Case No. 2012-1507 (Prost, Moore, Reyna)
Numbers matter in defining an invention, not only in simply counting how many of a particular limitation are present, but also in defining relationships among distinct populations of claim elements (e.g., one-to-one, one-to-many, all-to-all, etc.). Apple v. Samsung II returns to the Federal Circuit with some pretty important lessons in careful claim drafting, including cautions in using generic “plurality” language for interrelated element groups in a claim.
The appeal arose from the district court’s preliminary injunction against Samsung’s sales of the Galaxy Nexus and involved issues that are quite distinct from the design patent issues (“Apple has a patent on rounded corners!”) appealed earlier this year in Apple v. Samsung I. Much of the preliminary injunction discussion turned on the irreparable harm factor. This post does not address those issues any further, but they are definitely worth a read and analysis for litigators. Instead, this post looks at the Federal Circuit’s likelihood of success analysis and construction of claim 6 of Apple’s patent 8,086,604, directed to multi-source intelligent search, embraced by the maligned Siri application. The claim looked like this:
6. An apparatus for locating information in a network, comprising:
an interface module configured to receive an inputted information descriptor from a user-input device;
a plurality of heuristic modules configured to search for information that corresponds to the received information descriptor, wherein:
each heuristic module corresponds to a respective area of search and employs a different, predetermined heuristic algorithm corresponding to said respective area, and
the search areas include storage media accessible by the apparatus; and
a display module configured to display one or more candidate items of information located by the plurality of heuristic modules on a display device.
The district court interpreted the above-bolded plurality of heuristic modules to require two modules (or more) that each used a different algorithm. Other redundant or non-unique-searching modules could be present while meeting the claim. Android’s Quick Search Box, which uses user input to query multiple different sources: Internet; browsing history; and contact list, with appropriately unique searching methods for each source, infringed this construction. Quick Search Box includes other modules that search same sources in the same ways, but the district court treated these as mere additional elements that would not defeat infringement under the “comprising” transition.
On appeal, the Federal Circuit disagreed, finding that “each heuristic module” in the claim required every heuristic module to use a unique algorithm; no redundant module could be present and still infringe the claim. This was because “each” preceded the generic “heuristic module” and not the earlier “plurality of heuristic modules.” The court’s reasoning looked to an analogy from ResQNet.com, Inc. v. Lansa:
The word “each” appears not before “plurality of modules,” but inside the “wherein” clause and before the phrase “heuristic modules.” The district court drew support for its construction from ResQNet. . . . But ResQNet in fact counsels the opposite conclusion. That case involved two different claims, one of which recited “each field,” the other one “each of a plurality of fields.” ResQNet, 346 F.3d at 1377. We thought “[t]his difference is significant” and thus construed the two claims separately, holding that the first claim meant “all fields,” the latter “at least two, but not all.” Id. at 1382. Here, the district court eliminated the very distinction that we deemed material in ResQNet by plucking “each” from where it appears and planting it before the phrase “plurality of modules.”
The “comprising” transition would not save infringement, because the “plurality” was not indicated or recited as a subset of other heuristic modules that would open the claim up to inclusion of additional, non-unique modules. This language was clear enough to override weak broadening in the specification, which disclosed that multiple modules were associated with heuristic algorithms, while only at times stating that each module used a different heuristic algorithm.
The panel also found the prosecution history suggestive of all present modules having their own algorithm. In amending the claim to add the “plurality of heuristic modules . . .” and wherein clause at issue, the prosecutor distinguished over a piece of prior art in the following remark:
[the prior art] teaches that the processor can use the solution to a constraint satisfaction algorithm to formulate a search request and employ any appropriate combination of local and remote search operations. [The prior art] does not describe, however, that each of the local and remote search operations employs a different heuristic algorithm to search an associated relevant area of search for information that corresponds to the search request, in accordance with [claim 6]. That is, the algorithms described in [the prior art] go to the formation of the search request and not to how the local and remote search operations employed by the processor perform a search of the repositories on the network.
This strongly suggested to the court that the prosecutor was distinguishing the search actions of the prior art as lacking 100% unique algorithms for each search operation/module. If the prosecutor was merely distinguishing search input constraints from the actual search process, the second sentence’s detail was unnecessary. Thus, the district court’s construction of claim 6 was erroneous, and the Android’s Quick Search Box on the Galaxy Nexus, including redundant search modules, was unlikely to infringe claim 6.
Preliminary injunction reversed and remanded.
Lessons: On a very simple level, if trying to modify only members of a claimed plurality and leave a claim open to further unmodified elements, avoid generic language like “wherein each of the X has . . .,” as seen in the ‘604 patent from Apple II. Instead use language that ties the “each” descriptor to the plurality, like “wherein each of the plurality of X has . . .” Also, for software (or any process invention, really), I think it’s always a good idea to be more explicit in the specification that additional and distinct steps, algorithms, and calculations can be inserted into any part of, or used in combination with, example embodiments. Sometimes in algorithms it is easy for a description to seem closed. Describing a summation of several different gathered values to calculate, say, a quality indicator rather implicitly requires that you can’t interject a “multiply by zero” action to that summation. Thus, give some consideration to truly how exclusive a disclosed process is, and try to very clearly state where additional actions can be added based on different circumstances. It may have helped the patentee in Apple II. Lastly, although the prosecution history did not seem dispositive here, be careful when stating the magic “the applied art lacks X” or similar language that X is chosen as narrowly as possible and used consistently throughout the argument. It’s usually ok to discuss the context of X in the art, claim interpretation, and other claim elements, just keep the actual distinguishing / faulting of the art clear and limited to the single chosen narrow claim aspect.
On a more subtle level, consider outright avoiding use of generic “pluralities” of things in claims with fairly complex relationships among different plural things. Sometimes it’s very important to consider and claim, for purposes of avoiding the prior art, exactly how two different element populations are related or interact and whether that relationship must be exclusive or permits additional combinations. It can be much better in this instance to recite “a first X and a second X” and “a first Y and a second Y” and recite the individual relationship between individual group members, like “wherein the first X connects only to the first Y and vice versa,” and “wherein the second X connects to at least the second Y” or whatever the relationship may be. This still gives you a plurality, except with a more clearly claimed relationship, and without using “each” or “every” or “two of” or any other language that tries to reach back specifically into a generic plurality and risks being interpreted to require “all of” or otherwise closing the claim to further group members, as happened in Apple II.
Comment: While I think the opinion (and district court) were right in finding that the “each” language fairly required “all of” a plurality of modules to each use a distinct algorithm, the closing of the claim to forbid additional, non-unique-algorithm-using modules outside that plurality seemed harsh. The court’s approach on this does not feel grounded in the plain language of the claim or appear mandated by ResQNet; it looks more like amending the claim to use a “consisting of” transition or to recite “only a single plurality of modules,” or limiting the claim to a single aspect of the disclosed embodiment. As I read ResQNet, the exclusving/closing “each field” in that case was never introduced as “a plurality of fields” in the claim, like here. Also, I would quibble with the court’s use of prosecution disclaimer in this instance, because the statement quoted in the opinion was open to multiple interpretations, but 1) it looked like the prosecution disclaimer was just secondary and supportive to the court’s plain language reasoning, and 2) Apple repeatedly distinguished other pieces of art during prosecution using harsher language about the every-module-to-unique-algorithm requirement.