Opinion
CIVIL ACTION NO. 09-cv-11813-DPW.
March 31, 2011
MEMORANDUM AND ORDER
The Plaintiffs Red Bend Ltd. and Red Bend Software Inc. (collectively, "Red Bend") bring this action against Google Inc. for infringement of Red Bend's U.S. Patent No. 6,546,552, titled "Difference Extraction Between Two Versions of Data-Tables Containing Intra-References" ("`552 Patent"). It is alleged that Google's Courgette software infringes claims 8-12, 21-25, 28-34, 42-46, 55-60, and 62-67 of the `552 Patent (the "Asserted Claims"). (Def. Opening Constr. Br. ("Def. Opening") at 21; Pl. Opening Constr. Br. ("Pl. Opening") at 5.) Red Bend has moved for a preliminary injunction. Following briefing and oral argument on claims construction, this matter is ripe for resolution.
Red Bend initially asserted, in its preliminary injunction briefing, that Google had infringed only Claims 8-10, 21-23, 42-44, and 55-57. (Pl. Opening at 6.) Red Bend has since asserted infringement of the additional claims listed above.
I. FACTUAL BACKGROUND
I draw the factual background from the preliminary injunction record, which includes the parties' respective memoranda and the exhibits attached thereto.
A. The `552 Patent
Red Bend is a privately held software company with approximately 100 employees. The `552 Patent, issued on April 8, 2003, was invented by Sharon Peleg and assigned to Red Bend Ltd. (`552 Patent.) Red Bend's patented software updating technology is used both in Internet-delivered update environments and in firmware over the air ("FOTA") implementations. Its FOTA solution has been deployed on more than 500 million mobile devices and adopted by eight of the top ten handset manufacturers and dozens of other leading companies in the mobile, machine-to-machine ("M2M"), and WiMax markets.Red Bend's `552 Patent claims priority from an application filed on August 19, 1998, leading to issuance on April 8, 2003, and "relates generally to updating computer programs." (`552 Patent at 1:8-9.) When the `552 Patent was filed, the prior art techniques for updating software involved three steps, as described in the background of the invention.
First, the software provider "generate[s] a difference result representative of the difference between the old program and new program." ( Id. at 1:26-33.) The "difference result" is also called a "diff" or "patch." Second, the provider sends the resulting file through the Internet to a remote client site. Third, the client "invokes appropriate utility, which incorporates the differences in the old program, thereby generating the desired new program at the client site." ( Id.) An "obvious advantage[]" of this technique is that "only the difference result and not the entire new program is sent to the client." ( Id. at 1:36-37.)
The inventor of the `552 Patent noted that use of the prior art diff techniques "normally results in a relatively large amount of data, even if the modifications that were introduced to the old program (in order to generate the new program) are very few." ( Id. at 1:56-59.) For example, if an old program was updated by inserting only one or two new instructions and deleting only one or two other instructions, the difference result would reveal the inserted and deleted instructions, as well as "reference entries" which "specify a target address (reference) as an integral part of the command." ( Id. at 1:59-2:2.) The inventor therefore recognized "a need in the art to provide for an efficient tool which will result in significantly smaller volumes of difference results between old programs and new programs. . . ." ( Id. at 2:17-29.)
Google aptly analogizes the inclusion of reference entries in the difference result to the redline of a document: "If an author adds (or deletes) a single section heading to a document, all subsequent section headings are renumbered . . . [and] any references to those renumbered section headings in the text also change even though only a single section heading has been added or deleted."
The `552 Patent aims to reduce the size of the diff by generating a "modified" old program and a "modified" new program, in which internal references of corresponding entries in the old and new programs are replaced with "invariant entries." The diff is then performed on the modified old and new programs to generate a diff result that excludes the invariant references. At least two of Red Bend's products — vCurrent and vRapid — practice the technique described in the `552 Patent.
B. Courgette
In 2008, Google released an open-source web browser, called Chrome, as a free download and also made the code available in open source. To create updates for the Chrome browser, Google used an open-source software program called "bsdiff." On July 15, 2009, Google released a new open-source compression algorithm, called Courgette, to generate smaller updates for the Chrome browser on computers running Microsoft Windows. The corresponding Courgette source code was published on the Internet and is governed by an open-source license. Courgette cannot be used to update mobile devices or other computer systems that do not use the Intel x86 instruction set and the Windows portable executable file format.
According to Google, both Courgette and the invention described in the `552 Patent create diff or patch files to update software, but do so in different ways. Courgette uses a disassembly process to take apart the old and new executable programs into symbolic data structures that no longer resemble executable formats. The symbolic data structures separate references, i.e., addresses associated with instructions, that Courgette can recognize from those it cannot; Courgette does not recognize many types of instructions that contain reference addresses. Courgette also separates the disassembled programs into eight encoded streams that, like the symbolic data structures, are symbolic and not executable. Courgette preserves all unique references it can recognize in address streams, while the `552 Patent replaces unique references with invariant references. The unrecognizable references, instructions, and data are preserved by Courgette in a byte stream. From the eight streams corresponding to the old and updated programs, Courgette generates a patch that is a compressed comparison of those streams, and, unlike the `552 Patent, reflects all unique references found in the streams.
C. Procedural History
Red Bend became aware of Courgette on or about July 15, 2009, the date of Google's announcement, and immediately investigated Google's potential infringement of the `552 Patent. On September 3, 2009, Red Bend sent Google a cease and desist letter, requesting that Google "immediately cease and desist from using and inducing others' use of, such infringing algorithms, including by ceasing publication of the Courgette delta-update source code. . . ." On September 16, Red Bend and Google began discussions, geared toward arranging a meeting to discuss Red Bend's allegations, and agreed to a truce period lasting until October 23. During that period, Red Bend asked — and Google refused — to remove the source code from Google's website and to post a notice on its Chrome blog stating that users should contact Red Bend for a license before using the Courgette code.
On October 26, 2009, Red Bend sued Google for patent infringement. On November 5, the parties met to attempt to resolve their dispute to no avail. On November 17, Red Bend moved for a preliminary injunction, requesting an order requiring Google (1) to immediately cease implementation of Google Courgette updates or any colorable imitation that likely infringes the `552 Patent; (2) to immediately cease publication on the internet of the Courgette source code and accompanying instructions or any colorable imitations thereof that likely infringe the `552 Patent; and (3) to publish an announcement that Google's previously published Courgette code is not "open" and free for all to use, but instead is alleged to be protected by the `552 Patent.
I held a hearing on April 14, 2010, regarding Red Bend's motion for a preliminary injunction. The motion was taken under advisement and the parties were instructed to file a joint scheduling statement pursuant to Local Rule 16.6. On May 10, 2010, I adopted their joint statement in part, and ordered the parties to submit claims-construction briefs. A Markman hearing was held on August 18, 2010. This Memorandum and Order addresses and resolves both the preliminary injunction motion and claim construction of the disputed terms in the `552 Patent.
A Claim Chart concisely setting forth the disputed and agreed upon claim term constructions, the parties' respective proposed constructions and the construction I give the terms is attached on Appendix A to this Memorandum and Order.
II. CLAIM CONSTRUCTION
Claim construction of the disputed terms is necessary to determine Red Bend's likelihood of success on the merits for purposes of awarding a preliminary injunction, as well as for future determinations in this litigation. The parties touched on the issue of claim construction in their preliminary injunction submissions and, in their Markman briefs, refined their respective positions on which terms are disputed and their proposed claim constructions.A. Claim Construction Principles
B. Disputed Claim Terms
Markman v. Westview Instruments, Inc.52 F.3d 967970-71 i.e. Phillips v. AWH Corp. 415 F.3d 1303 1312-13 Id. Id. Vitronics Corp. v. Conceptronic, Inc.90 F.3d 15761582Id. Id. 1. "Compact Difference Result"
A "compact difference result" is claimed in Independent Claims 8, 21, 42, and 55, as well as in each of the asserted dependent claims, of the `552 Patent. Claims 8 and 21, for example, describe "a compact difference result between an old executable program and a new executable program." Red Bend requests that "compact difference result" be construed as "a difference result of a smaller size as compared to a conventional difference result (obtained by using techniques in existence prior to the invention of the patent-in-suit) in which the need to reflect changes to references due to delete/insert modifications is reduced or eliminated." Google, for its part, defines this claim term as "a difference result in which references that have changed due to delete/insert modifications do not appear." Alternatively, Google proposes a secondary construction of "a difference result in which substantially each reference that has changed due to delete/insert modifications does not appear."
