RPX: Taking on the patent trolls

Once an arcane feature of an arcane area of the law, patent trolls — or, more politely, “non-practicing entities,” or NPEs — have jumped to the top of the nation’s political, legislative, and judicial agenda. These are the companies that, while selling no products or services of their own, account for most patent litigation today. In their most controversial form, they purchase patents on the open market and then assert them against companies that do sell products or services, like AT&T, demanding licensing fees and, very often, suing to get them.

In the past two years the U.S. Justice Department, the Federal Trade Commission, the Patent and Trademark Office, and a White House task force have all launched studies of the growing NPE phenomenon. There are three important NPE-related cases awaiting argument before the U.S. Supreme Court this term and five major NPE-related bills pending in Congress. In January, President Barack Obama even trumpeted the need for patent reform to curb “needless litigation” — a veiled reference to trolls — in the State of the Union address.

The numbers generating all the attention are stark. AT&T is no anomaly. Google was hit with 43 NPE suits last year; Verizon, 42; Apple, 41; Samsung and Amazon, 39 each; Dell and Sony, 34 each; Huawei, 32; BlackBerry, 31. Every brand on this unenviable top 10 list was sued by an NPE at least once every 12 days.

NPEs sued more than 4,800 defendants last year, or — if one counts repeatedly sued defendants only once — more than 2,600 different companies. Last year they brought nearly six times as many lawsuits as they did in 2008, against 60% more companies, and thereby accounted for 67% of all patent suits filed — up from about 26% six years earlier.

At the same time, NPEs have their champions. They argue that giant tech corporations routinely pilfer innovations dreamed up by independent inventors and that NPEs simply give these powerless individuals the financial support and litigation muscle they need to vindicate their rights. NPEs therefore serve not only small inventors, the argument continues, but also society at large by preserving the incentive systems that our Founding Fathers wrote into the Constitution to ensure that the Thomas Edisons of the world would be motivated to provide the rest of us with the maximum possible benefit from their genius.

Still, the sheer numbers have many people skeptical. Is AT&T really stealing breakthrough ideas from various Edisons at a rate of more than once a week? Or might someone be gaming a flawed system, bringing dubious claims in favorable plaintiffs venues (40% of NPE defendants last year were sued in the populist Eastern District of Texas) and harvesting nuisance settlements?

This article is about a man who stands directly in the crossfire between NPEs and their corporate targets but claims to take no side in the combat. He co-founded and runs a public company whose raison d’être is to sidestep the hostilities, to defuse them, and, in the long term, to obviate them.

John Amster, 45, runs a company called RPX. (The “RP” stands for rational patent.) RPX seeks to provide a market-based solution to the NPE phenomenon. (RPX, whose statistics have been relied on by academics and government agencies, is the source of the statistics cited above.)

RPX is what’s known as a defensive aggregator. While it’s not the first or only such outfit, it’s the largest and most ambitious and the one that stands the greatest shot of making a dent in the NPE phenomenon. In exchange for an annual subscription fee, paid at the moment by 168 member companies, RPX preemptively buys up dangerous patents on the open market before NPEs can get their hands on them. Alternatively, when NPEs do sue RPX subscribers, RPX tries to wrangle settlements for its members at a “wholesale price,” as Amster puts it, leveraging the fact that RPX can strike deals on behalf of multiple members at the same time.

RPX’s current members include such giants as Apple, Amazon, Cisco, Dell, eBay, Google, Hewlett-Packard, HTC, IBM, Intel, LG, Microsoft, Oracle, Samsung, Sony, T-Mobile, and Verizon. Some members instruct RPX not to identify them, but several of those — like Apple and Amazon — have nevertheless been outed by stock analysts. (Apple and Amazon did not respond to inquiries seeking comment.)

Though tech companies were RPX’s first subscribers, any firm that sells products or services over the Internet wears a bull’s-eye on its back these days, so companies like Bank of America, Best Buy, Crate & Barrel, Target, and Wells Fargo have also signed up in recent years.

Founded in 2008, RPX has spent more than $500 million purchasing rights to more than 3,800 patents, which it pledges never to assert against anyone. It has bought its clients 430 dismissals from 60 litigations brought by NPEs, and it claims — though the figure is necessarily speculative — to have averted more than 2,000 suits against members by taking dangerous patents out of circulation before an NPE could snatch them up.

