Pages

Monday, March 31, 2014

How far will the Supreme Court go to stop patent trolls?

Today's case: An Australian patent holder's idea boils down to 7 lines of code.

What kinds of software—if any—deserve a patent?
It's a basic question, and one that many thought would be resolved by now, especially since the issue came up in the US Supreme Court's 2010 Bilski decision. But Bilski left the legal landscape more confusing and fractured. This was illustrated in dramatic fashion last year, when the nation's top patent court, the US Court of Appeals for the Federal Circuit, practically split apart at the seams trying to decide if four patents on financial software held up. The Alice Corp. v. CLS Bank case resulted in ten judges writing seven different opinions and resolving nothing.
Seven months later, the Supreme Court agreed to take up the controversial case, setting off the most intense round of legal positioning over patents and software yet seen. In the four years since Bilski, the spread of "patent trolls"—that is, companies that do nothing but sue over patents—has gone from a tech-sector talking point to a white-hot, Main Street controversy. Dramatic examples like Wi-Fi trolls suing coffee shops and "scanner trolls" threatening thousands of businesses for the use of everyday technology have spurred debate in Congress. Several proposals were combined to become the anti-patent-troll Innovation Act, which passed the House on a rare bipartisan vote and is scheduled to be debated shortly in the Senate. 
Whatever happens in Congress, the Supreme Court's decision in this case—or indecision—matters. The court is aware of patent trolls, and it seems to genuinely dislike them. The justices also seem aware that Bilski did not do the job, and they want to devise some legal means of executing patents they don't like. The question then becomes "how?"
"How" matters a lot here. Practically no observers expect Alice Corp. to keep its four patents, but the question of how many other patents get lassoed into today's decision will resonate for years to come. Some would like to see the Supreme Court end Alice's patents with sniper-like precision, touching no other patent; others are eager for a pronouncement that would deal some broader harm to the patent troll business model.
In theory, the Court could go so far as to simply say software is not patentable. That's a very unlikely outcome, but it can't be ruled out entirely. The tiny possibility of such an outcome has spurred companies that favor software patents—including Microsoft, Adobe, and IBM—to file briefs focused on their importance. (While these companies extoll software patents as a general category, no one defends Alice's patents.) 
Internet companies, meanwhile, are trying to steer the court's attention to the harm caused by vague software patents. A pointed, 12-page brief filed by LinkedIn, Twitter, Yelp, Newegg, Netflix, Rackspace, and several smaller software companies does not go so far as to call for the abolition of patents on software, but it makes the signors' distaste for them crystal clear. "Innovation happens despite software patents, not because of them," states that brief, written by Stanford professor and patent litigator Mark Lemley.

Seven lines of code, $5 trillion of trades

While many of the lawyers briefing the case, including CLS Bank's attorneys, are too polite to say it, Alice Corporation fits the classic model of a "patent troll." Despite a website with language about wanting to launch "an innovative form of derivatives market," the company appears to have never created any product. By its own admission, it has been focused on its lawsuit since 2007. The inventor of the patents and founder of Alice is Ian Shepherd, previously managing partner at the Melbourne branch of McKinsey & Company, a financial consultancy. National Australia Bank, one of Shepherd's former clients and the fourth largest bank in Australia, became a 50 percent owner of Alice in 2000.
Alice stands out from most patent trolls in that its litigation strategy is focused on a single company. Sometimes one is enough, though, and CLS is no ordinary defendant. The Economist called CLS Bank "the most important bit of the financial infrastructure you have never heard of." CLS is a membership organization, composed of a who's who list of more than 60 of the world's biggest banks, including the three largest US banks, Bank of America, JP Morgan Chase, and Citigroup. Each day, CLS settles accounts worth $5 trillion in various currencies. A major patent win against CLS could have raised the cost of international money-exchange worldwide.
Alice's brief spends seven pages describing Shepherd's inventions, dividing them between "system" claims, "method" claims, and "media" claims. It describes a complicated system of "shadow accounts," updated on a "real-time basis," which mitigates risk in "exchange institutions."
But other than Alice's lawyers and several Federal Circuit judges, nobody who has delved deep into the patents has been too impressed. Underneath bloated language, critics say, Alice's patents describe how anyone performing an exchange of goods with an unknown party would proceed: find a trusted third party.
"That's what people have looked for when they've needed to exchange things since the beginning of time," says Charles Duan, an attorney at Public Knowledge who has analyzed Alice's patent claims. "If you and I want to exchange some money—dollars for euros—there's some risk that you will give the money and then I don't give you the money."
The world's banking regulators were well aware of that risk, and CLS Bank was custom-designed to address it. The systems used by CLS, which stands for "continuous linked settlement," are undoubtedly complex, but the bank itself is a classic example of an idea whose time had come—it was "decades in the making," as CLS' lawyers explain in their brief. In a 2002 article, The Economist explained that banking regulators had a name for the risk that one party in a foreign-currency trade wouldn't pay up: Herstatt risk, named after a German bank that closed halfway through a trading day in 1974.  
The Alice patents are covered in complicated-sounding language about computers, but critics like Duan see the complex language as little more than smoke and mirrors. The bloated nature of the patents is cleverly addressed in the amicus brief filed by Public Knowledge and the Application Developers Alliance, written by Duan, which distills the "very simple, basic idea" that Alice hid "beneath a veneer of technical language." Duan took the most broadly accepted claim in Alice's batch of patents, claim 26 of US Patent No. 7,725,375, and wrote a computer program that satisfies all the instructions of the 200-word claim. It is seven lines long:
10 LET account1 = 200.00
20 LET account3 = 300.00
30 INPUT "Value to exchange for transaction"; exchange
40 IF account1 < exchange THEN PRINT "Inadequate value" : STOP
50 account1 = account1 - exchange
60 account3 = account3 + exchange
70 PRINT "Instruction to 1st institution: adjust 2nd account by "; - exchange
It doesn't require any programming experience to interpret what's going on in the above code. Check everyone's account to make sure there's enough money; if there is, do the trade. "This is basically a bank transfer," said Duan. But it's also the claim that split the patent appeals court down the middle. 
Alice describes the claim this way:
The computer in claim 26 is configured to receive transactions from the parties to an exchange, to make adjustments electronically to the shadow accounts independently maintained by the computer after ensuring that those accounts reflect “adequate value,” and to generate instructions to the exchange institutions to implement the exchange in the separate real-world accounts maintained by those institutions.
The company goes on to note that the specification "contains flowcharts that provide algorithm support for the specific programming required." That seems to have been effective. In the Public Knowledge brief, Duan posits an explanation of how four Federal Circuit judges came to embrace the claim:
Certain judges below were misled by the language of the claims and the patent. Judge Rader believed that Claim 26 involved “a complex problem” that could only be solved with a specialized system with “at least four separate structural components.” He reviewed the “at least thirty two figures which provide detailed algorithms” to conclude that “[l]abeling this system claim as an ‘abstract concept’ wrenches all meaning from those words.” Judge Moore likewise found a similarly worded claim “limited to one that is configured to perform certain functions in a particular fashion” and, based on one of the flowcharts, suggested that the claims demanded a dizzyingly long and complex algorithm. And Judge Linn concluded that, while they may be based on an abstract idea, “the claims here are directed to very specific ways of doing that.”
The common thread among all of these opinions is an assumption that, given the heavy use of technical language in the specification and claims, only a specific, complex, technical computer program could infringe the patents. As the seven-line computer program on the previous page demonstrates, this assumption was in error.
The brief goes on to suggest that the Supreme Court bar courts from repeating some of the Federal Circuit's mistakes. Judges shouldn't get lost in "obfuscatory language" from patent specifications and should instead decide based on claim language alone if a patent should be allowed, the brief says. As for the idea that Alice's "system claims" survived whereas its "method" claims could not, such differences are a "drafting formality," particularly for software, and shouldn't be relied on to decide patentability.

