Software patents poised to make a comeback under new patent office rules
arstechnica.com
arstechnica.com
The patent system was conceived in an age when innovation was represented by complex works like the steam engine. Invention was the result of applied engineering and ingenuity.
Now that system is weaponized by organizations claiming exclusive use over something as mundane as a UI gesture.
An institution intended to spur innovation is achieving the exact opposite, chilling effect.
Here's an interesting comment from a seasoned investor on the value of formal IP protection to the average internet startup: https://youtu.be/15iWltPLuPY?t=2274
While I can't find it now, I recall another I/O talk where an investor was asked by an aspiring founder how to deal with the fear of unwittingly treading on someone's patent. In response he opined the very act of innovating in today's world makes you likely to trip over some patent or another, and his guidance was to focus on building a great product.
I'd love a world where companies compete based on the merits of their product and talent of those who craft it. Let them stay relevant by continuously pushing the leading edge of innovation, rather than resting on the laurels of accomplishments from 20 years ago.
Kudos to companies which have commited to never using patents for offense, and bigger kudos to those who have gifted their IP to the world.
That gives credit to companies simply for being in particular types of industries. Facebook and Google have opened up large amounts of IP in the form of open source. They do so because they don't sell software; software is just a gateway to their ad platforms and user databases (which, unsurprisingly, are well-guarded). Does that deserve "kudos" compared to say, Microsoft, which doesn't open up IP as much because it's in the business of exchanging software for money? I don't see a moral distinction to be made there.
I believe you must have missed the news of Microsoft sharing 10,000 patents with their azure customers? Is it different from what you're referring to?
https://startups.microsoft.com/en-us/blog/azure-ip-advantage...
Patents aren't about morally owning property -- they're about their effectiveness. The case of patents needs to be made for their continuation.
Intellectual property is a misnomer, it's not property -- it's restrictions on what others can do with their own property.
The history of technology repeatedly shows that (temporary) monopolies are good things. The transistor was invented by the AT&T monopoly. GUIs, email, and networking were bankrolled by Xerox's patent monopoly on copiers. (PARC declined almost immediately after a DOJ suit forced Xerox to license its patents to Japanese competitors.) Google's panoply of innovations is bankrolled by its near-monopoly on search. By contrast, commoditization tends to kill innovation in an industry. PCs have languished for two decades now thanks to a highly competitive environment where companies like Lenovo struggle to eke out a few percent in profit margins.
What I really want to see is for obviousness tests to be strengthened. I believe that in order for a patentee to sue for infringement, they should first have to present objective evidence that their idea is actually nonobvious, along the lines of the "Graham factors" [0]: commercial success, long-felt but unsolved needs, or failures of others.
The nonobviousness requirement has become toothless in recent years; the Federal Circuit has reduced it to a requirement that no prior art exist explicitly mentioning the invention. But lots of things are obvious that haven't been explicitly mentioned in some publication; indeed, many things aren't mentioned because they are obvious. And the PTO is poorly placed to evaluate nonobviousness; for one thing, one of the Graham factors, commercial success, may not exist at patent application time — one may have to wait a few years to see whether the invention gives rise to a viable business.
So the right time to evaluate obviousness is at the commencement of litigation.
This triggers one of my pet soapbox issues: The U.S. patent statute requires a patent to be issued to an inventor unless the patent examiner can show that the claimed invention was not new, useful, or nonobvious. True, the inventor must tell the patent office about any prior art of which s/he happens to be aware, but the inventor has zero obligation to do any kind of prior-art search to see what else might have been out there. Contrast that with the Ph.D. degree system, where the burden is on the candidate to do a literature search and convince his/her committee that s/he didn't reinvent the wheel.
Another contrast: When a university grants a Ph.D. degree, the result isn't national industrial policy in the form of a statutory monopoly over the claimed subject matter.
And another one: Ph.D. degrees require a defense before not just an experienced advisor, but also a committee of people who know at least something about the subject matter. In contrast, patents and their statutory monopolies are typically approved by junior- or mid-level examiners generally either flying solo or (in theory) supervised by one more-experienced examiner. Only if the examiner issues a final rejection will the invention be reviewed by a board of appeals comprising senior examiners.
Yes, that's quite backwards, isn't it? Granted patents should be reviewed more carefully than rejections. Where the PhD system is built to minimize false positives, the patent system is built to minimize false negatives.
Here's how I've seen it put in the past. A patent is a deal the public makes with an inventor: the inventor gets a temporary monopoly in exchange for disclosing how the invention works. The problem we have is that no one in the system — neither the PTO nor the courts — is specifically responsible for representing the public's interest, by making sure that the information being disclosed is valuable enough to deserve the monopoly.
By law, that's the job of the USPTO. The PTO routinely rejects patent applications if the examiner thinks that the claimed subject matter wasn't novel or would have been obvious at the time the invention was made. Most rejections stick (I would guess), but sometimes examiners get worn down by patent attorneys. The problem was famously summarized nearly a century ago by the legendary federal judge Learned Hand, who said that "... the antlike persistency of solicitors has overcome, and I suppose will continue to overcome, the patience of examiners, and there is apparently always but one outcome." [0]
[0] Lyon v. Boh, 1 F.2d 48, 50 (S.D.N.Y.1924) (L. Hand, J.), at https://scholar.google.com/scholar_case?case=969659756696519... , rev'd on other grounds, 10 F. 2d 30 (2d Cir. 1926).