Claims 42 and 55 are identical to Claims 8 and 12, respectively, except that every reference to an "executable program" or "program" is replaced by the term "data table."
The parties agree that a "difference result" is "data representative of the difference between an old program (or data table) and a new program (or data table) used to carry out an update of the old program." (Joint Claim Construction at 5.) Their disagreement centers on how the invention renders the difference result "compact."
It is clear from the patent that the "compact difference result" is smaller when "compared to conventional, larger in size, difference results." (`552 Patent at 14:10-14.) The Background of the Invention describes "a need in the art to provide for an efficient tool which will result in significantly smaller volumes of difference results between old programs and new programs, as compared to hitherto known techniques for accomplishing difference result." ( Id. at 2:18-20 (emphasis added).) The compact result is generated by a method whose "net effect," as described in the specification, is that "the invariant reference entries (between the modified old program and the modified new program), will not appear in the difference result, thereby reducing its size. . . . ." ( Id. at 3:40-46.)
Similarly, a preferred embodiment explains that a modification in an entry caused by an insertion of entries "will be modified to invariant references with the obvious consequence that they are not reflected in the difference result, thereby keeping the latter relatively compact." ( Id. at 9:66-10:15.) Thus, the patent is explicit that the difference result becomes compact by eliminating the invariant reference entries from the difference result. This aspect is wholly ignored by Red Bend's proposed construction.
The question then becomes whether every invariant reference entry is eliminated from the difference result. The language in Independent Claims 8, 21, 42, and 55 expressly answer that question, stating that " substantially each reference in an entry . . . that is different . . . due to delete/insert modifications . . . are [ sic] reflected as invariant references. . . ." ( Id. at 16:31-36, 18:9-15, 20:6-11, 21:51-56 (emphasis added).) That the invention "reduced or eliminated" the need to include "all the changed references when computing a difference" is also an explicit basis for distinguishing the invention over prior art techniques in the prosecution history. (Doc. No. 97, Ex. 2 ("Prosec. Hist.") at RedBend150 (emphasis added).) Google's proposed construction, which eliminates all references that have changed due to delete/insert modifications, is too narrow in light of the above-quoted claim language and the prosecution history. To account for this claim limitation, however, Google proposes, as an alternative construction for "compact difference result," the definition of "a difference result in which substantially each reference that has changed due to delete/insert modifications does not appear." This alternative construction properly "gives meaning to all the terms of the claim" and, therefore, "is preferred over one that does not do so." Merck Co. v. Teva Pharm. USA, Inc., 395 F.3d 1364, 1372 (Fed. Cir. 2005).
Accordingly, I construe "compact difference result" as "a difference result in which substantially each reference that has changed due to delete/insert modifications does not appear."
2. "Executable Program"
Independent Claims 8, 12, 21, and 25 relate to an "executable program." Red Bend defines this claim term as "a program comprising machine language instructions and corresponding bytes of data used by the program that are ready to be run on a computer." Google defines this term as "a program comprising machine language instructions and corresponding bytes of data used by the program that are ready to be run on a computer, excluding source or other symbolic code." An alternative construction that evolved during the Markman hearing, and to which Google did not object, was: "a program comprising machine language instructions and corresponding bytes of data used by the program that are ready to be run on a computer, excluding source or other symbolic code to the degree such source or symbolic code is not necessary to the execution of the program." (Tr., Aug. 18, 2010 (" Markman Tr.") at 47-50.) The parties therefore agree on the construction of this claim term, except for Google's proposed limitation excluding source or other symbolic code to some degree.
The `552 Patent's glossary provides that "an example [of] a data table can be an executable program either as a loaded program in machine-memory or as an executable-file . . . [in which] entries are individual machine instructions of the program or the individual data elements used by the program." (`552 Patent at 2:61-65.) There is no dispute that this definition governs the construction of "executable program."
The specification does not explicitly state, however, whether the construction provided in the glossary excludes source or other symbolic code. I must therefore turn to other intrinsic evidence to guide my construction analysis. The prosecution history proves helpful in this inquiry. The Examiner initially rejected the `552 Patent's claims as anticipated by the prior art Okuzumi. In order to distinguish the two, the applicants focused on a key difference regarding the input used in the Okuzumi and the `552 Patent's methods of generating a difference result and updating the program. Specifically, the applicants emphasized that the Okuzumi patent "explicitly mention[s] `source'" as the input into the process, whereas the `552 invention operates on executable programs, which it defined on the basis that "[i]n extracting diff between 2 version of executable files as defined in amended Claim 1, there is no source involved, and neither statements, nor any textual or other symbolic representation of the program even exist [ sic]." ( Id. at RedBend151 (emphasis added).) Moreover, in the prosecution history, the applicants differentiated their invention from prior software-updating methods, including Okuzumi, that worked on source code: "A major problem arises when applying these methods [used on source code] to executable program files aiming to extract the differences at the byte-level, regarding the files as a list of data bytes rather then [ sic] list of text lines." ( Id. at RedBend150.) The applicant continued, "Okuzumi et al. explicitly mention `source', and even more, this prior art reference contains a step of sorting `statements' (a common name for an element of a source program) according to their `character strings.' In contrast, the present invention . . . defines an executable program." ( Id. at RedBend151 (emphasis added).)
Red Bend argues that this language in the prosecution history refers only to Claim 1, which is not an asserted claim, and is therefore irrelevant. However, the prosecution history clearly indicates at the beginning of the same submission to the Examiner that "the following arguments focus on Claim 1, and will later apply to the other Claims, mutatis mutandis." (Prosec. Hist. at RedBend149.) Moreover, the treatment of a term in unasserted claims is instructive of the intended meaning of the same term in asserted claims. See Phillips, 415 F.3d at 1314.
To incorporate that limitation, the word "executable" was inserted into Claims 8, 12, 21, and 25 to limit the claim scope to "executable programs" instead of just "programs." ( Id. at RedBend161-63.) Red Bend confirmed at the Markman hearing that the `552 Patent does not use source or symbolic code as input: "You start with executables and turn it into symbolic code. That's the invention." ( Markman Tr. at 36-37.) The prosecution history, therefore, strongly suggests that the `552 Patent did not envision source or symbolic code as part of an "executable program."
The intent to distinguish the `552 Patent by excluding source or symbolic code is also evident in Red Bend's submissions to the U.S. Patent and Trademark Office ("PTO") during the Reexamination process. In response to the PTO's initial rejection of many of its claims as anticipated by Wetmore, U.S. Patent No. 5,790,860, Red Bend differentiated the Wetmore patent from the `552 Patent by noting that Wetmore operates on object files containing symbolic or source code rather than on the executable, numeric, and linked references that the `552 Patent covers. Red Bend cautioned, however, that "[t]his is not meant to suggest that the claimed executable program must exclude all symbolic information. To the contrary, as is well-known in the art, executable files often contain header and other symbolic information to assist the loader and the operating system with loading the program image into machine memory and commencing execution." Together, the prosecution history and the reexamination submissions suggest that the `552 Patent intended to exclude source and other symbolic code that was necessary to the execution of the program, but not source or symbolic code that provides peripheral information not necessary to the execution itself.
Consequently, after considering the parties' submissions and engaging in an extended discussion of the term at the Markman hearing, I construe "involved" from the phrase "no source involved" to mean "that is necessary for the execution of the program." ( See Markman Tr. at 51:3-12.) I consequently understand "executable program," in the context of this patent, to exclude "source" or other symbolic code that is necessary to the execution of the program.
