Update: March 1, 2017 Today IBM told Ars Technica that it "has decided to dedicate the patent to the public" and it filed a formal disclaimer at the Patent Office making this dedication. While this is just one patent in IBM's massive portfolio, we are glad to learn that it has declared it will not enforce its patent on out-of-office email.

On January 17, 2017, the United States Patent and Trademark Office granted IBM a patent on an out-of-office email system. Yes, really.

United States Patent No. 9,547,842 (the ’842 Patent),“Out-of-office electronic mail messaging system,” traces its history to an application filed back in 2010. That means it supposedly represents a new, non-obvious advance over technology from that time. But, as many office workers know, automated out-of-office messages were a “workplace staple” decades before IBM filed its application. The Patent Office is so out of touch that it conducted years of review of this application without ever discussing any real-world software.

The ’842 Patent describes technology that would have been stupefyingly mundane to a 2010 reader. A user inputs “availability data” such as a “start date, an end date and at least one availability indicator message.” The system then uses this data to send out-of-office messages. The only arguably new feature it claims is automatically notifying correspondents a few days before a vacation so that they can prepare in advance for a coworker’s absence. From a technological perspective, this is a trivial change to existing systems. Indeed, it is like asking for a patent on the idea of sending a postcard, not from a vacation, but to let someone know you will go on a vacation.

It is worth considering the full prosecution history (PDF) of the ’842 Patent to understand how the patent system reached such an absurd result. There were two big oversights. First, the examiner never considered whether this patent’s software-related claims were eligible under the Supreme Court’s decision in Alice v. CLS Bank. Second, the Patent Office did an abysmal job of reviewing prior art and considering obviousness.

In Alice, the Supreme Court ruled that an abstract idea does not become eligible for a patent simply because it is implemented on a generic computer. That decision came down in June 2014, so the Patent Office had plenty of time to apply it to the application that led to this patent. If it had, it likely would have rejected the claims. The ’842 Patent goes out of its way to make clear that its method can be implemented on a generic computer. The final three columns of the patent recite at length how its claims can be implemented in any programming language on essentially any kind of hardware.

At one point, the examiner did reject some of the application’s claims under Section 101 of the Patent Act (which is the statute the Alice decision applies). But IBM overcame the rejection simply by arguing that the patent’s method was implemented in computer hardware. In January 2013, IBM noted that “it was agreed [between IBM and the patent examiner] that the rejection ... under 35 U.S.C. § 101 could be overcome by reciting that a hardware storage device stores computer readable instructions or program code.” Even if that was a reasonable response in 2013, it certainly was not after Alice. Yet the Patent Office never revisited the issue. We have submitted multiple rounds of comments (1, 2, 3, and 4) to the Patent Office urging it to be more diligent in applying Alice.

Even if the claims of the ’842 Patent were non-abstract, they still should have been rejected as obvious. We’ve written before about how the Patent Office does a terrible job of finding and considering real-world evidence when reviewing patents. In fact, it seems to operate in an alternative universe where patents themselves provide the only evidence of the state of the art in software. The prosecution that led to the ’842 Patent is a stark illustration of this.

You might think that a patent examiner faced with a patent application on an out-of-office email system might look at some real out-of-office email solutions. But the examiner considered only patents and patent applications. The Patent Office spent years going back-and-forth on whether IBM’s claims where new compared to a particular 2006 patent application. But it never considered any of the many, many, existing real-world systems that pre-dated IBM’s application.

A figure from IBM’s patent (left) and from its 1998 publication about Notes (right)

To take just one example, the Patent Office never considered this detailed specification from 1998 (PDF) from IBM describing the out-of-office agent in Notes. Nor did it consider other well-known email features like scheduling and signatures. If the Patent Office had taken a peek at the real world, and applied a modicum of common-sense, it would have quickly rejected IBM’s claims.

Some advocates for software patents have recently been pushing for legislative reform to undo Alice. Indeed, IBM is among those asking Congress to reopen the software patent floodgates. If they succeed, perhaps IBM can finally get a patent on shorter meetings (that application was rejected under Section 101). It’s clear that software patents do not help people who actually write software. And while Alice has caused some frustration for those who churn out software patents, it has not harmed the software industry.

The patent system is still far from perfect. But the last thing we need is to go backwards and encourage the Patent Office to issue more nonsense like IBM’s patent on out-of-office email.