The last decade was a boom time for patent trolls. Their names and lawsuits made the news; This American Life dedicated two hour-long episodes to them. The number of defendants in patent troll lawsuits increased sixfold from 2003 through last year. But now the tides seem to be turning for them: After growing very rapidly since 2009, the number of lawsuits filed by "non-practicing entities" will be significantly lower this year compared to 2013. Although the level of litigation will still be at a historic high, is this indicative that they are finally being reigned in?

I talked to several patent litigation experts—both legal scholars and industry experts—and all agreed that a series of recent Supreme Court decisions has put a crimp in the patent troll's business model. Two key ingredients of the patent troll's business model are: 1) a litigation process that is very costly for defendants, and, 2) patents that are overly broad or vague so that they can be interpreted to cover commonly used technologies and hence snare many defendants. Given the cost, many defendants are willing to pay the troll to avoid a lawsuit even if the suit is not justified.

These ingredients have provided ample business opportunity for patent trolls. Patent lawsuits can be very expensive to defend, especially because plaintiffs can, at very little cost to themselves, cause defendants to rack up millions of dollars in costs for legal discovery before the plaintiff even establishes what the alleged infringement is. Ever since the appellate court for patent cases (the Court of Appeals for the Federal Circuit) loosened standards for patenting software during the 1990s, the Patent Office has issued tens of thousands of vague, overly broad software patents, including patents that cover methods of doing business using software. Indeed, in a study last year, the Government Accountability Office attributed 89 percent of the increase in patent litigation to software patents.

Now, several Supreme Court decisions offer some help to those being trolled. First, in decisions issued in April (here and here), the Court made it somewhat easier for a defendant to recover its legal costs if the judge sees the plaintiff’s suit as frivolous. Then in June, the Court made it easier to challenge the validity of a vague patent.

Finally, and most importantly, in Alice Corp. v CLS Bank, also decided in June, the Court made it clear that you can’t get a patent just for implementing an existing business method on a computer. Alice Corporation is a patent troll established by Ian Shepherd, a former manager at the McKinsey consulting firm in Australia. In 1999, Shepherd received a patent on the general idea of handling escrow accounts on a computer.

Shepherd’s patent claimed the general idea of communicating these documents over a computer system. But the Supreme Court, following a string of decisions going back to 1972, ruled that the simple notion of “doing it on a computer” was an abstract idea that was not patentable.

The effect of this decision was quickly evident. First, the Patent Office tightened the spigot on issuing business method patents. Last month, they issued fewer than half the number that they had issued during months prior to Alice.

Patents Issued Each Month by Type

Data:  U.S. Patent and Trademark Office

Second, since Alice, courts have tossed out 15 business method patents for being not patentable. These invalidations are particularly important because they show that firms accused of infringing business method patents will often be able to get the suit dismissed before costly legal discovery.

According to Stanford law professor Mark Lemley, this ability to dismiss a suit—along with the greater risk that a troll might have to pay legal costs—have curbed trolls using the “nuisance value” business model. Defendants are also not as willing to settle a patent troll lawsuit, now that the costs might not be so stacked against them—rendering the nuisance business model less profitable. Lemley sees this change contributing to a decline in patent lawsuit filings this quarter:

Patent Lawsuits Filed Per Quarter

Data: Unified Patents​

While the number of lawsuits has been declining since spring of last year, the drop has been especially noticeable since Alice (the third quarter of 2014). Of course, other factors are likely contributing to this decline in litigation. Some experts cite the importance of new procedures for challenging patents at the Patent Office as a way to reduce the cost of litigation. Others cite the stalemate in the smartphone patent wars between Apple and other companies. Most patent lawsuits today, however, are not filed by operating companies, so the drop in litigation after Alice suggests that at least some trolls are hurting.

But don't shed too many tears for the patent trolls just yet. While these changes are a real positive step, they are also limited. While it's true that the Patent Office is granting several hundred fewer business method patents each month, it is still granting record numbers of software patents overall. Most of the litigation is over software patents, not business method patents: One study estimates that business method patents constitute only 10 percent of the patents used in lawsuits by “non-practicing entities” including patent trolls. And while the number of lawsuits has decreased, there are still over four times as many as in 1990, before the surge in software patenting began following court decisions in the mid-1990s. Furthermore, Matt Levy of the Computer and Communications Industry Association told me that in only one case has a victorious defendant recovered legal costs—thanks specifically to the April Supreme Court decisions.

It’s possible that the Supreme Court decisions will eventually have a greater impact on patent trolls than they have so far. For example, the Alice decision pertains to abstract software patents generally, not just to business method patents. It all depends on how the lower courts, particularly the Court of Appeals for the Federal Circuit, interpret and apply these decisions. But that is the very court which has allowed so many broad and vague software patents starting from the 1990s. Some patent experts see legislative reform as critical to really fixing the issue of patent trolls while others believe we should wait to see what the courts do before going to Congress.​ A patent reform bill passed the House last year, but died in the Senate. A new patent reform bill is seen likely in the new Congress.

Lemley thinks Congress should wait to see how the courts work things out: He thinks that they have been moving policy in the right direction for some time, and he is optimistic that they will continue to do so. But others, such as Levy, argue that new legislation is needed to effectively change the economics of patent litigation. He notes that some courts have been particularly friendly to patent trolls. “Without direction from Congress, they will not change what they’re doing,” says Levy.

It seems that both changes—in the courts and the law—will be needed to defeat patent trolls and restore the patent system as an aid, rather than a hindrance, to innovation.