By: Robert R. Sachs
What motivates the USPTO to consider the use of glossaries in patent applications is surely the hue and cry that software patents are frequently, if not inherently, vague and ambiguous, and that something must be done to help clarify—and more importantly, narrow—the scope of patent claims. See, Government Accountability Report “Assessing Factors That Affect Patent Infringement Litigation Could Help Improve Patent Quality,” August 2013, p. 28, 30.
My experience in prosecuting patents, and in particular in evaluating software patents for essentiality to technology standards, is that claims in software patents are no more or less ambiguous to one of skill in the art than claims in other disciplines. There is a difference between claims that are broad and claims that are vague, and this difference is lost on most observers. Indeed the examples used by critics of the patent system are typical of broad claims, not vague ones. See Chin, On Abstraction and Equivalence in Software Patent Doctrine: A Reply to Bessen, Meurer and Klemens, J. Int. Prop. Law, v.16, no. 2, pp. 197-240 (Spring 2009).
Second, to argue that the claims of issued software patents are frequently vague and ambiguous is to implicitly argue that examiners at the USPTO who examine software patents are systematically ignorant and unskilled at understanding the meaning of patent claims, something that I believe most software patent prosecutors and most examiners would dispute. It further suggests that the examination procedures that the USPTO has developed over the years have been and continue to be, fundamentally flawed. Again, this suggestion is without merit. While there are no doubt cases of bad software patents, there are equally bad patents in all other fields of technology. The studies showing software patents make up the bulk of litigated patents do not support the conclusion that software patents are vague; rather, they simply correlate with the relative rise in the pervasive nature of software technologies in all fields of commerce and the increasing penetration of digital technologies into every aspect of our daily lives. That is, the more that software is used to provide functionality—from washing machines to toasters, from smartphones to smart cars—the more patents we would expect on software-based inventions, and the more we would expect such patents to figure in litigation.
Third, even assuming that software patent claims are more ambiguous or vague than other types of patent claims, one should consider the source of the problem before proposing a remedy. Using glossaries to fix the so-called problem of ambiguity is like putting a Band-Aid on a birthmark: it covers up an unsightly but otherwise natural result, yet it cannot make it go away.
The primary problem with claims is that they are artificial linguistic products of a complicated set of rules and practices driven by tradition and case law, that is, rules that are not perfectly uniform, consistent or clear. The simple rule, set forth in MPEP 608.01(m) that a claim must be a single sentence results in more problems than anything else, with claims running hundreds of words, having multiple clauses and subclauses all interlocked together with wherein, whereby and further comprising. The Oxford Guide to Plain English suggests that a well-written sentence contain 15 to 20 words; by contrast, patent claims routinely exceed 200 words, and longer claims are common place. See, U.S. 7,593,860 (shortest claim is 1,019 words, longest claim is 1,053 words); U.S. 6,763,226 (2,552 words); and U.S. 6,953,802 (over 17,000 words!). For an insightful discussion of the history and alleged justification of the single-sentence rule, see, Stephen Scott, “An Appeal to the New Patent Office Directory, Repeal the Single Sentence Rule,” Patently-O, September 18, 2009.
The rigors of strict antecedent basis and the fear that a stray singular or an accidental plural will preclude infringement or lead to indefiniteness result in monstrosities such as the following:
a meta-policy for representing said network security policy, said meta-policy comprising:
an association with zero or more outcomes;
an association with zero or more relationships;
an association with zero or more network objects; and
an association with zero or more services;
wherein a relationship of said zero or more relationships is associated with at most one of said zero or more services and is associated with at most one of said zero or more outcomes, wherein a protocol of said only one of said zero or more services must match a protocol at said only one of said zero or more outcomes, and wherein said relationship associated with an initiator network object and a target network object
See, U.S. 8,074,256.
