by Lee Gesmer | Dec 16, 2017 | Trade Secrets
A recent opinion from an SDNY federal district court reminded me of a client I represented many years ago. The company insisted that the core functionalities and user interface of its software program were trade secrets. A competitor had copied these, and the client asked my firm to commence a lawsuit for misappropriation of trade secrets.
An investigation showed that the client’s software was disclosed to prospective clients in one-on-one meetings and at trade shows without any requirement that a confidentiality or nondisclosure agreement be signed. Once a customer licensed the software an NDA was part of the sale, but before then it was not.
A secondary concern was that the software was used by hundreds of employees (via a site license), but these employees were not told that the software contained trade secrets – only the manager that signed the license agreement agreed to this. Was this sufficient to protect the trade secrets in the software? I was concerned it was not.
I advised the client that we could not bring suit against the client’s competitor. The client was not pleased with this advise, but so it goes.
The court’s opinion in Broker Genius, Inc. v. Zalta reminded me of this engagement. In this case Broker was seeking a preliminary injunction against a competitor based on similar facts. Source code was not at issue – Broker claimed trade secret misappropriation only of the user interface and functionalities that would be viewed by end users of the software.
The court found that Broker did protect the confidentiality of its software before it entered into a license by limiting what prospects could see – in that respect it did a better job than my client.
However, Broker’s primary tool for protecting its trade secrets was the Terms of Use implemented via a clickwrap license. But, there were problems with this license. First, as was the case with my client:
Broker Genius licensed its software to its customers on a per company basis. Because these customers include companies with as many as 120 employees … only the person who initially sets up the corporate account on behalf of her organization is required to assent to the Terms of Use … other employees who may be more involved in everyday use of the software would not have the opportunity to individually assent and would not know what the Terms of Use were …
Second, the Terms of Use paralleled the Copyright Act, but didn’t contain a confidentiality provision. Broker’s Terms of Use stated that users of Broker’s software may not “[r]eproduce, modify, display, publicly perform, distribute or create derivative works of the Site or Apps or the Content.” This language tracks the exclusive rights language of Section 106 of the Copyright Act, but does not, as the court stated, “amount to a confidentiality or non-disclosure clause that notifies users of the secrecy of any aspect of [Broker’s software] or preclude [licensees] from describing to others the software’s functions, structure, and appearance.”
Third, the court questioned whether a clickwrap license was sufficient to protect trade secrets, although it did not base its decision on this fact.
Based on factors one and two above, the court denied Broker’s motion for a preliminary injunction.
Here are my takeaways from this case –
- Don’t disclose your trade secrets to prospective customers. Either limit what they can see, or require an NDA. (Broker actually passed this test).
- Don’t use a clickwrap agreement to impose a confidentiality requirement. Although the court did not base its decision on the fact that Broker had done this, clickwraps are suspect in the eyes of many judges, and I would not advise a client to base such an important contractual provision on a clickwrap.
- If your software is going be used by many employees under a site license, the employees need to be reminded that the software is confidential every time they access the program, or at the very least periodically. You can’t assume an employee agrees to confidentiality based on a manager’s signature on a contract.
- Make sure your contract, regardless of form (paper or clickwrap), contains an appropriate confidentiality/trade secret provision. Merely echoing the Copyright Act won’t accomplish this.
Broker Genius, Inc. v. Zalta (S.D.N.Y, December 4, 2017).
by Lee Gesmer | Nov 29, 2017 | Trials
Uber is in trouble.
The trial between Alphabet’s Waymo and Uber over Waymo’s self-driving car trade secrets was scheduled to begin on December 4th before Judge William Alsup, of Oracle v. Google fame. (Readers familiar with coverage of that case know how smart and tough he is).
According to published reports, at the last minute evidence (a letter) was discovered suggesting that Uber has a team dedicated to collecting trade secrets from competing companies. Allegedly, the people involved use disappearing-message apps, anonymous servers, and secret computers and phones to communicate without leaving a trail. The purpose was to ensure there was no paper trail that would come back to haunt the company in any criminal or civil litigation.
However, now that this has been disclosed, that strategy has backfired.
