There's not a lot of debate about whether abstract ideas deserve patents. The consensus is that they don't. The debate is about how to figure out what is an abstract idea and what isn't. On one side, you have an abstract idea like "hedging settlement risk by using shadow accounts" (the patent at issue in CLS Bank v. Alice Corp). Almost everyone on the Fed. Cir. thought this shouldn't be patentable. On the other side, you have something like whatever algorithm is the subject of the gbatteries patent: http://www.gbatteries.com/technology/ ("Our patent-pending process maintains the lithium ion diffusion at optimal levels and eliminates concentration polarization and thus allows the Li-ions to be more uniformly extracted from or reinserted into the cathode materials."). That probably should be patentable, because while the charging algorithm might be simple to describe and could easily be reimplemented, it probably took a lot of R&D to design it in the first place.
The question is, how do you draw the line between the two?
The question is, how do you draw the line between the two?
This is what has traditionally been protected by trade secrets - the problem is that software is easier to reverse-engineer than real-life problems. The integrated circuit community has dealt with the issue ever since they were invented. My feeling is that the possibility of reverse-engineering has in fact spurred innovation in that field, and it's another reason why the no-poaching agreement between Apple and Google is so abhorrent, it did not only depress wages, it also slowed progress in the field.
On the other hand, there is the issue of "gene patents". A pharmco takes out a patent on a gene, because you cannot patent abstract ideas, but what they are really doing is protecting the relationship between a gene and a medical condition. This is early discovery, and the price tag and timeline is that of a small space mission, several hundred millions and the better part of a decade. Once something is admitted to market the project is comparable to a flagship space mission, several billions and two decades of work. Patents are the wrong tool for that kind of job.
I persist in thinking that the problem is not with the subject matter test. A computer program is obviously a machine, just as the Federal Circuit and lots of other people intuitively assume.
The problem is with the obviousness bar, which is far, far, far too low. I haven't studied the Alice patent, but let's take what the NYT says about it here:
The issue in this case, Alice Corporation Pty. v. CLS Bank International, is whether using a computer to implement a well-established economic concept can be patented. [...] Alice Corporation obtained four American patents that cover a method of settling trades between investors in currency and other financial markets. The approach depends on a neutral middleman to make sure traders complete the transactions they have agreed to. [...] Alice Corporation has argued that its patents cover a specific computer-enabled system and method, not the fundamental idea of using intermediaries in financial settlements. But CLS Bank, which first sued Alice in an attempt to have its patents invalidated, says that Alice’s system does not add anything meaningful to a basic idea.
The NYT Editorial Board is using this argument to urge the Supreme Court to rule that this patent's subject matter is inappropriate. But I think this can just as easily, if not more easily, be read as an argument that the invention in question is obvious. Taking some process that was previously done by hand and computerizing it, or some part of it, is, in this day and age, a paradigmatic example of something obvious.
Again: I haven't studied the actual patent. I just think it's unfortunate that everyone is talking about this as a subject matter problem, when we have a screaming obviousness problem that the Federal Circuit is right at the center of.
I want the Supreme Court to tell the Federal Circuit that mere novelty does not entail nonobviousness. In such a highly generative field as software, which has an uncountable number of applications, simply coming up with another one contributes little or nothing to the state of the art. If someone else who encountered the same requirements could implement a system to satisfy them just as easily before reading a putative patent as after reading it, there was no intellectual contribution and the patent should not be valid. This was clearly the intent behind the patent system at its inception. It should be restored.
> we have a screaming obviousness problem that the Federal Circuit is right at the center of
Let me give a cite for that. From [0]:
The [Federal Circuit] watered down the rule against obvious patents. Supreme Court precedents prior to the creation of the Federal Circuit held that merely combining two previously-known technologies in a straightforward fashion was too obvious to merit patent protection. To be patent eligible, the Supreme Court held, an invention needed to be "greater than the sum of its parts."
The Federal Circuit developed a more permissive rule that only allowed a combination of known components to be declared obvious if there was specific documentation that someone had suggested that combination prior to the patent application. The court adopted this rule to guard against the danger that hindsight bias would lead people to over-estimate an idea's obviousness. But the court seemed less concerned with the possibility that some combinations are so obvious that no one would bother writing them down.
The question is, how do you draw the line between the two?