This ruling extends the basic rule of Bilski
that you can't patent an idea to the computer. Essentially, Alice Corp.
states that you can't patent a general idea simply by appending the term "on a computer." What the case doesn't say is that all software patents are invalidated. Rather, the software has to be more than just a generic business idea expressed "on a computer."
Alice Corp. was the assignee of several patents for mitigating "settlement risk" via software. Software claims in patent law usually occur in two parts: a method of performing the claimed function and a system for performing the prior claimed method. This basically lets a patent holder guard against people manipulating their way around the system or method claims to perform the exact same function, such as by using a remote server instead of a local hard drive, or querying before step A as opposed to after step A. Alice Corp. had both a system and method patent for mitigating settlement risk. Specifically, the claim contemplated two parties using a third-party intermediary, in this case a computer, to create account ledgers (or "shadow accounts," as the patent called them) based on the accounts of both primary parties, determining available versus unavailable funds, calculating a risk for a given set of transactions, and then issuing instructions to the parties telling them what transactions are permitted and what transactions are too risky to engage in.
The patent itself was to "facilitate the exchange of financial obligations between two parties by using a computer system as a third-party intermediary." If this sounds like an abstract business method, that's because it is. Alice Corp.'s patent was basically a claim to mitigating settlement risk by employing a third party on a computer. The intermediary was just the computer in this case. The district court found that the patent was too vague, because it really only contemplated an idea. The Federal Court on first hearing reversed, saying that it wasn't "manifestly evident" that the patent ideas were abstract, so the case should be litigated rather than dismissed on summary judgment. On second hearing en banc (i.e. with all the Federal Circuit judges present, the Federal Circuit changed its position and determined that the method claims were invalid. There was some internal dispute, however, as to whether the system claims -- i.e. a claim over a computer that performs the function -- was valid.
The Supreme Court determined, in an opinion written by Justice Thomas, that both the method and system claims were abstract and therefore invalid. The rule under a cased called Mayo Collaborative Services v. Promethius Labs requires that an abstract idea, to be patentable, must have some practically beneficial application to either the computer system implementing it or to some other kind of technology. For example, it might be common knowledge that plucking a guitar string emanates a harmonic frequency, so I can't patent plucking a guitar string, but if I find a new, beneficial use for plucking a guitar string, such as a patent on plucking guitar strings to encourage the growth and development of plants (yes, it's nonsensical -- I can't invent good, patentable things on the spot!), then I could patent that. Here, though, the court asked "whether the claims at issue here do more than simply instruct the practitioner to implement the abstract idea of intermediated settlement in a generic computer." Laconically, the Court concluded, "They do not."
So what does this mean for software patents? Well they still are valid as more than just math. Though the Court didn't address the issue directly, it has been generally held that software is sufficiently transformative to warrant patentability. While not a favorite opinion here on Slashdot, that's not altogether nonsensical; if software is sufficiently artistic enough to be copyrightable, then it stands to reason that it has been sufficiently transformed from some natural state to warrant patentability as well. That said, algorithms are not patentable. A software claim that only protects against doing specific math, as opposed to being comprised of math, is not valid and will be (and has been) invalidated. What's the difference? You have to take your tech beanies off and look at it from the perspective of very intelligent, but non-technical 60+ year-olds.
What this case does do is extend Bilski to the software world. You couldn't patent an idea. Now it's articulated that you can't patent an idea on a computer. So finally, all of those irate claims we've made about "that's just a patent for doing X on a computer!" has some valid case law supporting it.
Law school graduate, not a lawyer (yet), who works in a patent law firm. This isn't legal advice. Blah, blah.