Kristina Osenga, in Linguistics and Claim Construction, insightfully observes that one of the inherent problems with the single-sentence claim form is “the difficulty of explaining anything complex in a single, concise, and comprehensible sentence.” Even something as simple as a “sealed” peanut butter and jelly sandwich is reduced to a tangle of geometry and foodstuffs:
1. A sealed crustless sandwich, comprising:
a first bread layer having a first perimeter surface coplanar to a contact surface;
at least one filling of an edible food juxtaposed to said contact surface;
a second bread layer juxtaposed to said at least one filling opposite of said first bread layer, wherein said second bread layer includes a second perimeter surface similar to said first perimeter surface;
a crimped edge directly between said first perimeter surface and said second perimeter surface for sealing said at least one filling between said first bread layer and said second bread layer;
wherein a crust portion of said first bread layer and said second bread layer has been removed.
See, U.S. 6,004,596.
If this is what it takes to describe a sandwich, can we really expect claims to complex technology to be easily understood by judges and juries? Indeed, the problems created by the single-sentence rule motivated Judge Plager of the Federal Circuit to suggest that “Because claims in U.S. patents are written using words and phrases that purport to be in the English language, it might help if the rest of English language practice was used: short declarative sentences, careful and precise phrasing, and so on.” See, S. Jay Plager, Challenges for Intellectual Property Law in the Twenty-First Century: Indeterminacy and Other Problems. Of import, the single-sentence rule is not required by statute or regulation, it is merely customary, and the Office could cease “to insist” upon its usage.
The second major source of confusion and dispute about claims comes from the doctrines of claim construction themselves. Over the years, the Federal Circuit has provided incomplete, conflicting guidance on claim construction, particularly regarding whether to rely on the specification or dictionaries as the primary source of the meaning of the claim terms (the literature is replete with commentary, and I will not repeat it here; an excellent summary is discussed by Osenga, above). That conflict, along with the rule that claim construction is a matter of law and subject to de novoreview, has contributed to a reversal rate of district court claim construction that has ranged between 20 percent and 50 percent over the past 10 years. And while Anderson & Menell have recently shown that reversal rates have dropped, they argue that the “Federal Circuit’s adherence to the de novo standard has frustrated district courts’ distinctive capabilities for apprehending and resolving the factual disputes inherent in claim construction determinations, undermined the transparency of the claim construction process, discouraged detailed and transparent explanations of claim construction reasoning, and produced alarming levels of appellate reversals.” Further, the Federal Circuit’s claim construction doctrines are at odds with well-established principles of linguistics, which explain how readers extract meaning from text. As Osenga explains in detail, “the Federal Circuit is disregarding many of the linguistic intuitions that everyday speakers and listeners use to understand language, which may lead to inaccurate claim interpretations and decreased public notice.” Glossaries will not fix these wounds.
A third source of potential confusion comes from the legal requirements for infringement that result in separate claims for apparatus versus method, and user versus manufacturer versus seller. Patent drafters specifically craft claims that can be directly infringed by an individual or other entity in different ways, such as by using a method, operating a system, or providing software for download. However, software products and services increasingly rely on interoperability and interconnectivity between multiple different entities, from the user of the device, through a network provider, to an online service, as well as hidden third parties, such as content delivery systems, advertising networks and so forth. Different claims are often drafted for each of these, with subtle but important differences in scope, such as whether an entity is “sending” or “receiving” or merely “displaying” a given item of information. These differences often lead to potential ambiguity as to which entity is performing which claim element.
The rules of claim drafting, claim construction and infringement result in not individual words or terms being unclear, but rather in ambiguity in entire clauses. In my experience, disagreements with examiners over the scope of the claim focus on the meaning of a particular clause or entire limitation of a claim, and whether the examiner’s interpretation of that clause or limitation as a whole is reasonable—not on the meaning of individual words.
Most patent attorneys and agents come from the sciences; they are not English majors or skilled grammarians, and yet are required to draft some of the most complex and convoluted sentences in the English language. As Justice Brown observed, “The specification and claims of a patent, particularly if the invention be at all complicated, constitute one of the most difficult legal instruments to draw with accuracy.” Topliff v. Topliff, 145 U.S. 156 (1892). If anything, this difficulty has increased in the past 120 years. For numerous reasons, which I will expand upon in my next post, no glossary can make up for the problems inherent in the nature of claims themselves, or the rules that dictate their complicated forms.