Whether this Uber team targeted Waymo is not entirely clear, but there is enough suspicion that it did for Judge Alsup to have postponed the trial so Waymo can conduct additional discovery – discovery that would have already have taken place had Uber disclosed this earlier.
All of this has Judge Alsup royally peeved. According to press reports (links below) his comments at the hearing postponing trial included the following:
- “I can no longer trust the words of the lawyers for Uber in this case. … If even half of what is in that letter is true, it would be an injustice for Waymo to go to trial”
- “You [Uber’s lawyers] should have come clean with this long ago”
- “You’re just making the impression that this is a total cover up”
- “Any company that would set up such a surreptitious system is just as suspicious as can be”
- “I have never seen a case where there were so many bad things done like Uber has done in this case”
So, Uber has a federal judge who is about to start a trial (not next week, but probably very soon), who distrusts Uber and, perhaps more importantly, its lawyers.*
*This is not the first time Judge Alsup has had problems with Uber’s lawyers. See Judge blasts Waymo v. Uber lawyers, delays trial until December (Ars Technica, Oct. 3, 2017).
This is bad news for Uber. It’s important that a trial judge trust the lawyers trying a case to tell him the truth. Trial lawyers are constantly making representations to the judge about the law and evidence, and if the lawyers representing one side of the case lose the judge’s trust, they are at a huge disadvantage. Once a judge catches a lawyer in a lie the judge will question everything the lawyer says.
While Judge Alsup must follow the rules of evidence and civil procedure in his courtroom, there are countless ways he can tilt the scales against Uber. Studies have shows that jurors believe that the judge knows what’s “really going on” in a trial, and they watch the judge closely to try to pick up on how he views the case. The judge can communicate how he feels about the case through body language, tone of voice and a thousand other subtle clues that can’t be reviewed on appeal. He can also rule against Uber on motions and procedural issues on which he has discretion, without fear of being reversed.
All of this leaves Uber at a disadvantage before the trial even begins.
Presumably, Uber’s executives and in-house counsel (and perhaps the Board of Directors) are on top of this situation and trying to figure out how to proceed. What are their choices?
- Do nothing. Uber can decide to ignore its problems with the judge and proceed to trial. If the evidence supports Uber’s position that it has not actually used any of Waymo’s trade secrets (as Uber has repeatedly stated in its public announcements on this case), perhaps it can muscle its way through the case. It will have one strike against it, but that’s not an out.
- Retain new counsel. The judge may no longer trust Uber’s lawyers, but it’s probably too late to change law firms. However, it may not be too late to bring in a new lawyer (one that Judge Alsup is known to “trust”) to help try the case. Judge Alsup has not set a new trial date, and depending on when he does, this is not out of the question. If the new trial date is far enough in the future, it may be possible to bring in a new firm to take over as lead counsel. By doing that Uber would be signaling to the judge that it recognized his problem with current counsel, and took steps to replace them. Of course, this ignores the possibility that Uber hid information from its lawyers, and that all of the fault with the newly discovered evidence lies with Uber, something Judge Alsup may already suspect.
- File a motion asking Judge Alsup to recuse himself (withdraw from the case). A recusal motion may be based on evidence that a judge is biased or lacks impartiality. That said, recusal motions are always a long shot, and I don’t think Judge Alsup’s comments have crossed the line. And, an unsuccessful motion for recusal might only backfire and make things worse for Uber. Before filing a motion to recuse Judge Alsup, Uber would do well to keep in mind Emerson’s quote that “when you strike at a king, you must kill him.”
That said, I’m pretty sure some associate at Uber’s law firm is feverishly researching Ninth Circuit recusal law today.
- Settle the case. It’s almost certain that there have been settlement discussions in this case. Given its current problems, Uber’s cost/benefit analysis of settle/litigate has changed, and Uber should consider putting more on the table. This may include more money and some form of independent monitoring process that would give Waymo assurances that Uber is not using its trade secrets. It may even go so far as to put Uber’s self-driving car project on hold for some period of time.
None of these options is likely to be appealing to Uber, but these are the choices that face a party who has (and whose attorneys have) angered or alienated a judge. Uber is in a tough spot. How it will extricate itself – or whether it even can – remains to be seen.
