Jacobsen v. Katzer, (N.D. Cal.)

We recently covered the summary judgment arguments in the model railroad free software case. Just days later a decision from the court granted Jacobsen victories on virtually every issue suitable for resolution on summary judgment.

Summary judgment was granted in favor of Jacobsen on the Anticybersquatting Consumer Protection Act (ACPA) claim, on his copyright infringement claim, on Defendant’s counterclaim for copyright infringement, and partial summary judgment on the DMCA Copyright management information claim (on everything but the knowledge and intent elements).

We had previously highlighted the actual damages issue as particularly noteworthy, and while the court does not reach an award of damages in this ruling, the court wrote,

“[T]he Court may assume at this procedural stage that there is some value for the expenditure of time invested in the JMRI Project and that it therefore has some monetary value.”

and

“[T]he Court is unpersuaded that Plaintiff cannot prove damages.”

The parties are supposed to attend another settlement conference shortly and the writing is on the wall on this one, so if both parties adopt reasonable positions going in to that conference, they should come out with a resolution. Timing will be interesting, however, as the Federal Circuit appeal on irreparable harm is slated for oral argument on Jan. 4, 2010, and that appeal might not go forward if a settlement is finalized beforehand.

Jacobsen v. Katzer, (N.D. Cal.)

Occasionally we will blog about “Other cases to watch” that are not on our Top 10 list.

On Friday, Dec. 4, 2009, I attended the oral argument in Jacobsen v. Katzer on the parties’ cross-motions for summary judgment. This case has made one trip to the Federal Circuit already and has oral argument in another appeal scheduled for Jan. 4, 2010. It seems to be less recognized that at least two interesting issues remain in the case at the district court level: 1) the alleged infringement of the selection and arrangement of certain data, and 2) the appropriate method of calculating actual damages to a free software project that, through failure to timely register its copyrights, is not eligible for statutory damages.

Background

In Jacobsen’s first appeal, the Federal Circuit vacated and remanded the denial of Jacobsen’s motion for a preliminary injunction premised on an allegation of copyright infringement. Jacobsen, a UC Berkeley physics professor, leads the Java Model Railroad Interface (JMRI) free software project which, at the time of this dispute, was provided under the Artistic license. The key issue on the first appeal was whether the alleged breach of a free software license, such as the Artistic License, provided JMRI with a cause of action for copyright infringement or merely for breach of contract.

This distinction was rightly seen as critical to the ability of free software projects to enforce their licenses, because copyright actions can bring the possibility of injunctive relief, statutory damages, and attorney’s fees, whereas contract claims generally cannot, unless specifically provided for in the contract. Given the limited resources of many community-based free software projects, each of the above could make the difference between the practical ability or inability to resolve a license breach.

The Federal Circuit held that Jacobsen was entitled to bring a copyright infringement claim, and did so in language that broadly affirmed free software licensing schemes in general. The court wrote,

Copyright holders who engage in open source licensing have the right to control the modification and distribution of copyrighted material… The choice to exact consideration in the form of compliance with the open source requirements of disclosure and explanation of changes, rather than as a dollar-denominated fee, is entitled to no less legal recognition.

That decision was hailed as a strong endorsement of the enforceability of both free software and Creative Commons licenses, with Lawrence Lessig calling it “a very important victory.”

On remand, however, the district court again denied the preliminary injunction, this time for failure to show specific and actual harm suffered or imminent as a result of the alleged infringement. That decision has also been appealed to the Federal Circuit.

Proceedings in the District Court

While the appeal on the preliminary injunction has been pending, the case has proceeded in the district court. Both sides filed motions for summary judgment on the remaining issues in the case:

Spolier Alert: Judge’s Notice of Tentative Ruling.

Jacobsen moved for partial summary judgment on the liability but not damages portions of his claims for cybersquatting and copyright infringement. He also moved for partial summary judgment on his claims for removal of and distribution of false copyright management information under 17 U.S.C. § 1202, but not on the intent element of that claim. Finally, Jacobsen also moved for summary judgment on Katzer’s counterclaim for copyright infringement. Katzer separately moved for summary judgment on Jacobsen’s copyright and DMCA claims.