Red Bend admits that the prosecution history demonstrates "that the applicant intended to limit the `executable program' claims to exclude techniques that accepted source code files as initial inputs." (Reply at 7 (emphasis added).) Its expert, Dr. Stephen Edwards, similarly maintains that the prosecution history reveals that the "applicant[s were] only trying to convince the examiner that his invention did not start from source code, not that it avoided all forms of symbolic representation throughout its operation." ( Id. (emphasis in original).) Red Bend's attempt to avoid Google's construction on the basis that it is solely the inputs that cannot be source code is unsupported by the prosecution history. Most compelling is the applicant's insertion of the modifier "executable" into the claims to differentiate the `552 Patent from prior art which used source code to execute the programs. (Prosec. Hist. at RedBend161-63.)
Consequently, I construe "executable program" to mean "a program comprising machine language instructions and corresponding bytes of data used by the program that are ready to be run on a computer, excluding source or other symbolic code that is necessary to the execution of the program."
3. "Data Table"
The term "data table" appears in Independent Claims 42, 46, 55, and 59. Red Bend construes "data table" as "a table of entries, including references, where an entry is an addressable unit within the data table. Each entry may have a different size. An executable program is one example of a data table." (emphasis added). Google advances the construction of "a table of entries, each of which may have a different size. An executable program is one example of a data table. The data table cannot be source or other symbolic code necessary for the execution of the program." It is evident that the parties agree that a data table is a table of entries, that each entry may have a different size, and that an executable program is one example of a data table. They dispute whether a data table can be source or other symbolic code.
The italicized text in the construction above was added by Red Bend at the Markman hearing. ( Markman Tr. at 67:2-21.) At the hearing, Red Bend also proposed an alternate construction: "[I]f the Court were inclined to adopt something narrower . . . add[ing] . . . from Google's construction, `[the data table] cannot be source or other symbolic code that lacks references.'" ( Id. at 67:14-18, 73:14-16.)
Prior to the hearing, Red Bend had construed this claim as "a table of entries, where an entry is an addressable unit within the data table. Each entry may have a different size. An executable program is one example of a data table."
This construction was proposed by Google at the Markman hearing to be consistent with its amended construction of "executable program." ( Markman Tr. at 70-71.) In its prior submissions, Google had defined this term as "a table of entries, each of which may have a different size. An executable program is one example of a data table. It cannot be source or other symbolic code."
As the parties indicate, the `552 Patent's glossary defines "data table" as "a table of entries, each may have a different size," and provides, "[a]s an example, [that] a data table can be an executable program." (`552 Patent at 2:33-34, 2:61.) "Entry" is defined in the patent as "an addressable unit [in a data table] that contains data." ( Id. at 2:35-36.)
Google, relying entirely on the prosecution history of the `552 Patent, urges that a data table cannot include source or other symbolic code because Red Bend disavowed a broader claim scope to obtain issuance of the patent. To distinguish the `552 Patent from prior art, specifically the Okuzumi patent, the applicant asserted that "[i]n extracting diff between 2 versions of executable files as defined in amended Claim 1, there is no source involved, and neither statements, nor any textual or other symbolic representation of the program even exist [ sic]." (Prosec. Hist. at RedBend151.) This limitation was also applied to the data table claims: "Claims 35 to 68 are basically similar to claims 1 to 34, respectively, except for the fact that they recite data table instead of executable program. Data table . . . do[es] not embrace source code as in Okuzumi." ( Id. at RedBend154.) On this basis, the applicants submitted that those claims were not anticipated by Okuzumi. Google now seeks to apply that aspect of the prosecution history to the construction of "data table."
Red Bend argues that Google's proposed construction is "far too limiting" because such a data table could not contain any source or other symbolic code. In support of its position, Red Bend points to a specific embodiment in the patent that contains symbolic code: "[B]y way of non-limiting example, the specified graph . . . represents a map where nodes stand for cities and links for roads linking the various cities." (`552 Patent at 15:9-13.) This example, Red Bend insists, is not an executable program and, in fact, is "exactly symbolic code." ( Markman Tr. At 53:18-24, 63:12.) In response, Google asserts that an example of a data table cannot "survive[] if it is not in an executable format" because of the disclaimers and the prosecution history. ( Id. at 69:14-22.) Google admits that its construction effectively limits data tables to executable programs.
I agree with Red Bend that to adopt a construction of "data table" with a wholesale exclusion of source or symbolic code would impermissibly read the non-limiting example out of the patent. ( See id. at 63:12-16.) Such a construction would also render the patent internally inconsistent, because the patent clearly envisions a "data table" and an "executable program" as distinct terms and concepts. However, I must reconcile this non-limiting example with the prosecution history's statement that data tables do not "embrace source code as in Okuzumi." The prosecution history is intrinsic evidence that "can often inform the meaning of the claim language by demonstrating how the inventor understood the invention and whether the inventor limited the invention in the course of prosecution, making the claim scope narrower than it would otherwise be." Phillips, 415 F.3d at 1317. To disregard completely the limitation in the prosecution history would therefore be improper because, "by distinguishing the claimed invention over the prior art, an applicant is indicating what the claims do not cover [and] he is by implication surrendering such protection." Ekchian v. Home Depot, Inc., 104 F.3d 1299, 1304 (Fed. Cir. 1997); see also Spectrum Int'l, Inc. v. Sterilite Corp., 164 F.3d 1372, 1378 (Fed. Cir. 1998) ("That explicit arguments made during prosecution to overcome prior art can lead to narrow claim interpretations makes sense." (citation omitted)).
I construe the word "embrace" in a manner similar to that employed in construing "involve" in the context of the term "executable program." ( Markman Tr. at 57:19-21.) In the executable-program context, I determined that "no source involved" meant that the program contained no source or other symbolic code "that is necessary for the execution of the program." In the data-table context, I analogously interpret "does not embrace source" to mean that the reference entries are not in source or other symbolic code but rather are "actual references to physical locations" in byte and numeric form. Red Bend admits that the `552 Patent does not operate on source or symbolic code, unless "you had source code where you had included in the source code some actual references to physical locations in the source and those became changed." ( Id. at 59.) The references at issue in Okuzumi are purely symbolic or relational, meaning that they do not point to a specific address. (Doc. No. 94, Ex. 7 ("Okuzumi Patent") at RedBend5434.) The references relevant for the `552 Patent are addresses pointing to unique physical locations. Thus, a data table may only include source in so far as the source contains specific physical addresses within the source.
Second, I find that the prosecution history of the `552 Patent sets forth "clear statements of [the] scope" of the claim term "data table" such that this term, when applied to an executable program, does not involve source and other symbolic code that is necessary for the execution of the program. See Teleflex, Inc. v. Ficosa N. Am. Corp., 299 F.3d 1313, 1327 (Fed. Cir. 2002). I therefore give the claim term "data table" a construction, not proposed by either party, of "a table of entries, each of which may have a different size. An executable program is one example of a data table. The data table cannot be source or other symbolic code that lacks references to a specific physical location or that is necessary for the execution of the program." This construction is faithful to the construction of "executable program," as well as to the patent itself, which states that a data table can — but need not always be — an executable program. (`552 Patent at 2:61-62.) 4. "Invariant References"
The patent describes a method for generating a modified new program in which
substantially each reference in an entry in said old program that is different than corresponding entry in said new program due to delete/insert modifications that form part of the transition between said old program and new program are reflected as invariant references in the corresponding entries in said modified old and modified new programs.
( See, e.g., `552 Patent at 16:31-36, 18:9-15 (emphasis added).) Red Bend asks me to construe "invariant references" as "values made the same in the modified old and new programs (or data tables) for corresponding reference entries so that substantially each reference that has changed due to delete/insert modifications does not appear in the difference result." Google construes this term as "values made the same in the modified old and new programs (or data tables) for corresponding reference entries so that the references are excluded from the difference result." ( Markman Tr. at 92:14-16.) The only practical difference between the two constructions is the inclusion of "substantially each," which Red Bend insists is necessary to be consistent with the court's construction of "compact difference result" and to reflect that not all references must be excluded from the compact difference result.
Independent Claims 42 and 55 are identical except that all references to "executable program" or "program" are replaced by the term "data table." ( See `552 Patent at 20:6-11, 21:51-56.)