RPX is still a young company and is hardly a panacea for anyone at the moment. While it spent about $125 million last year buying patent rights for its members, it operates in a world in which, by RPX’s calculations, NPE litigation costs corporations roughly $11 billion annually in settlements and attorney fees — or almost 100 times RPX’s spending budget.

Even so, Amster has grand hopes for his company. “We think there can be a clearinghouse in this market that can be really quite big and efficient,” he says. “If every company just decided, ‘We’re going to have a line item in our budget for patents and patent risks, and that line item is going to be the RPX rate card’ — i.e., RPX’s subscription rate, which ranges from about $85,000 to as much as $7 million or more per year, depending on a member’s net operating income — “we’d have a couple billion dollars to spend. And we’d be able to probably reduce the risk [of NPE patent suits] by 85% to 90% and do it very profitably.”

Purists at either end of the spectrum scorn RPX’s approach. Some in the anti-troll camp revile RPX as just another breed of troll or, at best, a company that could easily metamorphose into one if its current business model fails to produce satisfactory results for shareholders.

On the spectrum’s other end, one NPE, Cascades Computer Innovation, brought an antitrust suit against RPX in 2012, accusing it of acting as a monopsonist — a buyers’ monopoly — that collusively drives down the value of the settlements to which NPEs are entitled. RPX has denied collusion or any other wrongdoing. (In December a federal district judge in Oakland permitted Cascades’ suit to survive a motion to dismiss.)

But whether or not RPX proves a viable solution to the NPE phenomenon, Amster’s story is instructive, for he lives in a world that most of us didn’t know existed.

In Amster’s world, patent suits against tech companies no longer turn on quaint moral issues like, Did Company X steal from Inventor Y? For the most part, they don’t even turn on legal issues like, Is patent Z valid, and if so, is it infringed by Company X’s product?

What counts today are probabilities, statistics, and, most of all, transaction costs. As a White House task force noted in February, a single smartphone today may implicate 100,000 patents, and the scope and validity of any one patent, academics have calculated, is so uncertain that roughly half of them are tossed out when tested in court. In her recent book Rethinking Patents, law professor Robin Feldman of the University of California Hastings College of the Law has suggested that patents should no longer be likened to land titles, with ascertainable metes and bounds. “A patent does not grant a definitive set of rights,” she writes. “Rather, it is an invitation to bargain over the definition of those rights.”

This world is one that, in Amster’s view, won’t change much even if many of the reforms now being bruited about take effect. While certain NPEs are certainly using abusive practices, he says, the problems his clients face run deeper.

“The really nuisancy stuff — the courts sort those things out,” he says. “People don’t pay a lot of money for them.”

The intractable problems stem from “shades of gray,” he says. Most patents involve “little-step improvements,” not breakthroughs, which are nevertheless patentable. “Maybe it isn’t teaching something that’s super-revolutionary,” he says, “but it’s teaching enough that’s different, and they came up with it first, and they published it.” In addition, the fact that a company may have designed and marketed a product from start to finish without ever having heard of a patent that its product is now accused of infringing is no defense under our system.

No judge or legislator is proposing the tectonic changes to our patent system that would be required to alter this status quo. Nor are Amster’s clients, who treasure their own patents. “You’d risk having a completely impotent patent system,” Amster explains.

So Amster is trying to build a patent clearinghouse that will lift the burden of NPE litigation from tech companies’ backs while enabling legitimate inventors to get paid quickly and reasonably. In navigating toward this goal, Amster declines to call NPEs trolls and strives, on the contrary, to maintain good relations with them, even though this sometimes nettles his clients.

“Yes, we’re friends with them,” Amster says he tells those clients. “Yes, we take them out to dinner. We do business with them. And as a result, we save you a lot of money.”

Personable and earnest, Amster is the son of a well-regarded New York patent defense lawyer. Upon graduating from New York’s Cardozo law school in 1995, Amster defended some patent cases himself at the New York law firm Weil Gotshal & Manges. But litigation was not his calling, he says, and after transferring to the firm’s Silicon Valley office, he began turning to corporate transactional work on mergers and acquisitions.