Cybersquatting

“decoderpro” is a registered trademark of Jacobsen’s and he alleges that Katzer violated the Anti-Cybersquatting Protection Act (ACPA), 15 U.S.C. § 1125(d), when he registered decoderpro.com and later transferred it to another individual, purportedly under the condition that it not be transferred to JMRI. Katzer writes, “Assuming that Plaintiff can prove ownership, Defendants concede that all of the other elements to prove a cybersquatting claim are met, with the exception of Defendants’ bad faith intent to profit from the mark.” However, Katzer’s Opposition does almost nothing to address the nine non-exhaustive statutory factors indicative of bad faith and instead simply alleges it is a disputed question of fact. A simple search of TESS makes Jacobsen’s ownership of the mark clear, so this claim should come down to whether the court feels sufficient evidence of bad faith intent to profit has been shown, much of the evidence for which was redacted in Jacobsen’s brief. The court’s Dec. 3 tentative ruling makes no mention of the ACPA claim and asks no questions regarding it. It also was not discussed at oral argument. This suggests to me that the court has made up its mind on this issue and it seems extremely likely to me that Jacobsen will prevail on the ACPA claim. Under 15 U.S.C. § 1117(d), the court may award “statutory damages in the amount of not less than $1,000 and not more than $100,000 per domain name, as the court considers just.”

Copyright Infringement

Jacobsen’s copyright infringement claim would probably be unremarkable in any other case that lacked so many precedent-making issues, and perhaps it remains unremarkable despite this context. However, it is an assertion of a fairly thin copyright on behalf of a free software project that otherwise is extremely generous in granting permissions for downstream copying, modifying, and distributing its works. To be fair, Jacobsen, who was also in attendance at oral argument, clearly does not relish litigation and would prefer to write model train software in peace. Only when Katzer sent demand letters requesting significant patent licensing fees and when Katzer’s allegations made Jacobsen feel that his reputation and even job were jeopardized did he determine that litigating the case to clear his name was essential. Furthermore, if Jacobsen’s allegations are true, it is one thing for a free software project to grant a generous license to those who agree to abide by its terms, including particularly attribution, and quite another thing for that free software project to be expected to acquiesce in someone stripping out all such attributions and passing off the project’s work as their own for commercial gain. Nonetheless, because the copyrighted work at issue in this case is a selection and arrangement of data, it represents a type of copyrighted work that many free software proponents would, in many cases, prefer not be copyrightable at all.

Originality in Compilations

Model trains can be outfitted with a decoder that intercepts signals sent along the track to control the train’s speed, direction, lights, etc. Train manufacturers provide decoder files that detail which signals their decoders are responsive to and the National Model Railroad Association (NMRA) publishes standards and recommended practices for these systems.

Jacobsen explains that 102 files defining 291 decoders of the 500 decoders available on the market are at issue in this claim. JMRI developers chose the decoders they found most interesting and useful, and then structured the resulting definition files in a unique way, not based upon manufacturer or NMRA approaches, but reflecting the judgment of the JMRI developers as to their preferred arrangement, particularly choosing an arrangement appropriate for integrating these files with the rest of the JMRI software.

Katzer admits that JMRI’s files were downloaded (but this is permissible under the Artistic License) and admits to modifying the files, and distributing those modified versions. Were the license’s conditions followed (i.e., were changes noted and the copyright notices not stripped out) even modifying and distributing the files would have been permissible. However, since Katzer did not abide by the terms of the license, his defense rests on the argument that the elements of the work he ultimately distributed are not copyrightable.

One example discussed at length, includes a comparison of a small portion of one of these decoder definition XML files called QSI Electric. The files define multiple “Configuration Variables” (CV). The first CV in JMRI’s version of the file (which runs for 16 pages) is:

<variable label="Primary Address" CV="1" comment="Short address"
 item="Short Address" default="03">
 <shortAddressVal/></variable>

The comparable CV in Katzer’s version of the file (which runs for 36 pages) is:

<CV INDEX="1">
 <DEFAULT>3</DEFAULT>
 <MIN>0</MIN>
 <MAX>255</MAX>
 <NAME>Primary Address</NAME>
 <BITS BITS_DEFINED="NO" />
 <BIT_DEFINITION>
  <BIT_0 />
  <BIT_1 />
  <BIT_2 />
  <BIT_3 />
  <BIT_4 />
  <BIT_5 />
  <BIT_6 />
  <BIT_7 />
 </BIT_DEFINITION>
 <DATE_MODIFIED />
 <BitMask />
 <stringList />
</CV>

