It's the case of Symbol Technologies/Cognex Corporation v. Lemelson Medical, Education & Research Foundation. There is an AP article posted on msnbc.com and pretty much all over the Internet about the case. It explains just exactly how the patent system works, or doesn't, depending on your point of view. I find it satisfying to know that all over America, in Kansas and in NYC, thanks to AP, folks are reading all about submarine patents this morning.

It is a story about Jerome Lemelson on one side, an inventor, a prolific one, with nearly 600 patents to his name. He didn't actually build his inventions; he just thought them up and described them at length in patent applications, and that may explain the number. Back in the fifties, he filed patents for an early machine vision/bar code system. You could call scanners at the supermarket an example of such a system. Today they are everywhere, but back then, they were just in his head. That's impressive in itself, but is it enough for a protectable patent?

What made Mr. Lemelson controversial was his way of sitting on a patent by means of tweaking it, thus prolonging its life and keeping it secret, until others had businesses that actually used similar technology, and then using the threat of patent litigation as a club to get license fees. Pay or I'll sue, and then you'll be sorry. And after watching the discovery process in the SCO litigation, does it amaze you that companies by the hundreds caved and paid rather than endure prolonged litigation? At least, that's my best understanding of the strategy. I gather he more or less brought submarine patents to their pinnacle of expression. Since we are trying to learn how patent law works, let's use this case, which is more interesting than most, to get a handle on it.

Here's an explanation of the expression paper submarine patent:

The patents were "paper" because Lemelson did not practice his inventions. In order to obtain a patent, there must be conception plus reduction to practice. Reduction to practice, however, does not require that the applicant have actually built a prototype or used the method. Reduction to practice can be "constructive" -- by including in the application a written description of the invention sufficient "to enable any person skilled in the art to which it pertains...to make and use the same." 35 U.S.C. § 112, 1.

The patents were "submarine" because they remained pending in the Patent & Trademark Office until the technology allegedly covered by the patents came into widespread commercial use, at which time Lemelson "surfaced" the patents (by allowing them to issue) and then collected royalties.

Now a patent filed in the fifties would normally expire by now, so I couldn't at first figure out how there could be litigation over 1954-1956 patents. This Boston.com article explains:

Ordinarily Lemelson's patent would have long since expired, but Lemelson filed a series of patent continuations that extended the life of the original patents and added 16 others, which were issued during the late 1980s and early 1990s.

It seems Mr. Lemelson would file a "continuation in part" application which added additional drawings and text to the applications, in order to extend the time. Meanwhile companies -- not aware of his patents -- did similar types of inventing and built companies around their technology. Then, finally, the submarine patent would surface. The law on patents was changed in 1995 in such a way that this kind of patent continuation strategy makes less sense today.

According to an article [PDF] in Fortune magazine, "The Patent King," made available by Cognex Corporation, a machine vision company, the US Patent Office examiners called Lemelson Black Box Jerry, because his incredibly long applications were so lacking in technical detail. The Fortune article's author sums up the criticism of the patents by analogizing to Leonardo Da Vinci. Should Boeing have to pay the Da Vinci estate because Da Vinci dreamed up a flying machine long before it could be actually manufactured by anyone? Critics of Lemelson say that is the level of some of his "inventions." But hundreds of companies paid to license them nonetheless. Lemelson had a long history of successfully suing corporations for violation of his patents, large ones too, who mostly paid up rather than fight:

Critics charge that for decades Lemelson manipulated the U.S. Patent Office. They accuse him of exploiting loopholes that forced 979 companies — including Ford, Dell, Boeing, General Electric, Mitsubishi and Motorola — to pay $1.5 billion in licensing fees.

Ford, for one, had fought for a bit, but when it lost on the District Court level and the appeals court wouldn't grant them an interlocutory appeal, they folded too and settled. A lot of companies ponied up for licenses after that. AP on Yahoo's article tells us more about how Lemelson's attorney, Gerald Hosier, eventually chose the wrong targets, the end users of machine vision and bar code technology:

With the licensing campaign in full swing, Hosier and Lemelson decided not to go after the smaller companies that produced the equipment, such as Cognex or Symbol Technologies, of Holtsville, N.Y., which made the bar code scanner.

