Columbia University v. Symantec Corp (Fed. Cir. 2015)
Back in 2013, the NPE known as Columbia University sued Symantec for infringing at least six of its data-analytics patents covering the process of detecting and blocking computer malware.[1] Following a 2014 claim construction it was clear that all of the asserted claims were either (1) not infringed or (2) invalid as indefinite. As a result, Columbia stipulated to such and filed its appeal of the claim construction order.[2] On appeal, the Federal Circuit has affirmed the indefiniteness holding and substantially affirmed the claim construction, but did find enough problems with the lower-court decision to re-ignite the case.
Here, I’m just going to focus on the claim term “Byte Sequence Feature.”
Defining Byte Sequence Feature: The claim construction issues in this case are parallel to those debated in many cases – when a term is understood by one skilled in the art, does the specification limit that definition by using the term in particular ways (but not expressly re-defining the term), or does the term retain its ordinary and customary meaning? Here, the court reiterated its en banc statements from Phillips that “the specification may define claim terms by implication.”
Columbia argued that the term “byte sequence feature” includes more than merely machine code instructions, and also extends to other non-compiled elements of an executable attachment. On appeal, the Federal Circuit rejected that argument — finding that the specification indicates that the byte sequence feature “represents the machine code in an executable.” The court also found that the machine-code limitation “most naturally aligns” with the inventor’s description of his invention.[3]
Since none of the accused products analyze machine code instructions, they don’t infringe (under this construction).
Loose Statements in the Provisional: Of interest to patent prosecutors, the associated provisional application included a statement essentially saying that the byte sequence feature does not include non-machine code “resource information.” That statement was removed from the non-provisional application, but the Federal Circuit indicated that the original definition still stands.[4] This fits within the normal approach to patent prosecution that the mere withdrawal of a mis-statement is insufficient. Rather, following a mis-statement or change-in-definition, the patent applicant should affirmatively identify the change for the examiner’s consideration.
Resource Code is Indefinite: I noted above that both courts agreed – the proper definition of “byte sequence feature” covers machine code and not ancillary resource information that may be part of the executable file. Now, the confusing part is that many of these same claims include a express limitation that the “byte sequence feature” includes this resource information — exactly opposite of the court’s definition of the term.
It is a problematic ambiguity to have a claim element include embodiments excluded from the element’s definition. As such, the court determined that those claims must be found invalid as failing the definiteness requirement of 35 U.S.C. 112. The court writes:
Claims 1 and 16 conflate a “byte sequence feature,” which is a feature extracted from machine code instructions, with the extraction of “resource information,” which is not a machine code instruction. Specifically, the claims describe the step of extracting machine code instructions from something that does not have machine code instructions. . . . The claims are nonsensical in the way a claim to extracting orange juice from apples would be, and are thus indefinite.
The court’s decision on indefiniteness appears correct but avoids the patentee’s actual argument that the claim terms should be construed so as to avoid this nonsense approach.
As noted in the intro, the Federal Circuit did reverse the claim construction as to the claims of two of the patents and vacated the non-infringement judgment. Thus, the patent case will continue on remand. In the case, Columbia also sued Symantec for fraudulent concealment, unjust enrichment, and conversion. Those state law claims are apparently still pending before the district court.
= = = = =
[1] The asserted patents include Patent No. 7,487,544 (“the ’544 patent”), U.S. Patent No. 7,979,907 (“the ’907 patent”), U.S. Patent No. 7,448,084 (“the ’084 patent”), U.S. Patent No. 7,913,306 (“the ’306 patent”), U.S. Patent No. 8,074,115 (“the ’115 patent”), and U.S. Patent No. 8,601,322 (“the ’322 patent”).
[2] You might question whether the claims are valid under 35 U.S.C. § 101 for being directed to an unpatentable abstract idea. Claim 1 of the ‘544 patent is directed to a “method for classifying” an executable file attached to an email by (1) extracting a byte sequence from the attachment “representative of resources referenced by the … attachment”; and (2) using a set of classification rules (not defined by the claim) to predict whether the byte sequence is malicious. I suspect that the full-throated Section 101 argument was not raised because the defendant Symantec also supports broad subject matter eligibility. In that situation, it may be the Court’s jurisdictional role to take up the mantle.
[3] See Renishaw PLC v. Marposs Societa’ per Azioni, 158 F.3d 1243, 1250 (Fed. Cir. 2003).
[4] See Advanced Display Sys., Inc. v. Kent State Univ., 212 F.3d 1272, 1282 (Fed. Cir. 2000) (provisional applications incorporated by reference are “effectively part of the” specification as though it was “explicitly contained therein.”). This suggests that the oft-used approach of ‘fixing it in the non-provisional’ may need to be reconsidered.
[5] In the ‘115 and ‘322 patents.