Katzer used this comparison to illustrate how certain default values or word choices were taken by JMRI developers from manufacturer manuals or NMRA standards, which Jacobsen acknowledges. However, Jacobsen objected to this bit-by-bit deconstruction of the work when analyzing originality, arguing that the issue instead was whether the selection and arrangement of JMRI’s decoder definition files, when viewed in the aggregate, is original. However, even when considering just this file, Jacobsen argued that evidence of creative selection is evident from the JMRI developer’s choices to omit or combine numerous CVs used by NMRA and in manufacturer manuals, particularly those for speed which JMRI handled differently. When Katzer engaged in wholesale translation of JMRI’s files into his format, he frequently duplicated these sorts of selections and omissions. The court seemed to be leaning towards a finding of sufficient originality and hence of copyrightability when Judge White asked something to the effect of: “Isn’t there enough in the record to show sufficient originality?”

One of the cases Jacobsen relies on is Key Publications, Inc. v. Chinatown Today Pub. Enterprises, Inc., 945 F.2d 509, 514 (2d Cir. 1991). In that case a telephone directory prepared for the Chinese-American community was held copyrightable because the selection and arrangement of the listings displayed sufficient originality. The selection of businesses to include was found original despite the fact that some were taken from another earlier restaurant directory and because of some decisions made to omit certain businesses. The arrangement of the listings were also held sufficiently original because the categories were not mere mechanical groupings or copied wholesale from elsewhere, and instead used some categories of special interest to the Chinese-American audience.

However, despite a finding of copyrightability for the directory, the allegedly infringing publication in that case was held non-infringing because it contained only 28 categories as opposed to plaintiff’s 260 and of those, only three appeared to be duplicate categories. Additionally, the selection of listings showed that only 17% of plaintiff’s 9,000 listings also appeared and that no sizable category was copied in its entirety.

Katzer claims that he only distributed decoder definition files for 107 of the 291 decoders covered by JMRI’s files. This then might be marginally comparable to using only 28 of 260 categories in a telephone directory. However, Katzer offers less evidence that the files for those 107 decoders do not duplicate the selections made by JMRI. That is, this would be like identically duplicating the listings in 28 of the 260 categories in a telephone directory. Had the defendant in Key Publications done that, the court’s reasoning suggests it likely would have found infringement.

Infringement of Compilations

Whether Katzer’s files substantially duplicate the selections made by JMRI developers for those 107 decoders is a factually-intensive inquiry that I have not undertaken. However, it does seem to me that Katzer should have attached an enormous exhibit to one of these summary judgment briefs detailing precisely how the files for those 107 decoders contain selections that differ from those made by JMRI developers. If there is an exhibit that would show this, then I cannot find it and it was not discussed at oral argument. Absent such evidence, I think Katzer fell into a trap of working from another’s compilation to create one’s own. Even where the underlying data is in the public domain or has an independent source, someone’s compilation of that data may reflect selections and omissions from those independent sources that one will inadvertently duplicate if one simply makes a wholesale translation of their compilation into your own format. Feist and Key Publications both acknowledge that, “Compilers operating under different principles of selection are not obligated to repeat factual research already undertaken and completed by others.” However, when working from another’s compilation, one must be sure to use a principle of selection other than “take everything.”

Thus, assuming one agrees with existing precedent that provides copyright protection to original selections and arrangements of otherwise uncopyrightable data, this appears to be a case that even those free software proponents who favor free and easy re-use of such data can still accept as a case of a thin copyright that deserved protection and that was infringed. It is, however, a fairly close case.

Actual Damages

Like many community-run free software projects, JMRI did not bother to register its copyrights until after they discovered a case of potential infringement. Under 17 U.S.C. § 412, registration before the commencement of the infringement¹ is a prerequisite to an award of statutory damages and attorney’s fees under sections 504 and 505. This means that Jacobsen can only seek actual damages for infringement as opposed to statutory damages. This case should serve as an enormous red flag to free software projects who would want to enforce their copyright in the event of an infringement that they should be taking the necessary steps to register their copyrights before any enforcement issues arise.

The district court is thus faced with another issue of first impression: What is the proper method of determining the actual damages incurred by a free software project whose copyright is infringed, when that project provides its software at no cost to the entire world (on the condition that the recipient abide by the accompanying license)?