By 2003, Amster had found his way to Ocean Tomo, then a newly launched company that specialized in patent valuation, and opened an M&A office for it in San Francisco. There, at age 36, he became involved in an event that helped shape the patent secondary market as it exists today.

In October 2004 a dotcom-bubble company called Commerce One, which once had 4,000 employees and a market cap of $21.5 billion, filed a prepackaged bankruptcy calling for one of its creditors to buy all its assets for just $4.1 million. Its general counsel, Paul Warenski, then got the idea of carving out the company’s patent portfolio and selling it separately. It quickly fetched a $1 million offer from Intellectual Ventures, the company co-founded by polymath and former Microsoft CTO Nathan Myhrvold.

IV was controversial. The year before, it had begun buying up patents — including many from imploded dotcom companies — using a war chest provided in part by strategic investors, including companies like Microsoft, Intel, and Apple. Those investors saw IV as performing a defensive purpose — keeping those patents out of the hands of hostile NPEs — but also hoped to earn a return because IV would seek licensing revenue on the patents it was acquiring. At the time, Myhrvold insisted that if he priced his licensing demands correctly, he would never need to sue anyone. Still, others thought Myhrvold’s business model would eventually force him to do so. IV was, after all, set up like a hedge fund, promising fat returns to investors while its management generally skimmed off 2% management fees and 20% of the profit.

At the suggestion of a friend, Warenski called Amster, who persuaded him to turn down IV’s offer and instead hire him, acting for Ocean Tomo, to market the portfolio more broadly.

Ironically, the inventor of the key patents in question, Robert Glushko, was horrified by what was happening. The whole point of his inventions was to facilitate “an open, frictionless infrastructure for e-commerce,” says Glushko, now a professor at the University of California at Berkeley’s School of Information. The prospect of a troll owning his patents would have been catastrophic. “Tollgates were antithetical to why we did the whole invention!” he says. He begged friends at IBM, Sun, Oracle, and Microsoft to buy the patents defensively, he recounts.

In December 2004, Amster led an auction among eight bidders, including IV, at the U.S. Bankruptcy Court in San Francisco. On the 51st bid, the portfolio sold for $15.5 million — more than triple what the entire company had first been slated to sell for. The winner was a mysterious outfit called JGR Acquisition. Glushko was depressed for days, he recalls, convinced that it was a troll.

But six months later the New York Times reported that it was actually owned by Novell Corp. Novell, in turn, as Fortune now reports for the first time, had been serving as a proxy for a defensive patent-aggregation consortium that didn’t exist yet but that was already under formation. In an interview for this article, Jerry Rosenthal, then head of intellectual property for IBM, explains that Novell was acting for the inchoate Open Invention Network — consisting of IBM, Novell, Philips, Red Hat, and Sony (later joined by NEC and Google) — which launched formally in November 2005 with Rosenthal as its first CEO. OIN buys patents to protect users of the Linux open-source operating system from patent aggression.

Remarkably, the Commerce One auction also catalyzed into existence a second defensive patent consortium. It was midwifed by a bidder called ThinkFire, which had been founded in 2001 to advise corporations on how to monetize their own patent portfolios. ThinkFire was then headed by its co-founder, Dan McCurdy, a former head of intellectual property licensing at Lucent. It was bidding on behalf of a group of its clients, McCurdy says, who feared that IV would otherwise grab up the patents.

“‘We don’t want IV to know we’re interested,'” McCurdy’s clients were telling him, he says. ” ‘That would clue them to come knocking on our doors.'”

Ironically, ThinkFire’s other co-founder, besides McCurdy, was Myhrvold himself. Myhrvold was actually still ThinkFire’s chairman while ThinkFire was bidding against IV at the auction — a spectacle the press gleefully highlighted. (Myhrvold resigned from ThinkFire’s board about a week later.)

