Raise your hand if this has ever happened to you: 

The other side thinks your client so plainly — and evilly — copied software source code that their lawyers can't imagine losing. They have such clear proof, they believe, they almost sneer at you. And the smugness infects all that they do in the case.

If you put your arm up just now, you may see the train coming. Yes, the folks who have your client dead to rights and can't lose will in short order manage to destroy their copyright infringement case in epic fashion.

How, you ask? Simple: They forget to compare the software code they registered copyright in with the code they claim your client copied.

That very thing snatched defeat from the jaws of victory in Airframe Systems, Inc. v. L-3 Communications Corp., No. 10-2001 (1st Cir. Sept. 14, 2011). Airframe accused a software customer, L-3, of infringing Airframe's copyright in software that Airframe had at one time licensed to L-3. (The software seems to have helped high-flying L-3 track maintenance of its aircraft.) Airframe registered copyright in old versions of the source code (from the 1980s) but not the 2009 one. And yet Airframe's sole evidence of copying by L-3 compared not the versions Airframe had registered but one from many years later. The district court granted L-3's motion for summary judgment on the ground that Airframe had failed to raise a fact issue on the "copying" element of its infringement claim.

The First Circuit affirmed:

Here, the only evidence of copying Airframe presented was [Airframe president Gordon] Rosen’s declaration. Rosen made no direct comparison between the allegedly infringing M3 program [that L-3 used] and the ARMS source code versions covered by Airframe’s copyright registrations, as would normally be done. Rather, he compared the M3 program to the updated 2009 version of the ARMS source code. While this would support a finding of substantial similarity between the M3 program and Airframe’s “current” source code (i.e., the 2009 ARMS source code), there is no claim that the 2009 source code was itself registered or that the 2009 version is the same as one of Airframe’s earlier registered versions–the 1981 “IBM version,” 1984 “PC version,” or 1988 “UNIX version.” Rosen’s declaration said nothing about similarities between the 2009 ARMS version and Airframe’s earlier registered ARMS versions. Indeed, Airframe admits that they are in fact not the same, because the 2009 source code is a version “that had been updated by Airframe in the ordinary course of business.” Appellant’s Br. 15. Having presented no evidence sufficient to prove the content of its registered source code versions, Airframe cannot show that any of its registered works is substantially similar to the allegedly infringing M3 program, and Airframe has failed to create a genuine issue of material fact as to its claim of copyright infringement.

Id. at 10-11 (footnote omitted).