IBM Patents Optimization... Sigh. | Hacker News Reader

IBM Patents Optimization... Sigh.

google.com

17 pointsjolie10 comments

IBM is so funny. There are large incentives to get patents there as an employee. Prior to when I left and went to another company, there were even expectations that senior level engineers would regularly generate patents and patentable work as part of their normal work.

I remember one funny year when they pushed the patent expectation down to the lower banded employees (of which I was one at the time). We all dutifully submitted (internally) a number of patent "inventions" to meet the requirement and get the magic bonus $$ if you actually got one awarded. At the time, it didn't feel like an honest effort to generate innovation, but rather just another out-of-normal-process to heap on employees who were mostly trying not to get laid off.

I'm usually skeptical about complaints of seemingly over-broad patents. Most people don't realize that the abstract isn't the patent, the claims are. And the claims tend not to be overly broad because the examiners aren't that bad.

But in this case, it really is as bad as the title declares. Claim 1 covers the process of any attempt at optimization.

The only things close to redeeming specificity are in the description and background. The first claim still covers everything.

One small caveat: IBM is attempting to patent optimization. They filed an application and that application has been published.

But, as of yet, it hasn't been granted. Fingers-crossed, let's hope this doesn't go through.

Yeah, the claims are weird here, even just in terms of format. Usually claims are these dense paragraphs you can't interpret without a background in patent law, but here, the entire first claim is a few short clauses:

A method for developing a computer program product, the method comprising: evaluating one or more refactoring actions to determine a performance attribute; associating the performance attribute with a refactoring action used in computer code; and undoing the refactoring action of the computer code based on the performance attribute.

The only really limiting factor is that it could be read narrowly to require a very close "associat[ion]" between the "refactoring action", the "performance attribute", and the "undoing"; e.g. some sort of "refactor-and-keep-only-if-faster" button in Eclipse.

Claims typically start out broad, and then refine further to get at the core of what exactly it is they are patenting. Since many patents are extensions or improvements on existing things this seems to be a common theme in the patents I've skimmed over. This patent application just keeps everything vague and broad, which I suspect is common in many software patents.

That's incorrect. An invention is infringing an issued patent if it includes all the elements of any independent claim.

Dependent claims, in this case 2 - 5 (the ones that refer to another claim), add specificity, but do not narrow the scope of the original independent claim. They exist as a legal tactic; helping to ensure that a more specific variant of the invention is covered, in the event that the parent claim is later invalidated (ruled too broad, revoked upon discovery of prior art, struck after re-examination, etc).

But in case of a legal dispute, it may be that only certain contested claims are ruled illegitimate. So it makes sense to have many claims ranging from less to more specific, instead of having a few specific claims, so that maybe a somewhat wider claim will make it through a legal dispute.

There's absolutely a lot of legal strategy in developing claims. There are very good reasons that top firms hire specialized IP lawyers. All that is, however, an aside.

The original point stands: If this patent issues, any competing invention that merely contains all the elements of that broad independent claim, would be infringing.

Those more-specific dependent claims are only a concern if you plan on prevailing in court against the broad one. As setting foot in a courtroom is a net loss for most innovators, the very existence of those dependent claims is essentially irrelevant.

"Sigh" is right. This isn't a patent. It's an application that hasn't been examined yet. Come on people, try to learn a little bit about the subject before trolling.

Hope they get it. And they can defend it. In fact I am hoping for a patent on computing (damn can't do that). We need to prove by extreme example that the patent system is broken.

If their war chest of patents is to discourage me from writing competitive software, then they win. Since I don't want to write enterprise software, I think I win.

Reply on news.ycombinator.com