See: Rebuking Uber Lawyers, Judge Delays Trade Secrets Trial (NYT, Nov. 28, 2017); Uber trial delayed after new evidence emerges (CNN, Nov. 28, 2017); Judge Tells Uber Lawyer: ‘It Looks Like You Covered This Up’ (NYT, Nov. 29, 2017).
Update: The case settled.
by Lee Gesmer | Nov 24, 2017 | Contracts
The first time I heard of a “buy/sell agreement” was around 1970 when I was 19 – just a few years before I fell in love with Steely Dan’s music.
Back then my father owned a textile business in Haverhill, Massachusetts with a 50-50 “partner,” and he explained that the business had insurance policies on both of their lives. If either partner died, the insurance would fund the purchase of the deceased partner’s shares, and the surviving partner would end up owning 100% of the company.
This, he explained was a “buy/sell agreement” – the deceased partner’s estate was bound to sell, and the surviving partner’s estate was bound to buy.
Fortunately, neither my father or or his partner ever had to exercise rights under this agreement – my father retired in 1981, and sold his half of the business to his partner.
It turns out that at around the same time that I first heard of a buy/sell agreement Donald Fagan and Walter Becker, the creative geniuses behind Steely Dan, were doing something similar.
In 1972 Fagan, Becker and several other band members, entered into a buy/sell agreement providing that upon the death of any owner of shares in Steely Dan, Inc., the company would buy back that member’s shares. A classic, simple buy/sell agreement that my father would have recognized. The most significant difference is that Steely Dan didn’t fund the purchase of the deceased member’s shares with insurance – instead the shares would be purchased based on the “book value” of the company, which would be determined by the company’s accountants.
According to Fagen this agreement has remained in effect and unaltered for 45 years, and by 2017 Fagen and Becker were the only two remaining shareholders in Steely Dan. If true, when Walter Becker died in October 2017 his heirs became obligated to sell Becker’s shares back to Steely Dan, leaving Fagen the sole owner of Steely Dan, Inc.
However, according to a lawsuit filed by Fagen this month (November 2017), Becker’s heirs have refused to honor the buy/sell agreement, claiming it is no longer in effect. Fagen’s lawsuit attempts to force them to sell Becker’s shares back to Steely Dan.
What to make of all of this?
Forty-five years is a long time, and perhaps Becker’s estate has some basis for claiming that the buy/sell agreement was terminated or superseded. At this point we don’t know their side of the story. But given the existence of this agreement (which is not notarized but appears authentic), Becker’s estate faces an uphill fight.
If the agreement is enforceable, what can Becker’s estate expect to receive for Becker’s shares in Steely Dan, Inc.? Book value is simply assets minus liabilities, and Fagan’s complaint doesn’t disclose what Steely Dan’s assets and liabilities are. However, it’s worth noting that the agreement excludes from book value the “good will” of Steely Dan, which if included would have complicated and increased the determination of book value – the monetary value of “good will” can be difficult to establish, and by excluding this the members of Steely Dan likely were trying to keep things simple.
The agreement also doesn’t make any reference to who owns the most valuable assets associated with Steely Dan – the copyrights in the band’s compositions and the band’s recording contracts. It seems likely that revenues from these sources go directly to Becker and Fagen as composers and performers, but at this point we don’t know. If this is in fact the case, the only assets in Steely Dan, Inc. may be the name of the band (and therefore the right to perform in concert under the name), the domain name “steelydan.com” (which is referenced as an asset in the complaint) and the right to make commercial use of the “Steely Dan” trademark.
Most likely this case will settle out of court. If it doesn’t, we may learn more about the legal and business relationship between these musicians.
Fagen v. Estate of Walter Becker (buy/sell attached)
Update: Becker family response
by Lee Gesmer | Jul 21, 2017 | Internet Law
“One country shouldn’t be able to decide what information people in other countries can access online”
David Price, senior product counsel at Google
A risk long anticipated by Internet law observers is that the courts might become more aggressive in regulating online behavior, not just in their own nation, but worldwide. Google, more than any company, has had a target on its back for this kind of case.
