There is perhaps no occasion where one can observe such profound ignorance, coupled with reckless profligacy, as when a software patent is discussed on the web. Note the recurring pattern, which is repeated every two weeks or so. A patent issues, or a patent application is published or patent infringement suit is brought, and within minutes the web is full of instant pundits, telling us what the patent covers, how it should not have been granted, how it is entirely obvious, or how it applies to everything in the world, and how it presages a self-induced mutually assured destruction that now leads us on to the plains of Armageddon. If I had a nickel for every time this happens…
By way of disclaimer, I am not a lawyer, but I am blessed that my self-avowed ignorance in this area is coupled with a certain knowledge of the limits of my understanding, a handicap seemingly not shared by many other commentators. I know what I do not know, and know when to seek an expert.
In the past few days we have had a bumper crop of pontification on the significance of two XML-related patents, one newly issued to Microsoft (7,571,169), and another older one (5,787,449) owned by i4i, whose infringement has resulted in a large judgment and injunction against Microsoft. I’ve found the web coverage of both patents to be an unmitigated muddle.
I’m not going to comment on the merits of either one of these patents, but I’d like to make a few basic observations that may be of some assistance to those who comment on future patent issues.
- A patent has a description known as the “specification”. And it has a list of numbered “claims”. Although the specification can define terms that are then referred to in the claims, it is the granted claims that define the scope of the patent, not the specification.
- If all you do is read the abstract and the first few paragraphs of a patent, then you may know the general topic of the patent, but you do not really know its scope. If you then go off and cry, “Oi vey, this patent covers XHTML, SVG, RDF, Pringles and Obama Healthcare Plan” then you do your readers a disservice. You must parse the very specific and often obtuse language of the claims in order to understand exactly what a patent covers. There is no short cut. This is not like a book where you can understand the plot by reading the back cover. But over and over again, I see people who just read the abstract, maybe glanced at a diagram, and then feel equipped to hold forth at length on the substance of the patent.
- When you try to understand patent claims, you will encounter a dense form of legal English. Claims are not written for a layperson and do not presume that you will understand it easily. The drafting of patent claims is a black art, like writing device drivers, and if you are not versed in its intricacies, then your statements on any given patent are apt to be wide of the mark. Claims are full of magic words. Know what you do not know. If you do not recognize at sight and know the interpretation requirements of a means-plus-function claim (which is key in the ‘449 patent), or you are not crystal clear on the distinction between the verbs “consist” and “comprise”, then you probably should not be the first (or loudest) person to speak on what a patent claims.
- If you are reading an application, know that during the “prosecution” of that patent, when it is reviewed by the USPTO, some of the claims of the patent may be thrown out, for any of several reasons, including prior art identified by the examiners. However, the specification of the patent is unlikely to change much. So an issued patent often has a very broadly-written specification, that covers the entirety of the originally claimed invention, though the the issued patent might have only a subset of the original claims allowed. So if you are have an issued patent and you look at only the specification, you can easily be fooled into thinking it covers far more than it does. For example, the ‘169 patent from Microsoft had half the original claims thrown out in the prosecution. If you don’t know that and are reading only the specification, not the granted claims, then you will incorrectly think the patent is far broader than it actually is.
- Know what a priority date is, and how that is affected by a continuation. I’ve read all sorts of nonsense based on not appreciating that. Take a look at the ‘169 patent, for example. It says it was filed in 2004. But if you look closely you see it was a continuation of a 2002 application. You can moan and groan all you want about prior art, but if you don’t get your dates right you’re off to a bad start in your analysis.
- In an infringement suit, like with the ‘449 patent, be sure to look at the actual court record. Typically there is a Markman (claim construction) hearing, where the court will determine the meaning of terms used in the patent claims. If you have not read the court’s claims construction opinion in the i4i versus Microsoft case, then your commentary is uninformed by probably the most important document in this case (well, next to the patent itself).
