Sunday, July 24, 2005

:-) = Patent infringement?

US Patent application #20050156873 was filed July 21, 2005 January 20, 2004, and made available online July 21, 2005. The basic lawyerese of the claim is

A method, comprising: selecting pixels to be used as an emoticon; assigning a character sequence to the pixels; and transmitting the character sequence to a destination to allow for reconstruction of the pixels at the destination.

Translated into English that means use of graphical representations of smileys in an instant messenger program. This one adds even more emphasis to what's wrong with the US patent system. It also underscores the danger of some “patent reform proposals” that change the rules for who is entitled to a patent from “first-to-invent” to “first-to-file”. Despite the proposals' best intentions, there will likely be a rush on the patent office with patents on trivial things that will become even harder to challenge (would prior art still matter at all?). While it's good that the procedure for challenging patents may be streamlined, the fact that the filer didn't invent it should be enough to knock down their exclusive use rights. Let alone that this patent covers an obvious extension of being able to send text and images over instant messengers.

Technorati tags: , , ,

A little late here, but although I'm *very* sympathetic to the sentiments you express, I don't think the sky is falling in this instance. The method described *could* meet the threshhold novelty and non-obviousness criteria and also may not be quite as broad as you've characterized it -- and the claims may yet end up being narrower. Amazon one-click this isn't. Also, a nit-picky point: the application was filed in 2004, although it became accessible more recently at the publication stage.

Nit-picky correction noted.

Software doesn't work the same way real-world inventions work. Most software "innovation" builds trivially on something that's already there, and all software is reducible to mathematical statements. As math, it's not patentable, but as software, it is?

Also, the fact that the same functionality happens in one place in a renderer and in another in an instant messenger program is really irrelevant to a software programmer.

Take, for example, transmitting a character (or group of characters) and displaying an image instead. That's how all "fonts" work. Fonts were not invented in 2004, nor were they invented by Microsoft. Multibyte characters and the fonts for those weren't either. The fact that the picture may be 19x19 pixels and look like something other than a letter doesn't change that it's a trivial application of the concept.

What I would be more curious about from you is your opinion on the proposed change from "first-to-invent" to "first-to-file."
Also seems that AOL IM was doing this as early as 2002 (if not earlier -- most of the screenshots don't seem to be coming though on the Internet Archive).
I don't prosecute patents, and as you know, I'm far from being a software engineer. However, I have some basic familiarity with some of the issues implicated.

Before I answer the question you want me to answer, I respectfully disagree with your plain-language "translation" of the claims and with your AIM example as relevant prior art. The claims on their face do not cover all graphical representations of simple emoticons. Further, they note that the prior art involves selection from a set of pre-fab emoticons.

The crux of MS’s claims is that their method better enables creation and dissemination of “customized” emoticons. I lack the expertise and the inclination to examine the drawings and research the prior art to decide whether the claim has any merit, so I’m completely agnostic on that point. However, if MS’s claims are in fact knocked out by prior art (and they may be), I don’t think you’ve identified the clincher as yet.

Also, the real problem isn’t that the nature of software development doesn’t fit the long-established criteria. The law does not require earth-shattering advances or even hard work to get a patent. It requires: 1. “novelty” (i.e., the precise thing sought to be patented does not otherwise exist); and 2. “non-obviousness” (i.e., it can’t just be the logical “baby step” from the prior art to a person in the field) – there has to be some “chiddush.”

Determining whether these criteria have been met requires a good understanding of the prior art – and *that* is the main problems here, in my view. Historically, PTO examiners have often been poorly trained and equipped to identify prior art in certain areas. There have been large-scale attempts at the PTO to address that issue in recent years, although more needs to be done, of course. Also, the phenomenon of private sector people offering bounties to find prior art is a good thing.

As to your question, I would not support switching to a “first-to-file” patent system, because there are already considerable incentives for inventors to file. Also, unlike the copyright context (where most application fees are $30 and the applications themselves are less likely to be fatally screwed up by non-lawyers), the expense of filing for a patent may well be prohibitive for poor inventors.

Adding a bolt to an instrument that improves its functionality is a small improvement that in some cases may be nontrivial/nonobvious even if it's a small incremental improvement.

But, software is totally different. A customizable icon set and a fixed icon set are the same technology, just like a font and a graphical emoticon are the same technology, and a 20x20 box and a 19x19 box are the same. I looked through the drawings, and IMHO they're laughable. The 2 flowcharts that actually show something substantive show exactly the same algorithm used to render fonts. The detailed structure of their program describes a set of standard techniques for graphical rendering and network communication hidden in obfuscating technobabble.

Both bars of what's novel and what's non-obvious are set way too low when it comes to software. The problem is that there is going to come a point (it's almost here) where software can't be written at all without infringing on patents.

Then, there's the argument that software should not be patentable in the first place, since all software is equivalent to statements of mathematics. about how Actually, I can find some references where implementations of basic mathematical theorems were patented.
Doesn't adding a bolt to an instrument amount to using the same technology as when the instrument had one bolt less? It's all in the implementation, it seems to me.

If the main concern is that too many software patents stifle innovation, I hear the argument, but overall, innovation continues apace. The PTO scales back claims during the examination process, and also re-examines issued patents. Courts also curb the scope of claims and can invalidate patents entirely.

In general, I think abuses can be curbed far better by pushing for rigorous attention to prior art and re-examination of questionable issued patents than by enacting new laws.

The new bolt changes the physics of the instrument. If it changes it enough, what you have *might* be a new and better instrument. Making a slightly different sized physical object, also, can mean a lot of nontrivial changes. In either case, there's an actual object involved, some real physical standard against which potential infringement can be judged.

Because of the way software is engineered from preexisting *virtual* components, there is no way to write a software patent that is not one of: (1) covers only the specific implementation/use of the exact components listed, and can never be infringed except by exact copying, OR (2) overbroad and covers some principle of mathematics or established standard programming technique.

Your first thought (although I acknowledge that you didn't commit to it), if I understand correctly, ends up making this patent a case of #1. I'm thinking it's #2.

Of course, none of this addresses how a computer algorithm is any different from a nonpatentable mathematical statement.

I realize that I've thrown together a jumble of arguments against this patent in particular and against software patents in general, so this comment thread is probably a bit hard to follow,
Epilogue -- MS is on a patent-filing binge:

This filing may have predated the binge, and my views remain pretty much as described further up in the thread.

That went through the computer-press too. They increased their patent goals from 2000/yr to 3000/yr. If the USPTO were to go through these properly, they need a team of examiners permanently assigned to Microsoft filings.
Post a Comment

<< Home

Links to this post:

Create a Link