Ex parte Martinez
(Appeal 2009-002619, Appl. No. 10/781,477, Tech. Center 2100)
Decided: February 22, 2010
The independent claim at issue was directed to a method of reporting computer reboots, and the feature at issue was "generating at least one non-maskable interrupt signal." The Examiner took the position that the feature was taught by this sentence in the reference: "Test equipment is connected to the processor-to-bus bridge through a three pin serial port, the test equipment able to set a bit in the critical interrupt register for requesting the nonmaskable pseudo-interrupt." (Emphasis added.) In the Appeal Brief, the Appellant argued that the rejection was improper because the reference "unequivocally states that nonmaskable interrupts are not used." The Appellant then referred to a dictionary definition of "pseudo" to show that the meaning was "false or counterfeit; fake." Though the Appellant didn't make the argument explicit, I think the implicit argument was this: "a fake X isn't the same as an X."
How did the Examiner get around this "fake interrupt isn't the same as an interrupt" reasoning? In short, by showing that the pseudo NMI in the reference operated exactly the same as a POSITA understood an actual NMI to operate. Namely, that the pseudo NMI transferred control to an interrupt routine and could not be disabled (masked) by the processor's interrupt disable instruction.
The Board summarily adopted the Examiner's reasoning:
We agree with the Examiner that "[t]he fact that Murthy called these non-maskable interrupts non maskable pseudo interrupt[s] is irrelevant . . . because an interrupt when implemented is an interrupt whether it is called pseudo or something else." (Answer 16.) "In other words, a result of implementing interrupts or pseudo interrupts within a processor is the same. The result is to call an exception, and execute an exception routine . . . ."(Id.)I actually agree wholeheartedly with the Board's decision. [Disclaimer: I happen to know a lot about this particular area of technology, because I used to be a software developer and have a lot of experience with interrupts. It's possible that this colors my thinking here.]
The way I see it, "nonmaskable interrupt" could be understood in two ways. One, as a functional description. This is how the Examiner interpreted it: an NMI transfers control to an interrupt routine and cannot be disabled (masked) by the processor's interrupt disable instruction. The other interpretation is structural: an NMI is the signal that is connected to the processor's NMI pin.
If you choose to claim only function, then any implementation of that function will anticipate. To distinguish function, you must show that the claimed function operates differently than the reference's function. Here, the Appellant should prevail if he could show that the pseudo NMI did not transfer control to an interrupt routine, or that the pseudo NMI could be masked. But what doesn't matter is that the reference calls the function something else, or uses different words to describe it – if a POSITA would understand the reference to teach that function, it anticipates.
Here, it seems like the Appellant could have claimed some structure along with the function, something like "generating an non-maskable interrupt signal connected to the NMI input line of a processor." Because the reference explicitly taught that the pseudo NMI was connected to a different processor input, that claim would have distinguished over the reference.
I realize there may be good reasons to avoid claiming structure, perhaps reasons related to infringement. In my mind, that's the toughest part of claim drafting: coming up with claims that get over the art and are still infringed.