After the auction, McCurdy continues, his clients decided to formalize their relationship so they’d be ready to move quickly the next time a dangerous patent portfolio came up for sale. That led to the formation of Allied Security Trust, he says, which launched in January 2007 and which McCurdy now heads. AST’s goals are similar to what RPX’s later became, though AST is smaller and not for profit. Basically, AST alerts its members to patents that become available on the open market, and if any clients want to put up money to buy one, he’ll make a joint bid for it on behalf of those members. If AST wins the patent, it licenses it to the bidding members and then sells it back into the open market — including to NPEs — returning any money collected to those who bid. This business model, known as “catch and release,” prevents free-riding and guarantees that AST can never itself become a troll because it retains no patents to sue upon. AST today has 26 members, most of which are also members of RPX, according to McCurdy.

The Commerce One auction had one other lasting impact on the patent secondary market. Within a month of its completion, IV offered Amster a job, and he took it.

At the time, Amster says, he saw IV as a benign clearinghouse that enabled companies to clear rights and inventors to get paid without litigation raising its ugly, inefficient head. But three years later, when Intellectual Ventures was gearing up to launch a new fund, Amster had become leery, he says. In January 2008 he quit. (IV did not start suing companies until December 2010. Since then it has sued more than three dozen companies, and its licensing targets have notoriously included at least one of its own early investors, Xilinx. Today many people consider IV to be the largest patent troll in the world. It acknowledges owning 70,000 “patent assets” — domestic and foreign patents and patent applications — and actively seeking to enforce about 35,000 of them.)

Upon quitting, Amster bandied new business ideas with Geoffrey Barker, a friend from IV who quit the same week he did, and Eran Zur, a patent broker with close ties to the NPE world. Zur had once worked for the Lemelson Foundation, the richer-than-Croesus licensing arm of controversial inventor Jerome Lemelson, who had more than 600 patents to his name when he died in 1997. (Over a 14-year period beginning in 1990 the foundation collected nearly $1.5 billion in settlements from a set of bar-code and machine-vision patents that were all eventually invalidated by an appeals court in 2005.)

What emerged was the idea for RPX. The three founders would pitch it initially to tech companies with NPE problems. The idea was “to go do good,” says Amster. “To build a business that can lower your costs, that’s 100% aligned with you, that’s never going to sue anybody. Basically do the technology world a big favor by trying to create a clearinghouse in the patent space.”

It was an effective pitch. Notwithstanding the looming financial crisis, RPX launched in August 2008 with VC backing from Charles River Associates and Kleiner Perkins Caulfield & Byers, and by the end of the year it boasted such blue-chip clients as Cisco, IBM, Seiko-Epson, and — according to a Cowen and Co. analyst report — Apple. Seventeen more clients joined the next year, propelling RPX toward its current complement of 168. In May 2011 it went public. (Amster was given options valued at $3.2 million that year. In 2012 his salary was $632,000.)

“Look at the data,” Amster says. “Look at the patterns. Quantify your decisions.”

These are Amster’s mantras. RPX is rich in data, thanks not just to today’s digitized public court dockets, but also to RPX’s growing stable of clients who share mounds of confidential cost data with it.

When Amster pitches his company to a potential new client, he often mentions a patent portfolio relating to software encryption that was being shopped by a broker in mid-2008, when RPX was just launching.

“They wanted $2.5 million,” Amster recounts. “We would’ve bought it if we’d had any money, but we hadn’t been funded yet. So we missed it.”

The patent holder, an NPE called PACid, “then hired a contingency counsel, and over the course of the next couple of years filed six separate suits, and they hit 117 defendants in total,” Amster continues, including Apple, Cisco, and Microsoft. PACid eventually collected $20 million in settlements — or just under 10 times the portfolio’s asking price — but not before the defendants spent an aggregate $58 million in attorneys’ fees. So a patent that could have been snatched up for $2.5 million ended up costing defendants $78 million, of which 74% went to defense attorneys. If one factors in the plaintiffs attorneys too — who probably got at least 30% of the $20 million in settlements — 82% went to attorneys.

So the most efficient use RPX can make of its clients’ money is to buy problematic patents before litigation begins. That way, says Amster, “the problem goes away for a lot cheaper.”

How does RPX know what to buy? Often a broker selling a patent portfolio will include in the offering materials a “claims chart,” Amster explains, showing how specific paragraphs in the patents allegedly “read on” — i.e., are being infringed by — products being marketed by identified corporations. Those are the companies that are most obviously in the cross hairs.

