magazine: Even so, does that relieve IBM of its Unix license duty to keep the code and methods for them confidential?
--Thomas C. Carey: With respect to the 1985 agreements, that is only true for AT&T code, not for code that IBM authored using proper procedures . The IBM code is not part of the "Software Product". The 1998 agreement between IBM and SCO forbids the preparation of derivative works or modifications, so that, to the extent the conduct of the parties is governed by that agreement, there is an express prohibition against preparing modifications, and the confidentiality provision is a secondary safeguard.
Please see the first two parts of our series about SCO-Caldera's IP claims plus its intentions to enforce and license its intellectual property rights.
Related MozillaQuest Articles
SCO-Caldera v IBM: