Ex parte Stolowitz
Appeal 2009006157; Appl. No. 10/885,960; Tech. Center 2100
Decided: October 24, 2010
The application on appeal was directed to disk drive controllers. Two representative claims read:
1. A method of interfacing a RAID storage device controller to a PCI bus host without modifying existing host driver software, the method comprising the steps of:The Examiner rejected all the independent claims as obvious over a combination of two references. The rejection relied on FIGs. 1 and 2 of Levy for teaching the above-emphasized limitations. The Examiner characterized Levy as follows:
in the RAID controller, implementing a host interface for logical connection to the PCI bus host, the host interface emulating an ATA controller;
in the RAID controller, emulating an IDE storage device as though it were connected to the emulated ATA controller, thereby forming a first logical IDE storage device interface;
20. A RAID storage device controller comprising:
a host interface compliant with ATA interface specifications for interfacing with a PCI bus host as though an ATA controller were attached to the PCI host bus;
a drive interface comprising a plurality of drive port interfaces for attaching a plurality of disk storage devices to the RAID storage device controller;
Levy is teaching that the disclosed method and design provides for RAID functionality while emulating existing IDE/ATA functionality such that the existing software drivers and software applications can still be used, this is a description of a system and method for emulating existing hardware and software functionality.On appeal, the Applicant argued several limitations, including those emphasized above. With respect to claim 1, the Applicant argued that a PCI host controller is not the same as a RAID controller. The Applicant then addressed the emulating limitation as follows:
(Emphasis in original.)
Simply disclosing "an existing controller ... around which all drivers and applications are installed" (Col.4, lines 32-34), as relied on by the Examiner in Levy, does not even suggest any sort of emulating, let alone specifically teaching that "in the RAID controller ... a host interface for logical connection to the PCI host bus [is implemented], the host interface emulating an ATA controller, as claimed.The Applicant treated claim 20 in a similar manner, arguing that "disclosing an intermediate adapter connected to an existing controller, as in Levy, simply fails to even suggest 'interfacing as though an ATA controller were attached to the PCI host bus'." (Emphasis added.)
(Emphasis in original.)
In the Answer, the Examiner elaborated on the teachings of Levy with respect to emulation:
[C]learly Levey is teaching that the MIR-95 system is emulating a normal IDE ATA disk drive channel, while actually performing RAID level 1 mirroring with a plurality of ATA IDE disk drives. ... Col. 8 lines 29-34 [teaches] more specifically, "...if the disk drive adapter MIR-95 determines that the system is in a mirror mode, then for each WRITE command it will cause the same data to be written to all of the hard drives rather than to only one hard drive ...", in other words, the MIR-95 emulates the behavior of a single disk drive, but actually performs RAID level 1 disk mirroring and performs a WRITE to more than one drive.The Answer also included a definition of emulate from a technical dictionary: "to represent a system by a model that accepts the same inputs and produces the same outputs as the system represented." The Examiner then tied this to Levy's teachings as follows:
(Emphasis in original.)
The MIR-95 accepts the same inputs and produces that same outputs as the system represented, which is a single disk on an ATA IDE channel. However, the MIR-95 is in fact another system that simulates the functionality of a single ATA IDE channel, but actually provides the functionality of a RAID level 1 mirroring system.According to the Examiner, "this is description of emulation as it is defined by the IEEE."
(Emphasis in original.)
The Board reversed the rejection of claim 1, but affirmed the rejection of claim 20. With respect to claim 1, the Board found that the Examiner had mischaracterized the MIR-95 interface between the IDE controller and the physical drives:
This misses the point; Levy’s MIR-95 is not a host interface because it does not connect to the host computer. Levy’s host interface is an IDE controller, which generally is an ATA compatible controller. There is no emulation; there is identity instead.While claim 1 actually recited "emulating an ATA controller," claim 20 did not. Claim 20 instead recited "interfacing with a PCI bus host as though an ATA controller were attached to the PCI host bus." The Board framed the issue for claim 20 as: "whether, as a matter of inferential logic, doing something as though A were true precludes A from being true."
According to the Board, the answer is No:
The Appellants’arguments appear to assume that the claim requires that an ATA controller was not attached to the PCI host bus. The claim contains no such limitation. As a matter of inferential logic, doing something as though A were true does not preclude A from being true. Levy has an ATA compliant IDE controller attached to a bus which may be a PCI host bus. The ATA compliant controller in Levy interfaces identically as though attached to a PCI host bus.Because claim 20 did not exclude a configuration in which the ATA controller was in fact attached to the PCI bus, the Board affirmed the rejection.
My two cents: The Board got the rejection of claim 1 right. From my own cursory reading of the reference, it appears that Levy teaches emulation, just not the emulation that was claimed. That is, Levy teaches a redundant disk controller that emulates a single disk controller. And maybe it's obvious to go from that to what the Applicant claimed. But even if the Examiner understood Levy's limited teaching and was relying on obviousness to carry the rejection the rest of the way, he didn't explain his reasoning well enough to make a prima facie case.
The Applicant got lucky here with claim 1, because the arguments were not very persuasive. The Applicant's argument amounted to little more than a quote from each cited portion of the reference, followed by a corresponding assertion that the particular quote did not amount to a teaching of the claim element at issue. That strategy doesn't usually work. This is one of those unusual cases where the gap between the reference and the claims was so large on its face that a mere assertion was all it took to convince the Board. In other words, the Examiner read way too much into the reference.
As for claim 20, I interpret it differently than the Board did. I read the phrase "as though attached" to exclude attachment. That is, I read the claim as expressing a counterfactual conditional rather than an indicative conditional. (If you're into linguistics, Wikipedia has more about conditionals here.)
The Board's affirmation of the rejection of claim 20 should have been designated as a new ground, because it definitely relied on a new claim construction, one not seen by the Applicant before the Board's decision. I say this not because the Examiner didn't make his claim construction explicit, but because at no point during prosecution did the Examiner treat claim 20 any differently than claim 1. So the only way it's fair for the Board to interpret the two claims differently on appeal is to designate an affirmed rejection as a new ground.
Finally, an important lesson here is that sometimes small differences in claim language can have big results. Here, expressing the distinguishing limitation with exactly the same language in all claims – a very common tactic for patent prosecutors – would have resulted in either a complete win or a complete loss. Instead, some of the Applicant's claims made it through appeal, even though others did not.