Well, that’s enough for tonight. Repent. Sleep on it. And realize that making sense of a complex patent takes time, if you’re going to do it right. Ergo, the first impressions you read from the instant pundits on the web will tend to be shallow, imperfectly informed and often wrong. Heck, even everything I said in this post may be wrong.
Excellent post – tons of stuff in there I didn't know about, not that I speculate on patents much.
However, I found it rather funny when I looked up the difference between consist and comprise in the OS X dictionary:
comprise |kəmˈprīz|
verb [ trans. ]
consist of;
It does however note confusion in modern times that comprise means "consists entirely of" rather than just "consists partly of".
I would like to know where I can read the court's claims construction opinion in the i4i versus Microsoft case, or at least where I should search for it.
Thank you in advance.
@jhominal:
On groklaw (http://www.groklaw.net/article.php?story=20090812144154814) there are links to the original complaint, Microsoft's response and i4i's response.
The real problem with software patents is that not only are they incomprehensible to lay people, they're incomprehensible to technical people. Their claims are described in "process" terminology that runs counter to how developers think about software operation. So it's not surprising that most of the commentary fails to make sense of what's going on.
If software relied simply on copyright for protection, there would be much less FUD about software development. As things stand, no one can tell what might violate a patent.
These are transition terms, which separates the claim preamble from the body of the claim: "comprising" is legally 'open-ended' which means that what the claim includes in scope/coverage can be other things not mentioned in the claim; "consisting" is legally considered 'close-ended' meaning that the claim is limited in scope to only what is explicitly mentioned in the claim. Claim construction (i.e., interpretation) is very complex, so much so that it invites that which attorneys love to do — argue. Ultimately, it takes a court case and judge/jury to finally determine the scope of any claim. Anything else is just an opinion. A patent is just a starting point.
Software patents are obtuse; software patent language is obscure
Security through obscurity forever!
There is another reason the specification is so broad, which is hinted at in the article. Companies file patents to put their competitors at a disadvantage. They claim as much as they can, so the competitor cannot work around the patent, but they also describe as much as they can, even stuff they can't claim and don't know how to engineer, to establish published prior art that makes it difficult for a competitor to file a patent in the same field.
With such a long description, they also have the option of trying to claim the more far-fetched aspects in subsequent continuation patents, or add further descriptions and claims via continuation-in-part patents.
So the general advice is to describe everything that you think you might use, and everything you think a competitor could possibly use, and claim as much as you can get away with. If it's a first patent, you may choose to limit the claims to guarantee it passes, but you can then extend it using continuations.
Interestingly, from what I can tell the Microsoft '169 patent is fundamentally useless against ODF. The two independent claims both have a limitation that the document format uses Microsoft's oddball method of doing formatting and not mixed content. These limitations were added during the application process – the original application doesn't have them.
This was almost certainly due to prior art issues, and my understanding is that this bars any use of the patent against things that match the original claims but not the narrowed claims. IANAL, though.
Very interesting article and some good comments. Even my knowledge of the English language has improved (I think).
I once had to review a software spec to see if it conflicted with a patent owned by my company. I had the opportunity to read our patent extremely carefully. It consisted of a number of claims. Each claim was comprised of one or more clauses. An impingement on any one of the claims was for claiming that the spec impnged on our patent. But to impinge on a claim it had to impinge on all of the clauses.
The person or company that gets the patent wants as broad a reach as possible, and rivals will want to construe the patent as narrowly as possible.
“Software patents are obtuse; software patent language is obscure”
I disagree. Many patents, including software patents are clear in their wording. Those patents, however, do not get much attention, because both the patentee and the potential infringer can assess the scope. In such cases they both know when not to bother the courts. The unclear patents, on the other hand, where neither the patentee nor any third party or even their legal counsels can asses the scope are likely to get all the attention because they end up in court. They end in court only because neither the third party nor the patentee knows whether there is an infringement or not.