Red Bend proposed this construction during and after the Markman hearing. In its Markman briefing, however, Red Bend construes this claim term as "values made the same" and disputed that the invariant references were necessarily excluded from the difference result. ( Markman Tr. at 88:4-18; Post- Markman Joint Statement at 2.)
I reject Red Bend's original construction of invariant references as inconsistent with the intrinsic evidence outlined above. Red Bend's initial position is also undermined by its preliminary-injunction briefing, in which Red Bend described the invention in the `552 Patent as follows: "The diff is then performed on the modified old and new programs, thereby generating a `diff' result that excludes the effect of these invariant references." Red Bend's original construction is further contradicted by its own expert, Dr. Stephen A. Edwards, who admitted that, in the preferred embodiment, "the reference addresses for such invariant references are excluded from the difference result." (Doc. No. 60 ("Edwards Reply Decl.") ¶ 21.)
In its Markman briefing, Google proposed: "Values made the same in the modified old and new programs (or data tables) for corresponding reference entries so that the reference addresses are excluded from the difference result." At the Markman hearing, Google agreed to remove "addresses" from the construction. Although Google argued against the addition of "substantially each" to any construction of "invariant references" during the Markman hearing, it also proposed an alternative construction including the phrase in a post-hearing submission: "Values made the same in the modified old and new programs (or data tables) for substantially each corresponding reference entry, that has changed due to delete/insert modifications, so that their references are excluded from the difference result." (Post- Markman Joint Statement at 2.) Because this construction is inconsistent with Google's reasoning and argumentation both during briefing and during the Markman hearing itself, I will consider the language recited at the Markman hearing to be Google's proposed claim construction for this term.
My construction of "compact difference result" does indeed reflect the possibility that not all references in an executable program or data table will be converted to invariant references and, consequently, be excluded from the difference result. At the Markman hearing, Google argued that the inclusion of "substantially each" in the construction of "compact difference result" obviates the need to include it in the construction of "invariant references." In fact, Google argues, to do so is inconsistent with the patent. Google explains that its construction is "faithful to the patent . . . because it deals with the `substantially each' term on . . . the front end," meaning that "there is a processing of substantially each of these corresponding references that are going to become the invariant references at the front end" and "once it goes into the process, there can be no selection of `substantially each' when it comes out of the process." ( Id. at 92:12-20.) By contrast, Google contends that Red Bend's proposed construction "would make the selection of `substantially each' on the back end" such that "[i]t might not be able to be processed in the first place." ( Id. at 92:25-93:7.)
I agree with Google that the inclusion of "substantially each" in the construction of "invariant references" is redundant, confusing, and contradictory. Patent `552 and its prosecution history recognize — as does the construction of "compact difference result" — that not all references may be replaced with invariant references. However, once corresponding references in the old and new programs are replaced with invariant references in the modified old and new programs, the object of the invention is to exclude those invariant references from the diff result, thereby rendering the diff result "compact." The construction of "invariant references," consequently, needs to reflect the characteristics only of those references made invariant — not those left unaltered from their original state.
The Summary of the Invention explains that "the invention aims at generating a modified old program and a modified new program, wherein the difference in references in corresponding entries in said new and old programs . . . will be reflected as invariant entries in the modified old and new programs." (`552 Patent at 3:36-41.) Accordingly, "[t]he net effect is that the invariant reference entries (between the modified old program and the modified new program), will not appear in the difference result," thereby reducing the size of the difference result or making it "compact." ( Id. at 3:41-43 (emphasis added).) Similarly, a preferred embodiment describes "an object of the invention" in which certain modifications, for example those caused by the insertion of entries, are "modified to invariant references with the obvious consequence that they are not reflected in the difference result. . . ." ( Id. at 10:1-14 (emphasis added).)
The prosecution history also suggests that all invariant references are to be excluded. In differentiating the `552 Patent's invention from prior art, the applicants explained to the Examiner that "[t]he modification [giving rise to modified old and modified new programs] is effected in such a way that references become `invariant' . . . thereby reducing the diff result and rendering it compact." (Prosec. Hist. at RedBend150.) That sentence in the prosecution history cross-references to the section of the Summary of the Invention that states that "the invariant reference entries . . . will not appear in the difference result. . . ." ( Id. (citing `552 Patent at 3:36-46).) Thus the prosecution history makes clear that the purpose of the invention is to render the difference result compact by excluding the invariant references. Should the invariant references appear in the compact difference result, it would, presumably, no longer be compact.
Consequently, I construe "invariant references" to mean "values made the same in the modified old and new programs (or data tables) for corresponding reference entries so that the references are excluded from the difference result."
5. "Modified Old/New Program/Data Table"
The `552 Patent contains four related terms: "modified old program" and "modified new program" in Independent Claims 8, 12, 21, and 25; and "modified old data table" and "modified new data table" in Independent Claims 42, 46, 55, and 59. Google proposes a single construction for all four terms, that is, "a version of the actual program or data table in its original executable form, with references replaced."
Google amended its initial proposed construction — "a version of the actual program or data table in its original executable form, with certain portions replaced" — at the Markman hearing by replacing "certain portions" with "references."
Red Bend initially sought separate constructions of each term in its claim construction submissions. It construed "modified old data table" as "an interim result, such as tables or data structures, related to the old data table." It requested that "modified old program" be defined as "an interim result, such as tables or data structures, related to the old executable program." The term "modified new data table," Red Bend argued, should be construed as "an interim result, such as tables or data structures, related to the new data table," and "modified new program" as "an interim result, such as tables or data structures, related to the new executable program." At the Markman hearing, Red Bend proposed an alternative single construction covering all four terms: "table(s) or data structure(s), related to the old/new program or data table, wherein substantially each of the references in corresponding entries that changed due to a delete/insert modification are reflected as invariant references." (Post- Markman Joint Statement at 2.)
My construction analysis begins with the claims themselves. The patent claims a three-step "method [or system] for generating a compact difference result between an old executable program and a new executable program." First, a "modified old program" is generated "utilizing at least said old program." (`552 Patent at 16:25-26.) Second, a "modified new program" is generated "utilizing at least said new program, said modified old program and modified new program" whereby the "delete/insert modifications . . . are reflected as invariant references in the corresponding entries in said modified old and modified new programs." ( Id. at 16:27-37.) Third, the compact difference result is generated "utilizing at least said modified new program and modified old program." ( Id. at 16:38-40.) Thus, it is clear from the claims that, as Google observes, the modifications are caused by the replacement of the "delete/insert modifications" with "invariant references."
The language of Claims 12, 21, and 25 is the same as that in Claim 8. The language found in Claims 42, 46, 55, and 59 is also identical, with the exception that "old/new data table" replaces "old/new executable program," and "modified old/new data table" replaces "modified old/new program."
The specification does not expressly define these disputed terms, but provides guidance on how the modified programs/data tables are generated. Patent `552 outlines two methods or systems by which to do this. The first method or system, which is not at issue in this litigation, involves replacing corresponding references with label markers. ( Id. at 3:47-4:22 (method), 5:4-46 (system).) The second involves replacing "substantially all" corresponding references with invariant references. ( Id. at 4:23-5:3 (method)), 5:47-6:26 (system).) Both methods or systems require the replacement of references either with label markers or invariant references.
The `522 Patent's Abstract states that "[t]he [label marker] method includes the steps of scanning the old [and new] program[s] and for each reference entry perform steps that include replacing the reference of the entry by a distinct label mark, whereby a modified old [and modified new] program is generated." ( Id. Abstract (emphasis added).) Analogously, the preferred embodiment provides an example of how, in the invariant-reference method or system, "the desired invariant references are accomplished by generating modified old and new programs wherein address references in entries are replaced by label marks." ( Id. at 10:47-50 (emphasis added).) This teaching of the patent is included in the claims. For example, Claim 1 (lable-mark technique) describes "replacing the reference of said entry by a distinct label mark, whereby a modified new program is generated," and Claim 8 (invariant-reference technique) explains that "substantially each reference . . . that is different . . . due to delete/insert modifications . . . are reflected as invariant references." ( Id. at 15:39-41, 16:31-37 (emphasis added).) As is apparent in the Abstract, the "reflect[ion of corresponding references] as invariant references" in the modified old and new programs and data tables is essentially the "replacement" of corresponding references with invariant references. Accordingly, the patent, by its own terms, supports the clause in Google's construction relating to the replacement of "references" in the programs and data tables.