Instead, they focused on the large corporations with deep pockets, customers of Cognex and Symbol that made use of the technology.

Getting them to pay, Hosier said, was based on a simple premise. “This business is not based on what’s right or what’s wrong,” Hosier said. “It’s based on fear. Nobody would pay you for a patent unless they feared that the consequences of not paying you vastly exceeded the consequences of paying.”

You see? This is the heart of the patent problem. Law is indeed supposed to be about right and wrong, and when the system can be gamed so that right and wrong isn't in the picture any more, it's time to reevaluate the law.

When Lemelson died, his patents went to a not-for-profit foundation, and it was charged with collecting the license fees and going after infringers. They have a page on the litigation too. There's a picture of Mr. Lemelson on the foundation's website.

When its attorneys eventually decided to sue clients of Cognex Corporation, accusing Cognex's customers of infringing the patents, the line was drawn in the sand. Suing end users instead of the manufacturer or vendor is like SCO suing AutoZone while crossing its heart and hoping to die before the judge in Delaware that it never had a thought of suing Red Hat.

Some bar code reader manufacturers, including Symbol Technologies, decided to bring a declaratory judgment action against the Lemelson Foundation, and when Cognex saw its clients being sued, the founder and CEO, Robert Shillman -- who had graduated from MIT and who understood machine vision technology -- decided to do something too, even though his company was not being threatened directly. Eventually, the two litigations were combined, challenging the validity, enforceability and infringement of Lemelson's "bar code" and "machine vision" patents.

This article tells why Shillman decided to sue:

Against the advice of his most trusted associates, Shillman decided to go forward with a lawsuit in September 1998. The suit, filed in Massachusetts, said the Lemelson patents were invalid, unenforceable and not being infringed by Cognex.

"If evil is being acted upon in the world, everybody has an obligation, especially if only you had the power. Only we understood the technology. Only we could defeat this guy," said Shillman.

Here's a bit from the 1998 press release Cognex put out about their decision to file suit:

"No claim is now, or has ever been, asserted against Cognex by either Lemelson or by the Lemelson Foundation…despite the fact that Cognex has for many years been the world's leading supplier of machine vision systems. However, for several years, hundreds of companies around the world, many of whom are Cognex's customers, have been faced with assertions of infringement by Lemelson simply because of their use of machine vision systems in their manufacturing operations," stated Dr. Robert J. Shillman, Cognex's founder and C.E.O. "The Lemelson patents have no bearing on Cognex's state-of-the-art machine vision systems, and, furthermore, we have concluded that the patents are invalid, unenforceable and not infringed by us." Dr. Shillman continued, "Cognex has served this complaint to defend our current and future customers against Lemelson's demands, and we will do whatever is necessary to protect their legitimate rights to use our products and our technology."

Founded in 1981 by three graduates from the Massachusetts Institute of Technology, Cognex pioneered the development of the machine vision industry. Cognex systems use Cognex's proprietary image analysis software together with special-purpose high-speed hardware. "Our technology…more than 1,000 man-years worth over the last 17 years…is the result of our own internal efforts carried out by our large engineering staff which currently includes over 150 engineers, of which 23 have a Ph.D. degree," stated Dr. Shillman. Cognex currently holds 30 patents and has more than 130 patents pending, all covering innovative advances in machine vision technology.

Here's the complaint [PDF]. Cognex wasn't being sued, because it was too small. In fact, if it lost the case, it likely would go bankrupt. But it didn't lose. Lemelson did:

The court found that 14 critical patent claims by Lemelson were unenforceable under a rare defense called prosecution laches — an unreasonable delay or negligence in pursuing a right.

The judge also said the claims were invalid for lack of a written description, and a person of ordinary skill could not build the inventions using Lemelson's patents as his own experts had asserted. He added that "Symbol and Cognex products do not work like anything disclosed and claimed by Lemelson."

The companies did not demonstrate that Lemelson had "intentionally stalled" getting the patents, the judge ruled. But he did say that "decades of delay preceded the assertion of patent claims, and Lemelson has offered no adequate explanation for that delay."