If that's true, then the Fed Circuit has judicislated the nonobviousness requirement out of existence, if it adds nothing to the novelty requirement.
— Christian Heilmann (@codepo8)
> On the flip side, the argument that without software patents modern tech-based companies can't thrive seems weaker every day. A 2015 article calculated that 30 percent of billion-dollar startups have zero patents.
This is an odd take on the original data source: https://www.ipwatchdog.com/2015/11/03/the-naked-truth-30-of-.... If you look at the table mid-way down the page, only one category of companies has relatively low patent-ownership share compared to valuation share: "consumer Internet." Of course, these companies don't need patents to protect their IP. They don't deliver a product to consumers that their competitors can disassemble and copy. Their IP is hidden in their servers behind a web services API. Uber's logistics algorithms, for example, are important trade secrets, but there is no need to patent them because they can't be reverse-engineered from the Uber app.
As shown in the chart, several kinds of companies deliver software products to consumers, or disclose technology through standardization, and have a relatively greater need for patent protection. Waymo, for example, has tons of patents: https://patents.justia.com/assignee/waymo-llc. So does CMU, which pioneered early work in self-driving cars.
> In a 2016 ruling called Enfish, the Federal Circuit ruling took a single sentence from the Supreme Court's 2014 ruling and used it as the legal foundation for approving more software patents.
This reads Alice backwards. Alice was about why patents on business methods are ineligible, even though they are implemented in software. I.e. patents on using software to automate methods of doing business that were already in use. That is why when Ars covered the decision in Alice, its headline was "Supreme Court smashes 'do it on a computer' patents in 9-0 opinion," not "Supreme Court smashes software patents in general." Alice assumes that real technological inventions that happen to be implemented in software are patentable.
The "single sentence" in Alice Ars that refers to is actually two sentences:
> The method claims do not, for example, purport to improve the functioning of the computer itself.
> In other words, the claims in Diehr were patent eligible because they improved an existing technological process, not because they were implemented on a computer.
Ars makes it sound like the Federal Circuit (and now PTO) took a narrow exception in Alice, articulated in a single sentence, and used it to swallow the rule. But in Alice, the Supreme Court viewed business-method patents as the exception to the general rule that useful inventions are patentable (even if they are implemented in software). These sentences simply explain why the patent in that case falls into the exception.
To put it in concrete terms, compare the patent in Alice (on the business practice of intermediated settlement), to something like an algorithm for power management in a laptop that improves battery life. Alice says that the former is not patent eligible, in part by distinguishing it from the latter. But it's not really talking about the latter category of eligible inventions, or making any statements on how broadly or narrowly that category should be defined.
First, in Flook, SCOTUS effectively said that software wasn't patentable, the software in question being a controller for a catalytic converter. In Diamond v. Diehr, SCOTUS said that software was patentable insofar as it was part of a larger patentable system, the software in question being a controller for a rubber mold.
Effectively, SCOTUS (in a time-honored tradition) overturned Flook in Diamond v. Diehr while simultaneously emphasizing that it wasn't overturning Flook.
The Federal Circuit then created the infamous State St. ruling, which held that anything was patentable if it produced a "useful, concrete, tangible result." This is generally recognized as the moment when the patent system broke.
In Bilski v. Kappos, SCOTUS again returned to striking down dubious patents. Of particular note, all of the opinions went out of their way to criticize the State St. formulation as exceptionally bad law. They also rejected the CAFC's alternative-to-State-St. formulation of "machine or transformation test" while being unable to articulate what the correct test ought to be. Mayo and Alice, as noted in the article, further slammed both the State St. notion of "pretty much everything you can think up is patentable" and the general class of business patents. At the same time, SCOTUS has been reluctant to fully close out what isn't patentable (so as to avoid accidentally shutting out Next-Big-Technological-Area-That-Really-Needs-Patents) while being unable to articulate what makes something in a really gray area like software able or unable to clear the bar.
The thrust of SCOTUS is pretty damn clear: if your patent is struggling to tie its innovation to something that's pretty clearly patentable under 18th century patent norms, it's not going to be patentable. That it can't formally articulate those guidelines into a coherent, clear statement is unfortunate but completely understandable. Software itself is in a gray area that really could use clarification, but SCOTUS isn't going to do that without an appropriate vehicle to rule on. Someone needs to try to obtain a patent that is clearly novel, useful, and pure software and have its validity challenged, and have this challenge be appealed up to SCOTUS, before you're going to get a decision where SCOTUS may try to decide it.
Bilski, Mayo, and Alice have all shown that several justices are inherently skeptical of software patents, and they've also shown that software patents are of major interest to the wider topic. But SCOTUS is generally unwilling to rule on a question that isn't directly presented and argued before them, so until a case that itself directly involves software patent 101 eligibility (and has no room to rule invalid on 102-104), there is unlikely to be much clarity here.
Not true. They said system could be patentable despite having software as a component. This is common sense. If an invention includes a pulley, a wheel, and a lever, you don't evaluate what's there if those non-novel elements are removed; you look at the whole system. SCOTUS said to look at the system as a whole, including the software. But the novelty and non-obviousness have to exist in the system as a whole. If it's only novel and non-obvious in the software, generally speaking, no patent. They did allow for the possibility that patentable software might possibly exist, but the gist of the ruling is that they hadn't seen such software yet. Also that it was up to congress to clarify, because they didn't feel 100% comfortable deciding if software could possibly be patentable.
So they basically stopped a millimeter short of saying software couldn't be patented at all, and the State St ruling stuck a crowbar in that crack.