Red Bend contends that it is improper to inject this "limitation" regarding "replacing" from "the unasserted independent claims [regarding the label-mark method or system] and the description of one embodiment." To the contrary, "[o]ther claims of the patent in question, both asserted and unasserted, can also be valuable sources of enlightenment as to the meaning of a claim term." Phillips, 415 F.3d at 1314 (emphasis added). This is "[b]ecause claim terms are normally used consistently throughout the patent, [and] the usage of a term in one claim can often illuminate the meaning of the same term in the other claims." Id.
Red Bend relies on the specification in support of its own constructions, but to no avail. Although the embodiment shows "exemplary old and new programs and the various interim results that are obtained by applying the sequence of operations" that "may be carried out at any known per se platform including, but not limited to, a conventional P.C., a computer network, etc., all as known per se," this statement does not support Red Bend's initial constructions. ( Id. at 9:16-18, 9:31-34.) There is no text expressly linking "interim results" and the modified old/new program/data table.
The prosecution history also undermines Red Bend's inclusion of "tables or data structures" in both of its constructions. To differentiate the invention from prior art, the applicants explained that Miller, U.S. Patent No. 5,832,520, does not "create[] modified old programs in the sense of the invention" because Miller creates, "in addition to an old file, an index or hash table," i.e., "auxiliary data," which is "in contrast to the invention" in the `552 Patent. (Prosec. Hist. at RedBend173.) Unlike the `552 Patent, the applicants explained, "Miller does not disclose the generation of modified old file, and a fortiori, not in the manner recited, for example, in step (a)(I), i.e., `replacing the reference of said entry by distinct label mark.'" ( Id. at RedBend174.) These statements by the applicants limit "the interpretation of claim terms so as to exclude any interpretation that was disclaimed during prosecution" because "[c]laims may not be construed one way in order to obtain their allowance and in a different way against accused infringers." Southwall Techs., Inc. v. Cardinal IG Co., 54 F.3d 1570, 1576 (Fed. Cir. 1995). Thus, the applicants' statement to the Examiner that the invention does not include auxiliary data, such as an index or hash table, undercuts Red Bend's proposal to construe the modification terms as "an interim result, such as tables or data structures."
Furthermore, Red Bend simply cherry-picks language from the preferred embodiments to craft its original and alternative constructions. In support of the word "table," Red Bend cites to an embodiment in which "the desired invariant references are accomplished by generating modified old and new programs wherein address references in entries are replaced by label marks" by creating a table. (`552 Patent at 10:47-11:3.) As for the word "data structure," Red Bend relies on the statement "P2 [the new program] was generated indirectly from the different result data D2, by the intermediary data structure. . . ." ( Id. at 13:61-64.) Red Bend is thus drawing its alternative construction entirely from the embodiments, which Red Bend itself, in another context, acknowledged to be inappropriate at the Markman hearing. See also Kara Tech. Inc. v. Stamps.com Inc., 582 F.3d 1341, 1348 (Fed. Cir. 2009) ("The patentee is entitled to the full scope of his claims [and is] not limit[ed] . . . to his preferred embodiment.").
For the reasons discussed above, I adopt Google's single construction for all four terms and construe each of the modification terms as "a version of the actual program or data table in its original executable form, with references replaced."
III. PRELIMINARY INJUNCTION
A. Standard of Review
"The grant or denial of a preliminary injunction under 35 U.S.C. § 283 . . . is within the sound discretion of the district court." Amazon.com, Inc. v. Barnesandnoble.com, Inc., 239 F.3d 1343, 1350 (Fed. Cir. 2001); see also 35 U.S.C. § 283 (permitting injunctive relief "to prevent the violation of any right secured by patent"). To obtain a preliminary injunction, the plaintiff must establish (1) the likelihood of success on the merits, (2) the likelihood that he will suffer irreparable harm in the absence of preliminary relief, (3) that the balance of equities tips in his favor, and (4) that an injunction is in the public interest. Winter v. Natural Res. Def. Council, Inc., 129 S. Ct. 365, 374 (2008) (citations omitted); see also Abbott Labs. v. Sandoz, Inc., 544 F.3d 1341, 1364 (Fed. Cir. 2008) ("`The standard for a preliminary injunction is essentially the same as for a permanent injunction with the exception that the plaintiff must show a likelihood of success on the merits rather than actual success.'" (quoting Amoco Prod. Co. v. Village of Gambell, 480 U.S. 531, 546 n. 2 (1987))). The Supreme Court has characterized "injunctive relief as an extraordinary remedy that may only be awarded upon a clear showing that the plaintiff is entitled to such relief." Winter, 129 S. Ct. at 375-76.
Before Winter, the preliminary-injunction standard evaluated the same four factors. The inquiry, however, was phrased as a balancing test, see Amazon.com, Inc. v. Barnesandnoble.com, Inc., 239 F.3d 1343, 1350 (Fed. Cir. 2001) ("These factors, taken individually, are not dispositive; rather, the district court must weigh and measure each factor against the other factors and against the form and magnitude of the relief requested." (quoting Hybritech, Inc. v. Abbott Labs., 849 F.2d 1446, 1451 (Fed. Cir. 1988))), with an emphasis on the first two factors, see id. ("[A] movant cannot be granted a preliminary injunction unless it establishes both of the first two factors . . .").
B. Likelihood of Success on the Merits
In a patent infringement case, to satisfy the likelihood-of-success prong, "the patentee must demonstrate that it will likely prove infringement of one or more claims of the patents-in-suit, and that at least one of those same allegedly infringed claims will also likely withstand the validity challenges presented by the accused infringer." Amazon.com, 239 F.3d at 1351. "In assessing whether the patentee is entitled to the injunction, the court views the matter in light of the burdens and presumptions that will inhere at trial." Titan Tire Corp. v. Case New Holland, Inc., 566 F.3d 1372, 1376 (Fed. Cir. 2009). If the alleged infringer "raises a substantial question concerning either infringement or validity, i.e., asserts an infringement or invalidity defense that the patentee cannot prove `lacks substantial merit,' the preliminary injunction should not issue." Amazon.com, 239 F.3d at 1350-51 (quoting Genentech, Inc. v. Novo Nordisk, A/S, 108 F.3d 1361, 1364 (Fed. Cir. 1997)).
1. Infringement
Red Bend argues that Google directly infringes the `552 Patent, as well as induces and contributes to others' infringement, by (1) using the patented methods in its Chrome Courgette updates; and (2) supplying code constituting a material part of the `552 Patent and instructing and encouraging others to use that code in an infringing manner. The determination of patent infringement involves a two-step analysis: "[T]he claim scope is first determined, and then the properly construed claim is compared with the accused device to determine whether all of the claim limitations are present either literally or by a substantial equivalent." Id. at 1351. Having construed the disputed claim terms above in Section II, I compare the construed `552 Patent to the allegedly infringing Courgette to determine whether all of the claim limitations are present. See id. a. Comparison of `552 Patent Claims to Courgette
To demonstrate a likelihood of success on the merits, Red Bend must show that it will likely prove infringement of at least one claim. Google admits that both Courgette and the `552 Patent address the problem of making software patches smaller, but contends that they do so in "a fundamentally different manner." More specifically, Google submits that Courgette does not infringe the `552 Patent for at least five reasons: (1) the symbolic tables and encoded streams that Courgette creates through a disassembly process are not "executable," as required by the patent claims; (2) Courgette does not generate modified old and new program data tables, as required by the claims; (3) Courgette does not process "substantially each reference"; (4) Courgette does not distinguish references that change "due to delete/insert modifications" from those that change for other reasons; and (5) Courgette does not create the required "invariant references."
