"According to some of those who have had a look at the offending code, it DID come from IBM after all.  There are reportedly many lines of identical code, and at least some of the Linux code even carries an IBM copyright notice.  Well, this is a surprise to me and a delight at SCO headquarters in Utah, I'm sure, but I'll bet my house that SCO does not prevail and here's why....a case of a single person who did two very similar implementations based on his earlier research.  Both his UNIX and Linux versions (works B and C) were derived from his original research (work A) which was not exclusively limited to UNIX.  His paper shows that was the case and while SCO may see it as the smoking gun, IBM will see it as the proof of innocence.

"What SCO owns (forgetting for the moment Novell's contrary ownership claim and perhaps AT&T's) is the copyright on this particular work as applied to UNIX.  But Linux is not UNIX, so applying the same ideas -- even the same code if it comes originally from an upstream source -- is not necessarily copyright infringement. 

"Say I write a new high-level programming language, then do nearly identical implementations of that language for UNIX and Linux and the UNIX version is made part of some official UNIX distribution.  Does that mean the Linux version violates the UNIX copyright?  No.  But I wrote both versions and the code is identical.  Surely that is a copyright violation?  No.  This isn't a matter of clean rooms and virgins and reverse engineering, it is a matter of precedence and authorship.  Sequent (now IBM) did not give up all its rights to the code when it was made part of UNIX.  They were very careful to plan it that way."

There is one part of this story that has me puzzled, although I'm happy he thinks SCO will lose. The puzzle is, how can it be an offense for IBM to donate code to Linux that IBM has a copyright on Maybe I am missing something. On the other hand, Cringley reports a conversation with DiDio in th same article.