Good points all — and not contradictory to mine.

http://en.wikipedia.org/wiki/In_re_Bilski currently is a pretty good read. The parts on dissenting opinions outline the issues neatly.

Well, some of the issues. They don’t speak well to algorithm-gadget equivalence — i.e., the problem you noted in distinguishing among the patentability of multiple versions of the Dolby technology.

Both Dolby versions were implemented as circuitry – a couple of rack units for Dolby A, a clever circuit board for Dolby B. But at heart both were just digital signal processing algorithms. Today, you could implement them in software with little time and effort.

By traditional standards, Dolby sound processing was clearly patentable (and was, indeed, patented). But suppose DSP’s had been more advanced at the time, and the algorithms were implemented in software rather than by soldering parts together. Should that have affected their patentability? Why?

It’s interesting that the Bilsky case – now before the Supreme Court – returned to a much early approach for judging patentability, looking at “the kind of thing” that is being done in the patent. It looks positively on patents involving transformations to the real world, negatively on things that just manipulate abstractions inside the computer. Dolby software would probably have passed muster under the proposed Bilsky standards.

Two other things:
- At one point I was involved in reviewing engineer’s proposals for patent filings at a large corporation. One judgement we made was: Is this something visible from the outside, or is it purely an internal algorithm? We tended to reject the latter, because as a practical matter, it’s not useful: It’s too difficult to tell if someone is using your algorithm if there are no reasonably unambiguous outward signs. Many of the algorithmic patents out there that people raise such a ruckus about are of this nature. If they play a role in the real world, it’s usually as part of large portfolios that corporations lob at each other: Lawsuits get started over visible patents, and the internal algorithms then show up, or threaten to show up, during discovery. It’s not that they aren’t an issue – but given the realities, even if you’re a strong believer in software patents, you shouldn’t be too concerned about these kinds of patents. In the best of circumstances, they do damage without actually helping inventors.

- People have the feeling that the patent office is incapable of reviewing software patents and just accept anything. I had the experience of having a software patent rejected by the examiner four times based on prior art he found. The first time, what he found was actually pretty close, and it took some effort to show that what I and my fellow inventor were proposing was actually different. The second rejection was somewhat easier to deal with, and by the third and fourth the examiner was clearly in the realm of diminishing returns. Still … at least some software patents are now receiving significant, knowledgeable review. Anecdotal evidence, of course – but that’s what we have for all the bad examples of software patents as well.

— Jerry

It used to be that some competitor came out with a great new product and you could put some engineers in a clean room for a month and tell them to build the same product, tell them nothing about how and that was ok.

For some reason, that isn’t ok in software patents.

Also, you could actually look at the mechanical patents, design around them and bring that new product to market in a non-infringing way. There’s no real way to do this with software patents either because of the broad and vague claims they’re allowed to make.

Overall, perhaps the ideal of the idea of software patents isn’t broken but any implementation of it seems fundamentally flawed to me. (Especially the ridiculous mess we have now.)