The obvious example would be a court in one nation ordering Google to takedown (“de-index”) search results for users worldwide. This is exactly what happened in the Canadian Supreme Court’s decision in Google Inc. v. Equustek Solutions Inc. (June 28, 2017). This case represents the first time the highest court in a country has ordered a search engine to de-index worldwide in the context of a purely commercial two-party dispute. And, as we shall see below, this is a crucial issue for Google – one that it will not concede without a fight.
The facts of the Canadian case are straightforward. Equustek, a Canadian company, sued its former distributor, Datalink, on various grounds, including trade secret misappropriation. Datalink moved its offices and web host outside of Canada, but continued to sell its product online. A court order directed at Datalink was ineffective, so Equustek turned to Google, demanding that it remove links to Datalink from Google’s search engine. Google was prepared to comply in Canada, but Equuestek insisted that Google de-link Datalink worldwide.
After a series of lower-court decisions granted Equustek’s demand for worldwide de-indexing of Datalink sites Google appealed the case to the Supreme Court of Canada.
The Supreme Court upheld the lower Canadian court. It rejected Google’s argument that the Canadian courts did not have the legal authority to issue a worldwide takedown injunction. Noting that Datalink would not be commercially viable without Google’s search results, the court found that Google was “facilitating Datalink’s breach … by enabling it to continue carrying on business through the Internet.” Since the Internet has “no borders,” only a worldwide injunction would be effective to protect the Canadian company.
Google argued that a worldwide takedown order could run afoul of laws in other jurisdictions, but the court held that Google could seek modification of the order should that prove to be the case: “we are dealing with the Internet after all, and the balance of convenience test has to take full account of its inevitable extraterritorial reach when injunctive relief is being sought against an entity like Google.”
While Google has no further appeals in Canada, it’s seeking a partial reprieve in the United States. On July 24, 2017, a month following the decision in Canada, Google filed suit against Equustek (the plaintiff in the Canadian case), asking a California federal district court to:
Declare that the Canadian Order is unenforceable in the United States as inconsistent with the First Amendment, the Communications Decency Act, and the public policy surrounding enforceability of foreign judgments pursuant to international comity.
Google opened its complaint as follows:
Google brings this action to prevent enforcement in the United States of a Canadian order that prohibits Google from publishing within the United States search result information about the contents of the internet. As part of a Canadian lawsuit brought by Canadian plaintiffs against Canadian defendants, a Canadian trial court enjoined Google (a non-party based in California) from including in its search results links to dozens of the Canadian defendants’ websites—not just on Google’s www.google.ca site for Canada, but worldwide, including within the United States. As a result, Google, alone among search engines and other providers of interactive computer services, is compelled to censor the information it provides to its users around the globe about the existence of the Canadian defendants’ websites.
How Google’s U.S. case will proceed raises some interesting and potentially important issues, including the following:
- Google’s complaint is based on three legal grounds: that the Canadian ruling violates the First Amendment, the Communications Decency Act* and U.S. public policy. Each of these contentions is likely to implicate complex legal arguments. The last issue raises difficult issues of international comity – in this case the willingness of a U.S. court to override a court decision from another jurisdiction affecting a U.S. corporation.
*The Communications Decency Act states that: “No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.” 47 U.S.C. § 230(c)(1). Likely, under this U.S. statute Google could not be forced to de-index search engine links to Datalink’s web pages.
- Even if the U.S. court does override the Canadian case in the U.S. and Google restores links to Datalink’s web pages in the U.S., Google would still be in violation of the injunction in Canada. Will Google petition the Canadian courts to modify the Canadian order to permit re-indexing in the U.S., and how would a Canadian court respond to such a petition?
- At a practical level, who will defend this suit, which could easily go to the Ninth Circuit, and perhaps the Supreme Court? Equustek, not the Canadian government, is the defendant, and it may not even have the financial resources to defend this case in the U.S. If it doesn’t, who will?
- Again, as a practical matter, what if Datalink (the original bad actor) goes out of business or removes the offending pages while the case is pending? Will the U.S. case be moot, leaving the Canadian decision standing as a potentially important precedent?
While the presence or absence of links to Datalink is of no financial significance to Google, Google v. Equustek represents an existential threat to the Internet that Google takes seriously – the Balkanization of the Internet. Countries with different views of permissible online speech could lead to an Internet where what you see in one country (such as the U.S.) may be different from what you see in another (such as Canada).