All of the independent claims that Red Bend contends are infringed (Claims 12, 21, 25, 42, 46, 55, and 59) require that:
substantially each reference in an entry in said old data table [or program] that is different than corresponding entry in said new data table [or program] due to delete/insert modifications that form part of the transition between said old data table [or program] and new data table [or program] are reflected as invariant references in the corresponding entries in said modified old and modified new data tables [or programs].
( See `552 Patent Claims 8, 21, 42 55 (independent claims) (emphasis added); see also Claims 9-10, 22-23, 43-44, 56-57 (dependent claims).)
I focus my infringement analysis on "invariant references." Google contends that Courgette does not involve the claim limitation of invariant references because "Courgette preserves all unique references that it can recognize in a symbol table for the old and the new programs" and those unique references "retain their original values and are not changed to invariant." Furthermore, Google argues that Courgette's unique references are, in fact, reflected in the compact difference results, unlike the invariant references in the `552 Patent, which are excluded from the compact difference result.
Relying on a misapplied quotation from the deposition of Google's expert, Dr. Martin Walker, Red Bend claims that even under Google's construction, Courgette is still infringing. Dr. Walker, however, testified that in Courgette, "every address that is unique to the updated program appears in the difference result, which is in contrast to the method described in the Claim 42, for example, which says that . . . the addresses are replaced by invariant references." (Walker Dep. at 178:19-24.)
I am not satisfied that Courgette necessarily infringes on the `552 Patent because, at the very least, it does not encompass the claim limitation regarding invariant references. With insufficient evidence that Courgette infringes at least one patent claim, Red Bend cannot prove a likelihood of success on the merits.
Curiously, language in Google's opposition to the motion for preliminary injunction concludes that "Courgette does infringe because it does not use invariant references to replace references that change due to delete/insert modifications, and does not prevent them from appearing in the difference result." (Opp. at 22.) I assume Google inadvertently omitted the word "not" in that sentence.
b. Inducement or Contributory Infringement
Red Bend also argues that Google induces and contributes to infringement by Internet users and software developers. See 35 U.S.C. § 271(b) ("Whoever actively induces infringement of a patent shall be liable as an infringer."); 35 U.S.C. § 271(c) (defining contributory infringement). Even if Google is not liable for direct infringement, it may still be liable for inducement or contributory infringement. See Linear Tech. Corp. v. Impala Linear Corp., 379 F.3d 1311, 1326 (Fed. Cir. 2004). Critically, "[t]here can be no inducement or contributory infringement without an underlying act of direct infringement." Id.
Google argues that Red Bend fails to provide any evidence of direct infringement by third parties and therefore is unlikely to be successful in proving that Google induced or contributed to the infringement of the `552 Patent. Red Bend, for its part, claims that Google's open-source distribution of Courgette has led other companies to post the Courgette code on their websites and that at least two non-parties have executed and used the Courgette code, thereby constituting direct patent infringement. Merely downloading the Courgette source code, however, does not constitute infringement of the `552 Patent. See 35 U.S.C. § 71(a) (defining infringement as making, using, offering to sell, or selling a patented invention). Red Bend does not show that any third party has used the Courgette code in a way that infringes its patent. Moreover, "[t]he mere knowledge of possible infringement by others does not amount to inducement; specific intent and action to induce infringement must be proven." DSU Med. Corp. v. JMS Co., 471 F.3d 1293, 1305 (Fed. Cir. 2006) (internal quotation marks and citation omitted). Red Bend provides no evidence that Google intends for others to use Google's Courgette code to infringe the `552 Patent. Therefore, I find it unlikely that Red Bend will succeed on the merits of a claim that Google induced or contributed to patent infringement.
2. Validity
A patent has a presumption of validity. 35 U.S.C. § 282. "[W]hen the question of validity arises at the preliminary injunction stage, . . . the patent enjoys the same presumption of validity during preliminary injunction proceedings as at other stages of litigation." Titan Tire Corp., 566 F.3d at 1377. Because Google, the alleged infringer, has responded to Red Bend's preliminary injunction motion by challenging the validity of the `552 Patent, Google has the burden "to come forward with evidence of invalidity." Id. Red Bend, "to avoid a conclusion that it is unable to show a likelihood of success, then has the burden of responding with contrary evidence." Id. "Instead of the alleged infringer having to persuade the trial court that the patent is invalid, at [the preliminary injunction] stage it is the patentee, the movant, who must persuade the court that, despite the challenge presented to validity, the patentee nevertheless is likely to succeed at trial on the validity issue." Id. The trial court does not resolve, at the preliminary injunction stage, whether the patent is valid; rather, the court assesses the persuasiveness of the challenger's evidence, recognizing that more evidence may come out at trial. Id.
Google contends that the `552 Patent is invalid because it is both anticipated and obvious, citing several prior art references. Irrespective of Google's purported production to Red Bend of more than 250 prior art references, a number Red Bend calls "inflated," Google's validity arguments focus primarily on the Wetmore Patent. Google argues that "Wetmore discloses all of the steps required by the asserted independent claims of the `552 patent": specifically, that Wetmore generates a modified old program, a modified new program, and a difference result between the modified programs.
On January 22, 2010, Google filed and served a Request for Reexamination at the PTO, challenging the validity of the `552 Patent. The Federal Circuit counsels that district courts "should consider the current posture of the inter partes reexamination proceedings at the PTO when evaluating [the movant's] likelihood of success on the merits." Procter Gamble Co. v. Kraft Foods Global, Inc., 549 F.3d 842, 847 (Fed. Cir. 2008). "[A] requestor's burden to show that a reexamination order should issue from the PTO is unrelated to a defendant's burden to prove invalidity" at trial. Id. at 848.
On March 23, 2010, the PTO granted the request, explaining that "Wetmore raises a substantial new question by providing teachings that were not considered in a previous examination and that are relevant to the designated allowable subject matter." On May 28, 2010, after the parties appeared before me at the preliminary injunction hearing, the PTO preliminarily rejected Claims 1-4, 8-11, 14-17, 21-24, 27-28, 35-38, 42-45, 48-51, 55-58, and 61-62 under 35 U.S.C. § 102(b) as being anticipated by Wetmore. (Doc. No. 81, Ex. A at 5.) Additionally, Claims 2, 3, 9, 10, 15, 16, 22, 23, 36, 37, 43, 44, 49, 50, 56, and 57 were rejected as obvious under 35 U.S.C. § 103(a) as being unpatentable over Wetmore in view of U.S. Patent No. 5,790,796 ("Sadowsky"). ( Id. at 18.) Consequently, the PTO rejected each claim asserted by Red Bend in its preliminary injunction motion. Google contends that the PTO's rejection of these claims warrants denial of Red Bend's preliminary injunction motion. (Doc. No. 81 at 2.)
After the PTO office action, Red Bend asserted new claims in its Markman briefing, filed on July 15, 2010. Newly asserted claims 12, 25, 29-34, 46, 59-60, and 63-67 remain patentable and part of this infringement litigation. Google argues that "the status of these claims has no bearing on the Motion for Preliminary Injunction" because they were not at issue in that motion. In the absence of separate briefing, I do not consider these newly asserted claims in the preliminary injunction calculus.
However, the PTO had reached its preliminary determination before Red Bend had provided any submissions to the PTO in connection with the Reexamination. (Doc. No. 83 at 2.) On August 11, 2010, the PTO conducted an in-person interview with Red Bend's representatives and discussed Claims 1 and 8. (Doc. No. 111, Ex. 2.) As memorialized in the PTO's interview summary, the "examiner agree[d] that Wetmore does not teach the claimed invariant references or distinct label marks as claimed and defined by the specification." (Doc. No. 111, Ex. 2.) Accordingly, as of the Markman hearing, the only claim asserted in the motion for preliminary injunction that the PTO had allowed was Claim 8; Claims 21, 42, and 45 remained, at that time, preliminarily rejected. The PTO has since issued a Reexamination Certificate as to all claims. ( See U.S. Patent No. 6,546,552 C1, Feb. 15, 2011.)