Katzer argues that, in the Ninth Circuit, the appropriate approach to determining actual damages is to ask “what a willing buyer would have been reasonably required to pay to a willing seller for plaintiff’s work.” Frank Music Corp. v. Metro Goldwyn-Mayer, 772 F.2d 505, 512 (9th Cir. 1985). Katzer then argues that since JMRI’s decoder definition files have always been available for download at no cost, they have no market value, and JMRI has not suffered any actual damages. Again, since so many free software projects fail to timely register their copyrights and would also be forced to pursue actual damages, a decision that accepted Katzer’s argument here could permit widespread consequence-free infringement.²

Jacobsen responds that he is entitled to recover the value of the misappropriated work, citing to Sid & Marty Krofft Television Productions, Inc. v. McDonald’s Corp., 562 F.2d 1157, 1174 (9th Cir. 1977), which explains that a copyright owner may recover as actual damages “the reasonable value, if any, of plaintiffs’ work including the publication and republication rights therein, and the value, if any, to defendants of the use of plaintiffs’ works”, where “the value of use” “amounts to a determination of what a willing buyer would have been reasonably required to pay to a willing seller for plaintiffs’ work.”

Thus the parties largely agree on the appropriate legal standard for determining actual damages. They simply have a different view on how that standard applies to the facts of this case and Jacobsen contends that there is a factual dispute regarding damages that precludes summary judgment on this issue. Jacobsen has provided estimates of the time spent by JMRI developers and testimony regarding appropriate hourly rates for such work. Katzer dismisses this as the “acquisition cost” and not what a willing buyer would be willing to pay for a work that is provided at no cost over the internet. Jacobsen counters that the work is provided at no cost only under the terms of the license and so the appropriate hypothetical is to imagine that Katzer had come to JMRI and asked, “What would I have to pay to take a substantial portion of your decoder definition files and translate them into my own format, if I intend to strip out the attribution information, purport that the work is my own, and sell it to the world?” Faced with that question, JMRI developers probably would have been unwilling to provide the work at nearly any price as one of the few rewards they receive for their work is reputational, which can only arise through accurate attribution.

Instead of an approach that values the time of JMRI developers at zero or an approach that imagines JMRI developers being asked to “sell out” the more reasonable approach seems to be the one taken by Jacobsen’s expert which asks what Katzer would have had to spend to duplicate the work that he took. If Katzer had hired developers to make all the creative selections that JMRI developers made, Jacobsen’s expert concludes he would have spent between $153,000 and $157,600.³

What’s at Stake?

The most interesting issue remaining at the district court is how to evaluate actual damages for infringement of a free software project’s copyrighted work, which is distributed at no cost online, under the terms of a free software license. One of the lessons of the Federal Circuit’s opinion in this case was that exacting “consideration in the form of compliance with the open source requirements of disclosure and explanation of changes, rather than as a dollar-denominated fee, is entitled to no less legal recognition.” Thus when that compliance is found wanting, it cannot be that no remedy is available. Instead, it makes sense that the value of the use to the infringer can be reflected in an estimate of what he would have spent had he prepared the work without the benefit of infringed work. It will be interesting to see how the court comes out on this issue.

See also:

• JMRI, Jacobsen v. Katzer Docket.
• SFLC, SFLC Files Amicus Brief in Jacobsen v. Katzer, SFLC News (Jun. 15, 2009).
• Mike Madison, Jacobsen v. Katzer Continues, madisonian.net (Jan. 13, 2009).
• Pamela Jones, Court of Appeals for Federal Circuit Overturns Jacobsen v. Katzer – Ruling as text, Groklaw (Aug. 13, 2008).

1. Unless the work is registered within three months of publication. [↩]
2. If the infringement were done for profit, free software projects could alternatively seek to recover the infringer’s profits, but Katzer claims he suffered a net loss on his allegedly infringing product and in either that scenario or with infringement done not-for-profit, free software projects would likely be left with no monetary remedy. [↩]
3. I have been unable to find a rebuttal expert report from Katzer. [↩]

• Rescuecom Corp. v. Google, Inc. (2d Cir.), on remand to (N.D.N.Y).
• Ezzo v. Google, Inc. (M.D. Fla.).
• FPX, LLC v. Google, Inc. (E.D. Tex.).
• John Beck Amazing Profits, LLC v. Google, Inc. (E.D. Tex.).
• Google, Inc. v. John Beck Amazing Profits, LLC (N.D. Cal.).
• Stratton Faxon v. Google, Inc. (New Haven, Conn. Sup. Ct.).
• Soaring Helmet Corp. v. Bill Me Inc. (W.D. Wa.).
• Rosetta Stone Ltd. v. Google, Inc. (E.D. Va.).
• Flowbee Int’l, Inc. v. Google, Inc. (S.D. Tex.).