How to kill a patent made easy: The importance of Section 101

Four years after Bilski, the Supreme Court again seems poised to limit patentable subject matter, and it has again chosen to hear a case regarding financial patents. Dozens of companies have filed their input on the issue in the form of amicus briefs. There are two key themes: the harm caused by "patent trolls," and perhaps more importantly, what legal tool is most appropriate for eliminating bad patents.
Several briefs delve into the importance of Section 101 of patent law, which bars "abstract ideas" and "laws of nature" from being granted patents. It's not the only way to kill a bad patent; Section 102 dictates that patents must be new (novel), Section 103 states they must be non-obvious, and Section 112 requires that they be claimed in a specific way. But amicus briefs from companies that find themselves constantly facing patent troll lawsuits—including Internet companies large and small, retailers, restaurants, and supermarkets—want a Section 101 test for patents that requires them to be tough and come early.
Having a tough Section 101 test up front would undoubtedly be a boon to patent defendants. A retailers' brief notes that treating the basic patentability question of Section 101 first "is not a distinction without a difference." From the brief:
[A]nticipation, obviousness, and the other invalidity defenses found in Sections 102, 103, and 112 generally require extensive, expensive discovery, prior art, searching invalidity analysis, claim construction, expert reports and depositions, and the other accoutrements of patent litigation before the district court can address them, either on summary judgment or at trial. In contrast, patentability under Section 101 can—and should—be addressed as a legal issue at the outset of the lawsuit on a motion to dismiss.
High-stakes patent cases, considered those with more than $25 million at issue, typically cost $3 million through discovery and $5 million through trial. Compare those costs to recent cases where defendants have won on Section 101 grounds, such as FindTheBest's win against Lumen View Technology, a patent troll wielding a patent on "multilateral decision-making." That patent was found to be abstract, and FindTheBest didn't have to go through discovery or hire expensive experts. (FindTheBest still paid about $200,000 in legal fees.)
Conversely, a well-known case last year in which a defendant failed to win on Section 101 grounds meant that a patent holder was able to push forward with a lawsuit over watching ads online. (The Electronic Frontier Foundation, dismayed by that outcome, wanted the Supreme Court to take up that case rather than Alice v. CLS.)
A brief filed by larger Internet companies, including Google, Amazon, Facebook, Netflix, and Rackspace, also pushes Section 101 vigorously and quickly. Making Section 101 analysis wait for formal claim construction is a "novel procedural barrier," inappropriately championed by the Federal Circuit, the companies argue. 
Other parties want to see Section 101 used less, not more. The American Intellectual Property Law Association, an IP lawyers' group, bemoans the "current undefined rubric of Section 101" being "repeatedly invoked to deny patent claims." And IBM, the nation's largest patent-holder, argues that many of the court's concerns would be better addressed under Section 103, which bars non-obvious patents. 
Several briefs also emphasize that abstract patent language that simply adds a general-purpose computer to a claim—no matter how detailed—is still abstract. The Microsoft-Adobe-HP brief goes so far as to suggest that the Alice patents are scarcely computer technology at all. The technology at issue is simply "Bilski on a computer," those companies suggest.
Oral arguments today may signal which of these arguments are gaining traction with the Supreme Court justices. The court is clearly dismayed that such patents, many wielded by patent trolls, are proliferating. The question remaining is how aggressive they will be in treating the problem. A decision in the case is expected by June.

No comments:

Post a Comment