BPAI cases on silence in a reference

Takeaway: Case law from the BPAI suggests that when a POSITA could draw one of two different inferences from the implicit teachings of a reference, but not enough information is present to determine which one, then this is not enough for anticipation. However, case law also suggests that when the number of inferences a POSITA could draw from the implicit teachings is relatively small, the feature may still be obvious.

Details: There seems to be something about negative limitations that leads to a rejection even when the reference is silent. In Ex parte Mallis, the negative limitation at issue was "wherein is a torque is applied such that plastic deformation of the positive stop torque shoulder does not occur upon final makeup." The Applicant argued that the reference was silent about deformation, but did teach bending of the shoulder, and that a POSITA would understand this bending to include plastic deformation, which was prohibited by the claim. The Examiner conceded that the reference was silent as to plastic deformation, but argued that the Applicant's inference that the reference taught the opposite of the claim was wrong. The Examiner said a POSITA would not conclude that plastic deformation would occur. The Board characterized both assertions as speculative, and found no anticipation: "If Klementich lacks adequate disclosure to determine whether plastic deformation occurs, then it also lacks adequate disclosure to determine that plastic deformation is avoided."

Ex parte Bar-Sadeh involved a claim with a negative limitation: "A diaphragm comprising: a sheet of material formed on a substrate having a hole, the sheet of material covering the hole and including one or more corrugations that are substantially free of defects." The Applicant's position on appeal was simple: the references did not discuss defects, and so did not teach "free of defects." The Examiner's Answer agreed that the reference was silent about defects, then explained why features taught in the reference would "further contribute to corrugations having substantially no defects." The Board said the proper inference to draw from the reference's silence was no defects (as claimed), because the reference was an issued patent with a presumption of operability. Given this legal presumption, case law shifts the burden to the Applicant to show that the corrugations in the reference were defective. [Comment: I think the Board confused operable with defect-free. Surely the diaphragm in the reference could have defects and still operate.]

Several cases show that you can lose on obviousness even when the Board agrees that the silence in the reference would defeat anticipation. Ex parte Bett related to a method of controlling a manufacturing line, and the limitation at issue was storing a digitized image of workpieces in permanent memory. The Board agreed that the reference was silent as to the memory being permanent, but found the claim was obvious since "artisans would have known about permanent as well as volatile memories and the choice of one over the other is dictated by many well known factors."

The Board used similar reasoning in Ex parte Choo. The claim was very concise: "An Internet Key Exchange method, comprising performing an IDE Main Mode in hardware in parallel with system software loading." According to the Examiner, every computer has system software that runs at power up, and because the reference taught that key negotiation occurs at power up, that reference also implicitly taught that key negotiation was performed "in parallel with the system software loading" (as claimed). The Applicant argued that the reference was silent about the timing of the session key negotiation, and that the Examiner had "leapt to a conclusion that was unsupported by the reference." The Board agreed that the reference was silent as to whether the negotiation and the system software load occurred sequentially or concurrently. However, the Board found that it would be obvious to try both options.

In another case involving a negative limitation, the Board not only agreed the limitation wasn't taught, but also found the claim not obvious. The claim in Ex parte Gragg involved coordinating access to a storage device through an "uncoordinating communication protocol," negative because the Applicant argued this meant "protocols that do not provide notice of access to other protocols or devices." The Examiner's position was that the primary reference taught clients with different protocols, and that these clients could complete for access without notifying other clients. The Applicant's position is that different doesn't mean uncoordinating, and these protocols could be coordinating protocols.

As a preliminary matter, the Board found that the reference was silent about whether protocols coordinated or not. Notably, the Board specifically rejected the idea that the silence in the reference taught the limitation "by negative implication. The Board's obviousness analysis was interesting. The Board noted that if the issue was viewed as a choice between two protocols (coordinating and uncoordinating), the claim was obvious under KSR's "finite choices" rationale, assuming there was a design need or market pressure to solve the problem. But the Board framed the issue differently:

[T]here may be other types of protocols that could include attributes of both coordinating and uncoordinating protocols. As such, we cannot say that there would be a finite number of identified, predictable solutions, let alone that it would have been obvious to try them to arrive at the claimed invention.

Related posts: I first discussed this topic in a post about Ex parte Grath, a case in which I strongly disagreed with the inference that the Board drew from silence in a reference. I followed up on the topic in this post, in which I discussed two Federal Circuit cases that seem to be the most relevant to silence in a reference. Note that one of those cases (Upsher-Smith Labs., Inc. v. Pamlab) discusses multiple inferences. But in Upsher-Smith, the POSITA had enough information to infer that both options are taught, as opposed to not having enough info to know which inference to draw.