Background:

Google’s AdWords program, its primary source of revenue, generated over $20 billion in 2008. AdWords allows advertisers to purchase keywords that trigger the display of ads, or “sponsored links,” in response to user search queries. AdWords enables advertisers to purchase not only generic keywords to trigger their ads, but trademarks as well. So Boylan Bottling might purchase both “soda” and “Pepsi” as part of its AdWords buy.

In addition, Google’s Keyword Tool recommends keywords to advertisers; these recommendations likewise include trademarks. And although Google does not permit the use of marks in the titles of sponsored links, it does allow their use in the text of ads under some circumstances. When a mark is used in a descriptive or generic way, rather than as a reference to the trademark owner or its goods or services, its use is permitted. More recently, Google began allowing use of marks in ad text for certain classes of advertisers. Those who resell trademarked goods or services, or sell components, or replacement parts related to trademarked goods or services, or provide non-competitive informational sites that refer to trademarked goods or services are now able to use the relevant marks in the text of their ads.

Trademark holders, unhappy with the use of their marks by both Google and their competitors have filed numerous suits. As a result, AdWords has proven not only a remarkable source of revenue, but a reliable generator of litigation as well. By Eric Goldman’s count, after two recent voluntary dismissals by plaintiffs, there are eight pending AdWords cases.

What’s at stake:

These cases vary in a number of respects: the size and sophistication of the plaintiffs,¹ the strength of the marks, the particular components of the AdWords program to which plaintiffs object, and the constellation of legal claims relied upon to challenge them. Nonetheless, we can make some generalizations about the issues the AdWords cases collectively raise.

At a doctrinal level, these cases are likely to revisit the threshold question addressed in the Second Circuit’s Rescuecom decision, namely whether the inclusion of trademarked terms in the AdWords program constitutes a “use in commerce” under the Lanham Act. There the court rejected Google’s argument that use of marks within its internal coding falls short of “use in commerce.” This holding would seem to seriously undermine Google’s best strategy for disposing of AdWords suits on the pleadings, and open the door to some interesting fights over how to define, measure, and prove consumer confusion in the context of online keyword advertising. A previous AdWords case that reached these issues, GEICO v. Google,² raised as many questions as it answered about the appropriate standards for establishing confusion.

From a practical perspective, these cases have the potential to shape the structure of online advertising. The advertising programs of Google, Microsoft, Yahoo, and AOL have all been crafted, with varying tolerance for risk, with an eye to potential trademark infringement liability. Google’s success or failure in these cases will shape not only its own practices but those of its competitors.

Indirectly, these cases have important implications for consumers as well. The ability to communicate effectively about products and services depends in part on consumers’ ability to unambiguously identify the information they seek and competitors’ ability to unambiguously share that information with consumers. The trademark worries facing Google impede that communication. One need look no further than the absence of comparative advertising within AdWords. If Honda wants to tell consumers searching for the Toyota Prius that its own Insight hybrid is significantly cheaper, it has no easy way of doing so using AdWords.

In this sense, the AdWords cases push on a more fundamental question about the nature of trademark rights. Should we understand trademarks as a tool for consumer protection and empowerment? Or should this traditional justification for trademark law yield to the propertization of marks, the view that all uses of a mark fall within the trademark holder’s exclusive control?

See also:

• Eric Goldman, Flowbee Latest Trademark Owner to Sue Google–Flowbee v. Google, Technology & Marketing Law Blog (Aug. 14, 2009).
• David Johnson, Inside Rescuecom Corp. v. Google, Inc.: Does Google’s Use of Trademarks to Trigger Advertisements from Competitors Violate Federal Trademark Laws?, (Jul. 11, 2009).
• Joe Mullin, Of Patent Trolls and Trademark Champions: A Tale of Two Spangenbergs, The Prior Art (May 15, 2009).

Footnotes

1. Two recent cases have been styled as class actions. [↩]
2. Gov’t Employees Ins. Co. v. Google, Inc., 77 U.S.P.Q.2d 1841 (E.D. Va. 2005). [↩]