A while ago, when Google’s Android mobile OS first came out, I wrote an analysis of its use of Java, speculating that there were not only technical reasons for Google to come up with its own virtual machine (named Dalvik, if you care) but also legal ones.

I knew that because I had been one of the original proponents for the Apache Harmony project and I had watched with great interest (and other feelings, not all of them positive) the way the opening up of the Java platform unfolded (and how Sun is constantly trying to preventing it from unfolding it even more).

So Android came along and while I was excited to see somebody calling Sun’s bluff on their ability to prevent others to implement Java or even fork it, I was worried that Sun might actually sue Android phone makers preemptively and stop them from shipping.

Well, that fear got reduced substantially when the T-Mobile G1 phone hit the market with no (public) word or action from Sun.

Now, I don’t know if HTC (the handset maker) and Google (the software copyright owner) paid some licensing fees to Sun to keep them happy or at bay, but I couldn’t find any information on such a thing so I assume they didn’t or they would have found a way to leak it out.

Certainly though, if I were a cell phone maker or even a mobile service provider, I would be pretty nervous to pump so much hype into a platform with such an uncertain legal foundation.

That was until a few weeks ago, when the United States Court of Appeals for the Federal Circuit, in a decision that is likely going to shift a lot of power in many IP-heavy industries (such as the software one), basically reverted a previous court decision to allow business processes to be patentable and, well, now says they are not.

In very short, before this ruling, something was patentable if it created “useful, concrete and tangible result”. This new ruling now says that such a test “is insufficient to determine whether a claim is patent-eligible” and “reaffirm that the machine-or-transformation test outlined by the Supreme Court is the proper test to apply”.

This apparently simple decision has game-changing effects in how effective a patent claim could be in protecting an idea or an algorithm when it’s not tied to a special-purpose machine that executes it.

Which means, for Android and for the open Java ecosystem in general, that Sun’s legal sword against it is way less sharp than it was before the Bilski case, because, by definition of virtual machine, there is no special-purpose hardware machine, nor physical transformation that can justify the patentability of how the software runs.

The game is not over for software patents yet, as the court explicitly avoided stating something definitive over the patent un-eligibility of software and the equivalence of algorithm and processes to ideas, but it’s going to be much harder to obtain a software patent without any hardware associated to it and to enforce such an existing patent or use it as a deterrent in an IP portfolio.

And yes, this also means a lot less beach-front property for a lot of software companies that invested in patent portfolios.