The risk that worldwide takedowns will be the future of the Internet is very real, and Google appears to be committed to preventing, or at the least minimizing, this outcome by challenging it, as it did in Canada. If it can’t achieve that, as it could not do in Canada, it will accept Balkanization as a second choice, as it is now seeking to do through its U.S. suit.
For readers who may be interested in the legal theories Google is advancing in the U.S. suit, I recommend the Electronic Frontier Foundation’s amicus brief in the Canadian case, which is available here.
Update 1: Google has filed a motion for a preliminary injunction prohibiting enforcement of the delisting order in the U.S. Link here
Update 2: on November 2, 2017, the federal district court for the district of Northern California enjoined enforcement of the Canadian order in the U.S. on the grounds that it was inconsistent with the Communications Decency Act.
The court’s opinion is available here. As I suggested might be the case, Equustek did not enter an appearance in the case to oppose the lawsuit.
Update 3: On December 14, 2017 the court entered a default judgment and permanent injunctive relief (here).
by Lee Gesmer | Jun 7, 2017 | DMCA/CDA
While many performing artists and record companies complain that the Digital Millennium Copyright Act (the “DMCA”) puts them to the unfair burden of sending endless takedown notices, and argue that the law should require notice and “stay down,” supporters of Internet intermediaries and websites argue that court decisions have unreasonably narrowed the DMCA safe harbor.
A recent decision by the influential Ninth Circuit Court of Appeals (which includes California) adds to the concerns of the latter group.
LiveJournal, the defendant in this case, displayed on its website 20 photographs owned by Mavrix. Mavrix responded, not by sending DMCA “takedown” notices, as you might expect, but by filing suit for copyright infringement. LiveJournal responded that it was protected by the DMCA. However, to successfully invoke the DMCA’s safe harbor LiveJournal had to satisfy all of the legal requirements of the DMCA.
A key requirement is that infringing content have been posted “at the direction of the user.” In other words, the DMCA is designed to make websites immune from copyright infringement based on postings by users; it doesn’t protect a site from content posted or uploaded by the site itself – that is, by the site’s employees. The photos at issue were submitted by users and posted at their direction and therefore, LiveJournal argued, it satisfied this DMCA requirement.
However, when it comes to the DMCA, the devil is in the details, and the outcome in any case depends on how the courts interpret those details. In the case of LiveJournal photos are submitted by users, but they are posted only after they are reviewed and approved by volunteer moderators. For this reason, Mavrix argued, the photographs were not “posted at the direction of the user,” rather they were posted by moderators who selected them from user submissions. Further, Mavrix argued that the moderators were “agents” of LiveJournal, and therefore their actions were legally attributed to LiveJournal. In other words, as “agents” of LiveJournal their actions were the same as if they were employees.
The district court rejected Mavrix’s arguments and ruled for LiveJournal, but the Ninth Circuit reversed, holding that Mavrix had a point – the moderators might very well be “agents” of LiveJournal, in which case LiveJournal would have failed this requirement of the DMCA and be liable for copyright infringement. In reaching this conclusion the court emphasized that the critical inquiry is not who submitted content, but who posted the content. The court rejected LiveJournal’s position that the words “at the direction of the user” include all user submissions, even when they are reviewed and selected by agents or employees of the service provider.
In the case of LiveJournal, because moderators screened and posted user submissions the issue is whether the moderators are “agents” of LiveJournal whose actions should be attributed to LiveJournal. In effect, the court equated agents with employees.
To make matters worse for websites hoping to use volunteer moderators, the legal “test” to determine whether moderators are agents gets into the arcane subject of agency law, a topic that rightly triggers the limbic system of any lawyer who suffered through agency law in law school. In this case the question is whether the level of control LiveJournal exercised over its volunteer moderators created an agency relationship based on “actual” or “apparent” authority. Trust me when I say that these are complex issues that no website owner would want to have to parse out while running its business, much less have to present and argue to a jury.