Some of the claims asserted by Lemelson in this case will not expire until 2011, fifty-five years after the 1956 application was filed and forty-eight years after the application issued as a patent. The evidence adduced at trial is abundant that during that period, machine vision and bar code technology was developed by many who had never heard of the Lemelson patents. If the defense of prosecution laches does not apply under the totality of circumstances presented here, the Court can envision very few circumstances under which it would. To conclude otherwise would remove from the public domain subject matter arguably disclosed in Lemelson's applications, but not timely claimed in a patent, and by any meaningful standard would unreasonably delay the time when the public would be free to use Lemelson's claimed inventions.

March 21, 2000 – In a series of pre-trial motions, U.S. District Court Judge Philip M. Pro consolidates the Cognex and Symbol Technologies lawsuits into a combined case -- Symbol Technologies, et al.

In addition, Judge Pro dismisses Cognex’s prosecution laches defense from being included in the trial on the grounds that the issue had been resolved in Ford Motor Company’s 1997 lawsuit against Lemelson.

September 1, 2000 – U.S. Court of Appeals for the Federal Circuit grants Cognex’s request for an interlocutory appeal to decide whether the prosecution laches is an appropriate defense to bar enforcement of patent claims.

January, 2002 – The U.S. Court of Appeals for the Federal Circuit grants Cognex the right to raise the doctrine of patent prosecution laches as one of its defenses to the Lemelson Partnership’s infringement claims.

The District Court judge thought he couldn't accept such a defense as a matter of law, because of the Ford decision, because Ford had unsuccessfully tried to raise that very defense.

Another case this judge could have thought of is Advanced Cardiovascular Systems v. Medtronic (N.D.Cal. 1996) 41 U.S.P.Q.2d 1770, where Medtronic's affirmative defense was that the patent was unenforceable due to "unreasonable, improper and undue delay" in the prosecution of the patent, but the court disagreed, essentially saying that if a party has followed all the rules, then how can it be too long? They struck down that defense for Medtronic:

Furthermore, since there is no allegation that plaintiff violated any of the statutory or regulatory rules for prosecuting patents, there is no improper delay in the prosecution of this patent to justify a laches or inequitable estoppel defense. The '346 patent issued from a series of "continuation" applications relating back to the original application filed on January 6, 1987. Under 35 U.S.C. Section 120, the prosecution of this patent dates back to the filing of the original application. [citation] By providing this relation back doctrine, Congress evidenced a clear intent to regulate the timing of continuation applications. Accordingly, only Congress can determine what constitutes unreasonable delay in the filing of such an application. It is not for this Court to decide that the prosecution of a patent according to the rules of the PTO is unreasonable and inequitable. Since defendant's laches and equitable estoppel defenses ask the Court to make precisely this determination, the Court strikes these defenses with prejudice. Id. at 1774-75.

So Shillman and the others had a point in time when they too could have folded. Instead, they were willing to pay to appeal the issue, and that issue is eventually what led to the victory. Lemelson took that interlocutory issue all the way to the Supreme Court, by the way, unsuccessfully. Here's another article [PDF] about the case and about the attorney representing Symbol and who earlier had represented Ford, Jesse Jenner, and there is a picture of Mr. Jenner as well. I gather from the article that Mr. Jenner was involved, if not the originator, of the idea of developing the defense of prosecution laches:

Jenner decided to attack the Lemelson patents with the doctrine of prosecution laches. The theory held that Lemelson took an unreasonable amount of time in actively obtaining patents, rendering them unenforceable. Lemelson had submitted his original patent applications for the technology in 1954. His first patent issued in 1963. But Lemelson was able to keep receiving related patents well in to the early 1990s. Lemelson would file so-called continuation applications that kept his claims alive in the patent office. He would amend his applications, according to Jenner, based on technological breakthroughs made by others. At the time, a patent lasted 17 years from the day it was issued. So Lemelson was able to start demanding royalties decades after he had originally applied for a patent.

Jenner believed Lemelson was manipulating the patenting process. “There was something wrong with the way Lemelson seemed to have gamed the system,” says Jenner. “He’d watch what real companies were doing, and craft these claims. Then 30 years later these submarine [patents] would pop out of the water.”