The "current posture" of the Reexamination process, therefore, suggests that Red Bend is likely to establish validity of the `552 Patent at trial. See Proctor Gamble, 549 at 847-48 (noting that, to the extent that the defendant "challenge[s] the validity of the claims of the [patent] in the district court on the same bases it raised in the reexamination, the examiner's confirmation of the patentability of every claim in that [patent] may be relevant to [the plaintiff's] likelihood of success" in the preliminary injunction analysis). Consequently, Red Bend's success on all claims during Reexamination has satisfied me that "despite the challenge presented to validity [Red Bend] nevertheless is likely to succeed at trial on the validity issue." Titan Tire Corp., 566 F.3d at 1377. Red Bend thus has rebutted Google's invalidity challenge.
Overall, I conclude that Red Bend is unable to establish a likelihood of success with respect to infringement, but is able to establish such a showing as to validity of the patent. However, Red Bend's successful validity showing cannot overcome my determination that Red Bend has not demonstrated a likelihood of success on the question of infringement, without which there can be no likelihood of success on the merits.
C. Likelihood of Irreparable Harm
The Supreme Court's "frequently reiterated standard requires plaintiffs seeking preliminary relief to demonstrate that irreparable injury is likely in the absence of an injunction." Winter, 129 S. Ct. at 375 (emphasis in original). To that end, Red Bend argues that, as a result of Google's alleged infringement, a likelihood of irreparable harm may be shown by: (a) "an unquantifiable loss of market share for Red Bend's patented software update technology"; (b) "lost revenue and market opportunities"; and/or (c) "an unrecoverable loss of goodwill," including its reputation. To obtain a preliminary injunction, Red Bend must demonstrate more than "a possibility of irreparable harm." See id. at 375-76.
1. Availability of Money Damages
A threshold issue under the umbrella of irreparable harm is the availability of money damages. As a general rule, plaintiffs seeking injunctive relief must demonstrate "that remedies available at law, such as monetary damages, are inadequate to compensate for that injury." eBay Inc. v. Merc Exch., L.L.C., 547 U.S. 388, 391 (2006); see also Automated Merch. Sys., Inc. v. Crane Co., 357 F. App'x 297, 301 (Fed. Cir. 2009) (holding that the patentee bears the burden "to demonstrate that its potential losses cannot be compensated by monetary damages"). Red Bend must be able to demonstrate why its purported injuries entitle it to injunctive relief at this preliminary stage, when the defendant, a substantial public company, is perfectly capable of compensating Red Bend if found liable for infringement. See Nutrition 21 v. United States, 930 F.2d 867, 871 (Fed. Cir. 1991) (declining to find irreparable harm for purposes of a preliminary injunction where the alleged infringer "is acknowledged to be a large and financially responsible company which would be answerable in damages").
2. Presumption
Red Bend argues that there is a presumption of irreparable harm upon a showing of a high likelihood of success on the merits. But, in eBay, the Supreme Court reversed the Federal Circuit's departure from the traditional four-factor test and from its articulation of "a general rule, unique to patent disputes, that a permanent injunction will issue once infringement and validity have been adjudged." 547 U.S. at 393-94. The Court held that the Federal Circuit "erred in its categorical grant of [injunctive] relief." Id. at 394 (citation and internal quotation marks omitted).
After eBay, the Federal Circuit suggested that "[i]t remains an open question `whether there remains a rebuttable presumption of irreparable harm'" in the context of granting injunctions in patent infringement suits. Broadcom Corp. v. Qualcomm Inc., 543 F.3d 683, 702 (Fed. Cir. 2008) (quoting Amado v. Microsoft Corp., 517 F.3d 1353, 1359 n. 1 (Fed. Cir. 2008)). See generally Everett Labs., Inc. v. Breckenridge Pharm., Inc., 573 F. Supp. 2d 855, 865-69 (D.N.J. 2008) (collecting post- eBay cases). Nevertheless, slightly more than one year ago, a panel of the Federal Circuit held, in a non-precedential decision, that eBay "discarded" "the presumption of irreparable harm, based just on proof of infringement," and, therefore, the district court erred by granting a preliminary injunction in a patent infringement case. Automated Merch. Sys., 357 F. App'x at 301.
The First Circuit has not addressed the application of eBay to the presumption of irreparable harm. Two district courts in the First Circuit, however, have applied a presumption of irreparable harm post- eBay in the context of issuing preliminary injunctions for trademark infringement. See Nationwide Payment Solutions, LLC v. Plunkett, 697 F. Supp. 2d 165, 168 n. 1, 171-72 (D. Me. 2010) (presuming irreparable harm based on the plaintiff's showing of a substantial likelihood of trademark infringement); Operation Able of Greater Boston, Inc. v. Nat'l Able Network, Inc., 646 F. Supp. 2d 166, 176-77 (D. Mass. 2009) (concluding that because eBay "was confined to permanent injunctions issued under the Patent Act," the plaintiff was "entitled to a presumption of irreparable harm" in a trademark infringement case).
By choosing not to publish Automated Merchandising Systems, the Federal Circuit designated the opinion as "nonprecedential." See FED. CIR. R. 32.1. Nevertheless, I look to this case for guidance and find its reasoning persuasive.
After Automated Merchandising Systems, it appears that the presumption of irreparable harm is at best on life support. See Acoustics Processing Tech., Inc. v. KDH Elec. Sys., Inc. 697 F. Supp. 2d 146, 157 (D. Me. 2010) (holding that "to obtain injunctive relief, a party can neither rely upon a presumption of irreparable harm nor point to merely possibly harm" and "must show that irreparable harm is likely"); Graceway Pharm., LLC v. Perrigo Co., 697 F. Supp. 2d 600, 604 (D.N.J. 2010) (finding that "whether or not the rebuttable presumption of irreparable harm continues post- eBay, . . . the rebuttable presumption, to the extent it still exists, has been rebutted" in this patent infringement case). I decline to find continued vitality to the presumption. Instead, I find it prudent to follow the Supreme Court's guidance in Winter that a plaintiff must demonstrate "that irreparable injury is likely in the absence of an injunction." 129 S. Ct. at 375. Therefore, I will evaluate whether Red Bend has established this requirement for obtaining a preliminary injunction. 3. Showing of Irreparable Harm a. Loss of Market Share
First, Red Bend argues that it will suffer "an unquantifiable loss of market share" because an unascertainable number of Internet users could use the Courgette source code posted by Google in an infringing manner to update software programs. In a patent infringement case, "lost market share must be proven (or at least substantiated with some evidence) in order for it to support entry of a preliminary injunction, because granting preliminary injunctions on the basis of speculative loss of market share would result in granting preliminary injunctions `in every patent case where the patentee practices the invention.'" Automated Merch. Sys., 357 F. App'x at 301 (quoting Nutrition 21, 930 F.2d at 871).
Red Bend neither argues that Google is a market competitor nor that Courgette itself has caused Red Bend's products to lose their market share. Rather, Red Bend focuses on third parties who have posted the Courgette code on their respective websites for others to download, and at least two non-parties that have executed and used the code. But Red Bend offers no evidence that any of these non-parties has — or is inclined to — create software that competes with Red Bend's products. Red Bend can only speculate that those who access the Courgette source code are direct competitors in the market of developing software-updating technology and that Red Bend's products will lose market share as a result. Cf. i4i Ltd. P'ship v. Microsoft Corp., 598 F.3d 831, 861 (Fed. Cir. 2010) (upholding district court's conclusion that "i4i was irreparably injured by Microsoft's infringement, based on its factual findings that Microsoft and i4i were direct competitors in the custom XML market, and that i4i lost market share as a result of the infringing Word products"). Red Bend has made no showing of direct competition and lost market share simply because someone accesses and uses the Courgette code, and, therefore, I decline to find irreparable harm on this basis.
While Red Bend argues that the parties compete because "the market for connected devices has sufficiently converged," Red Bend's allegations of irreparable harm focus on adapting Courgette to update software other than on Google's Chrome browser. Such an injury is speculative, at best.
b. Lost Revenue and Market Opportunities
Second, Red Bend argues that, as a result of Google's free publication of the Courgette source code, "Red Bend's licensees will likely object to paying fees for software that is substantially similar to that which is being used for free by Google and others." For the same reason, Red Bend also argues that its ability to enter into new customer relationships will be hampered.