This ruling was a blow to LiveJournal, but the Ninth Circuit had more bad news to deliver. Even if LiveJournal was able to establish that the moderators were not agents of LiveJournal, Mavrix might be able to show that LiveJournal had “actual” or “red flag” knowledge that the postings were infringements. While the Ninth Circuit stated that “red flag” knowledge requires that the infringement be “immediately apparent to a non-expert,” the court ruled that the fact that some of the photos contained watermarks could have created red flag knowledge. Whether they did will be up to a jury to decide. If a jury decides that LiveJournal had the requisite knowledge with respect to one or more photos, LiveJournal will lose DMCA protection for those photos.
However, after this ruling the Ninth Circuit was still not done with LiveJournal. The DMCA requires that LiveJournal not have received a financial benefit from infringements that it had the right and ability to control. The court held that this benefit “need not be a substantial or a large proportion” of the website’s revenue, and added to the confusion around DMCA law by suggesting that such a benefit could be established based on the volume of infringing material on the site, even if this material did not belong to Mavrix and was not the subject of the current litigation.
What lessons can website operators draw from this case?
First, this is a very bad decision for any social media website that would like to moderate content, whether through the use of volunteers or by employees.
Any business based on LiveJournal’s business model – volunteer moderators who review user submissions and decide whether or not to post them – is at serious risk of losing DMCA protection. The degree of planning, administration and ongoing legal supervision necessary to be confident that moderators are not agents would be daunting.
It’s worth noting that this decision will be difficult to evade – the fact that a site may not be in one of the states included in the Ninth Circuit is not likely to provide protection. A site incorporated and operating outside the Ninth Circuit can be sued in the Ninth Circuit if it has minimum contacts there, and this is often easily established in the case of popular websites. The Mavrix case is so favorable for copyright owners seeking to challenge a DMCA safe harbor defense that it is likely to motivate forum shopping in the Ninth Circuit.
Second, the irony of this case is obvious – Mavrix creates an incentive to engage in no moderation or curation and post “all comers.” If there is no moderator (whether an agent or employee) to view a warning watermark, there can be no knowledge of infringement. Unregulated postings of user generated content is likely to result in more copyright infringement, not less.
Third, to add to the confusion over how the DMCA should be applied to websites that host user-generated content screened by moderators, the Court of Appeals for the Tenth Circuit issued a decision in 2016 that appears to come to the opposite conclusion regarding the use of moderators. BWP Media USA Inc. v. Clarity Digital Group., LLC. This case may give LiveJournal a chance to persuade the Supreme Court to accept an appeal of the Mavrix case (based on a “circuit split”) – assuming, that is, that LiveJournal has the stomach, and the budget, to prolong this case further, rather than settle.
Lastly, it’s important to view this decision in the context of the DMCA as a whole. Any service provider hosting user-generated content has to pass through a punishing legal gauntlet before reaching the DMCA’s safe harbor:
(i) content must be stored at the direction of a user;
(ii) the provider must implement a policy to terminate repeat infringers, communicate it to users and reasonably enforce it;
(iii) the provider must designate an agent to be notified of take down notices, register the agent online by the end of 2017 and post the contact info for the agent online on the site;
(iv) the provider must respond expeditiously to take down notices;
(v) the provider may not have actual or red flag knowledge of infringement, nor may it be willfully blind to infringements;
(vi) the provider may not have the right and ability to control infringing content; and
(vii) the provider may not have a direct financial interest in the infringing content.
This is a challenging list of requirements and, as illustrated by the Mavrix case, each requirement is complex, subject to challenge by a copyright owner and subject varying interpretations by different courts. If the service provider fails on even one point it loses its DMCA safe harbor protection.
After the Ninth Circuit’s decision in Mavrix the chances that a service provider will be able to successfully navigate this gauntlet are significantly reduced, at least in the Ninth Circuit.
Mavrix Photographs, LLC v. LiveJournal, Inc. (9th Cir. April 7, 2017)
Update: The Ninth Circuit clarified its position on whether the use of moderators deprives a website of DMCA protection in Ventura Content v. Motherless (2018). In Ventura the court held that screening for illegal material is permissible, and distinguished Mavrix on the ground that in that case the moderators screened for content that would appeal to LiveJournal’s readers (“new and exciting celebrity news”). “Because the users, not Motherless, decided what to post — except for [its] exclusion of illegal material . . . — the material . . . was posted at the direction of users.”