Here's a bit from Cognex's press release when they won at the District Court level in January of 2004:

In summarizing his 30-page decision, which was issued late Friday, Judge Pro wrote, "Having concluded that Lemelson's patent claims are unenforceable under the equitable doctrine of prosecution laches; that the asserted patent claims as construed by the Court are not infringed by Cognex because use of the accused products does not satisfy one or more of the limitations of each and every asserted claim; and that the claims are invalid for lack of written description and enablement even if construed in the manner urged by Lemelson, the Court finds that Judgment should be entered in favor of Plaintiffs."

Since the early 1990s, the late Jerome Lemelson and the Lemelson Partnership…which never built or sold a single vision system or bar code reader…have collected in excess of $1.5 billion in license fees by asserting their patent portfolio and the threat of complex and costly patent litigation against hundreds of companies around the world that use machine vision or bar code readers. As a result of Friday's ruling, every company will now be able to use machine vision systems and bar code readers, anywhere in their operations, without the threat of litigation from the Lemelson Partnership, and without having to pay licensing fees to them.

"This ruling is truly a cause for celebration…for Cognex and for every company around the world that makes, sells or uses machine vision systems or bar code readers," said Dr. Robert J. Shillman, Chairman and CEO of Cognex. "It is also a victory for consumers everywhere, because every one of us has paid a hidden tax to Lemelson each and every time that we made a purchase of virtually any item that was manufactured since the start of Lemelson's licensing onslaught."

Dr. Shillman continued, "Much of the litigation in our society is motivated by greed, but that was not Cognex's reason for suing Lemelson. Cognex has spent millions of dollars on this case, and even though we won, we won't receive a single cent from Lemelson. We pursued this litigation because we knew that what Lemelson and his Partnership were doing was terribly wrong, and only Cognex had both the technical knowledge and the fortitude to put an end to their campaign of legalized extortion. Our only reward is the knowledge that we did the right thing…we stood up to wrongdoing and we triumphed. Our success in this case is a testament to two of Cognex's core values: Integrity and Perseverance…we do the right thing, and we don't quit until the job is finished."

Lemelson appealed the ruling, and it was heard on June 8, 2005 by the Federal Circuit. You can read a transcript of the hearing [PDF], prepared by the Lemelson side from a recording of the event. On my reading, it doesn't sound like the judges altogether agreed with the District Court decision, although you can't always tell what a judge is thinking, just by his questions. One issue they seem to struggle with was where to draw the line. When is it too long to wait to prosecute a patent? One of the judges asks the Cognex side's attorney that very question in the hearing:

Judge Bryson: My problem is that we have already, at least retrospectively I mean, recognize there's been statutory changes that affect this issue. But retrospectively, we have already bought into a systtem in which you can, as long as the applications are copending, you can come back and effectively farm a disclosure for new claims. And we just haven't had cases in which the period is as long as this. Now my question is, if we are going to say well this is just too long, I have a very hard time finding a handhold on what constitutes the measure of too much and where we say, well that's permissable, but this is not.

Applying the foregoing standard to the preponderance of evidence adduced at trial, the Court finds that Lemelson's 18- to 39-year delay in filing and prosecuting the asserted claims under the fourteen patents-in-suit after they were first purportedly disclosed in the 1954 and 1956 applications was unreasonable and unjustified and that the doctrine of prosecution laches renders the asserted claims unenforceable against Symbol and Cognex.

They also assert that although Lemelson filed the tweaks, what they ended up suing over wasn't in the original patents, and that language was not brought forward in the tweaks, which I gather it should have been. The AP stories made it sound like this case is already decided, so I looked and looked, but I found no ruling yet from the court of appeals, and then I checked with the law firm that won the case, and they don't expect a ruling until the fall. I wish I could say it's all over, but until the last piece is in, it's not, and the court of appeals isn't the highest possible court either. I think the articles are preparing the way for the day the ruling arrives from the appeals court. Here's the page where the US Court of Appeals for the Federal Circuit publishes its decisions, if you want to keep track.

Isn't patent law unbelieveably dry and depressing? I'm trying my best to make it readable and understandable for you, but I must say, particularly when reading the hearing transcript, I feel like Alice in Wonderland trying to make sense of the conversation at the Mad Hatter's tea party. And I certainly wish Microsoft would pick another weapon, so we wouldn't have to learn all this patent stuff, but until they tack and choose a different course, we do need to understand patent law, and heaven only knows, I'm giving it my best.