In general, "lost sales standing alone are insufficient to prove irreparable harm; if they were, irreparable harm would be found in every case involving a `manufacturer/patentee, regardless of circumstances.'" Automated Merch. Sys., 357 F. App'x at 300-01 (quoting Abbott Labs. v. Andrx Pharm., Inc., 452 F.3d 1331, 1348 (Fed. Cir. 2006)). "Lost sales (without more) are presumed to be compensable through damages, so they do not require injunctive relief." Id. at 301. As a result, "no matter how much evidence of lost revenue [Red Bend] presented, this evidence by itself could not support a finding of irreparable injury." Id. In any event, Red Bend has not claimed a single lost sale nor has it produced any evidence that it lost even one customer due to the availability of Courgette. The only assertion that Red Bend offers at this stage is speculation that its business might suffer. Furthermore, statements without any follow-through — especially where Red Bend concedes it has "not yet changed" its business model or strategy — are simply insufficient to demonstrate irreparable harm.
c. Loss of Goodwill
Finally, Red Bend argues that it will likely lose "goodwill" if Google continues to use and post the source code for Courgette. First, Red Bend claims that it has achieved a reputation as a leader in software updating and that Google is "tainting" its reputation by offering its patented technology for free. Second, Google purportedly "diverts and usurps the credit and reputation enhancement Red Bend deserves" by giving the "public perception" that Google invented Red Bend's patented technology.
While I agree with Red Bend that harm to goodwill may constitute irreparable harm in certain contexts, see, e.g., Bio-Tech. Gen. Corp. v. Genentech, Inc., 80 F.3d 1553, 1566 (Fed. Cir. 1996), I do not find that Red Bend has shown a likelihood of loss of goodwill here. Red Bend presents no evidence that its reputation has been tarnished in any way. See Tech-Wear, Inc. v. Acme Laundry Prods., Inc., 38 F. Supp. 2d 1147, 1152 (C.D. Cal. 1998) (declining to find loss of goodwill where plaintiffs provided "no evidentiary facts" to support that conclusion). Red Bend simply cites anonymous Internet comments that Courgette has been well received, which goes toward Google's reputation and not toward its own. Furthermore, mere general statements that Red Bend's reputation will be eviscerated by Courgette and Google's open-source policy are insufficient to demonstrate irreparable harm. See Abbott Labs. v. Selfcare, Inc., 17 F. Supp. 2d 43, 50 (D. Mass. 1998) (denying motion for preliminary injunction where plaintiff's argument for loss of goodwill was premised on a "somewhat general and conclusory" declaration).
In sum, Red Bend's theories only show a possibility of irreparable harm, falling short of the Supreme Court's standard that requires a demonstration that irreparable injury is likely in the absence of an injunction. Winter, 129 S. Ct. at 375.
D. Balance of Equities
The third factor — the balance of equities or hardships — "assesses the relative effect of granting or denying an injunction on the parties." i4i Ltd. P'ship, 598 F.3d at 862. To that end, Red Bend claims that, like the patentee in i4i Limited Partnership, 598 F.3d at 362-63 its "patented technology is central to [its] business" and most of its revenues are based on the `552 Patent, such that its market share and business strategy are "tied to the `552 Patent." Red Bend also urges that it "will suffer irreparable harm to its name, irretrievable loss of market share, and potentially the loss of its business," if injunctive relief is not granted. Red Bend argues, by contrast, that an injunction would hardly harm Google because Courgette only relates to a small portion of Google's large business; Google can easily remove the published source code with minimal cost; and Google can resort to a non-infringing alternative — namely, bsdiff, the predecessor to Courgette — to generate software updates.
Asserting that Red Bend's "argument relies entirely on its speculative irreparable injury assertions," Google explains that it will, in fact, suffer harm from all three aspects of the preliminary injunction sought by Red Bend. First, Google claims that ceasing use of Courgette would affect its ability to provide updates to Chrome web browsers quickly, securely, and at a reduced cost for tens of millions of users. Second, Google claims that taking down the Courgette source code and publishing an announcement that the code is subject to Red Bend's patent would "disproportionately harm Google's business and disrupt its open source efforts." Third, Google questions whether the requested relief would benefit Red Bend because, even if Google did not post the Courgette code, Red Bend's `552 Patent is already publicly available.
When balancing the asserted harms, I find the risk of patent infringement outweighs Google's open source efforts with respect to the Courgette code. Notwithstanding, I find Google would suffer harm from a preliminary injunction because millions of Chrome users would not have an immediate means to quickly and securely update their software, thereby rendering the Chrome browser stale within a relatively brief amount of time. While both parties have advanced compelling hardships, I find the scale tips in favor of Red Bend on this factor because most of Red Bend's software offerings practice the `552 Patent, and Courgette is just one small aspect of Google's technology empire.
In his deposition, Red Bend's chief executive officer, Yoram Salinger, conceded that certain Red Bend software offerings — e.g., vDirect mobile — do not practice the `552 Patent.
E. Public Interest
Regarding the fourth factor in the preliminary-injunction inquirty, in a patent infringement case, "the focus of the district court's public interest analysis should be whether there exists some critical public interest that would be injured by the grant of preliminary relief." Hybritech Inc. v. Abbott Labs., 849 F.2d 1446, 1458 (Fed. Cir. 1988); see also Mass Engineered Design, Inc. v. Ergotron, Inc., 633 F. Supp. 2d 361, 394 (E.D. Tex. 2009) ("Where products do not relate to a significant compelling public interest, such as health or safety, this factor weighs in favor of an injunction."). Red Bend argues that an injunction may be granted because the `552 Patent does not relate to such a critical public interest. The patent also covers Red Bend's updating technology business, Red Bend argues, and the public benefits when patentees are allowed to enforce their patent rights. Finally, Red Bend maintains, the public would be harmed — not benefitted — from "false and confusing signals from Google that its `open source' Courgette is free to be copied, when it is not."
Google, for its part, argues that this factor is "neutral." First, Google concedes that the `552 Patent — relating to software updates — does not affect a critical public interest, like medical devices or pharmaceuticals, that would be harmed by granting preliminary relief. Cf. Sanofi-Synthelabo v. Apotex, Inc., 470 F.3d 1368, 1383-84 (Fed. Cir. 2006) (finding a "significant public interest in encouraging investment in drug development and protecting the exclusionary rights conveyed in valid pharmaceutical patents" tipped in favor of the patentee (citation and internal quotation marks omitted)). Second, Google observes that a generic interest in protecting patent rights has only a "superficial appeal" because accepting that argument "at face value would render much of the four-part injunction analysis unnecessary because it would make the likelihood of success question dispositive." Kimberly-Clark Worldwide, Inc. v. Tyco Healthcare Grp. LP, 635 F. Supp. 2d 870, 881 (E.D. Wis. 2009).
Here, I too decline to "place great weight on the public interest in upholding the exclusive rights of patentees," see id., recognizing that the Supreme Court purposely requires movants to establish each factor in four-part test to obtain injunctive relief, see Winter, 129 S. Ct. at 374 (preliminary injunction); eBay, 547 U.S. at 391 (permanent injunction). Furthermore, I acknowledge Red Bend's concern that Google's open-source policy sends conflicting signals that infringing software is free to be copied; in this case, however, Red Bend has not shown a likelihood of patent infringement, thereby rendering the availability of free Courgette downloads harmless. I find the public interest factor neutral in this case.
F. Conclusion on Four Factor Test
I find that Red Bend is unable, at the preliminary injunction stage, to prove a likelihood of success on the merits or a likelihood of irreparable harm. Although I have determined that Red Bend has not established the requisite four factors for a preliminary injunction, I note that these "preliminary conclusions" neither constitute final rulings on the merits of the case, nor bind this court for purposes of further proceedings. See Titan Tire Corp., 566 F.3d at 1385.
IV. CONCLUSION
For the reasons set forth more fully above, I DENY the Plaintiff's motion (Doc. No. 8) for preliminary injunction.