“So, for example, when we look at a reputable patent broker who’s got a track record of selling to NPEs,” says Amster, “to us it’s 100% clear that if that guy has a patent and it’s a decent patent and it’s got claim charts, an NPE’s going to buy it. It’s a litigation waiting to happen.”

Unfortunately, Amster concedes, it’s often hard to convince a prospective client that it needs to buy a patent when it hasn’t yet even received a demand letter.

In practice, the easier way to recruit clients is when they have already been sued.

By then, however, the patent has usually become too expensive for RPX to buy anymore.

“If there’s 20 defendants in a litigation, and 10 of them are our clients,” Amster explains, “it’s usually the case that we can’t justify, based on our valuation algorithm, buying the whole [patent portfolio]. So we’re much more likely to buy the [licensing] rights for our 10 clients, and options for the other 10.”

By buying options, RPX can then go to the remaining defendants in the case and offer to buy them out of the litigation if they will join RPX. In late 2008, for instance, an NPE called IPAT sued 23 companies in Marshall, Texas, on a patent portfolio relating to computer security. By late 2009 RPX had reached settlements for about half the defendants, who were already RPX members, and it then enticed about another 10 into joining in exchange for buying them out of the case.

One holdout defendant, the Russian software firm Kaspersky Lab, declined RPX’s invitation, however. Instead, it reported RPX to the FBI. In its letter of accusation, which is available on the Internet, CEO Eugene Kaspersky alleged that RPX — which was offering his company a three-year membership at $160,000 per year — was conspiring with IPAT to commit extortion.

According to Amster, the FBI never contacted RPX about Kaspersky’s letter, which Amster characterizes as “frivolous” and “bizarre.” Kaspersky went on defending the IPAT case, and in June 2012 won a dismissal. He hailed his victory in a press release, stressing the need to fight unmeritorious cases and arguing that all the other defendants had forked over money needlessly to IPAT or RPX. But in response to Fortune’s inquiries, a Kaspersky lawyer acknowledges in an email that the company spent about $2.5 million fighting the IPAT case (compared with $480,000 for the three-year RPX membership the company turned down). Meanwhile, Kaspersky Lab has been drawn into three more NPE litigations. It won a dismissal in one and is still litigating the others.

Those who share Kaspersky’s perspective — that RPX is just another breed of troll — also focus on another occasional practice of the company: “divestiture.” Once in a while RPX will sell its patents — after licensing them to all its members — back into the open market, including to NPEs. (This is the same procedure that the defensive patent conglomerate Allied Security Trust follows universally and calls catch and release.)

RPX has done this fewer than 10 times since its inception, according to Ned Segal, who left Goldman Sachs to become RPX’s chief financial officer last year. “But we’re doing them more frequently now,” he acknowledges. “Our clients want to see us recycle their capital to go buy more patents for their benefit as opposed to letting their competitors free-ride on their capital.”

The practice is used in part, Amster admits, to drive prospects into the network that otherwise feel no need, because RPX, at the behest of their competitors, has already purchased the portfolios that most threaten their industry sector. Such prospects sometimes claim to see no value in the portfolios RPX has already purchased.

“We call them back a few months later,” Amster says, “and say, ‘Look. We put it up for sale. There’s an NPE who’s willing to pay us for it.’ And sometimes they say, ‘You’re no different than they are’ [likening RPX to the trolls]. And we say, ‘Hey, wait a minute. Back up. We bought this on behalf of our clients. You told us there was no value to it. The market has spoken, and there is value. Somebody is going to sue you. We didn’t call them and tell them to sue you. We said, ‘Here’s an asset. It’s for sale.’ If you don’t think it’s a risk, then fine.”

“For the most part,” Amster continues, “it’s worked out that somebody goes, ‘You know what? You’re right. There is real risk in this crap, even if it’s crap.'”

This story is from the March 17, 2014 issue of Fortune.

Editor’s Note: The magazine version of this article contains certain charts that we were unable to reproduce here for technical reasons. As those charts disclose, certain NPE statistics referenced in the text of this article were impacted by the passage, in September 2011, of major patent reform legislation, the America Invents Act.