Citation patterns in temporal United States patent data
Schumpeter (1942) identified innovation as a primary driver of economic growth. Integral to innovation is the creation of new machines, new methods of production, and new products by inventors. Throughout human history, new technologies have transformed societies (Lenski 1966). Economists have long studied links between innovation, technological change, and economic development. The importance of particular inventions depends on the consequences of implementing them. Estimating the magnitude of technological changes, let alone establishing their wider societal impacts, is extremely hard. Inventions are intrinsic parts of technological change with many inventions being patented. Since the pioneering work of Schmookler (1966), patents have been studied to gain more insight into technologically driven economic and social changes (see Griliches (1984,1990); Trajtenberg (2002)). Patents differing in importance provide indirect evidence regarding large and small sources of the impact of technology on economic change. One potential indicator of a patent's importance is the extent to which it is used (cited) by later patents. This had led to the development of indices of the importance of patents (e.g. Carpenter et al. (1981); Trajtenberg (2002)). We do not dispute the value of this line of work but adopt a complementary approach. Consistent with the objectives of this book, we try to understand the structure of patent networks while delineating temporal changes in them.
A patent is an intellectual property right granted by the Government of the United States of America 'to an inventor to exclude others from making, using, offering for sale, or selling the invention throughout the United States' or importing the invention into the United States for a limited time in exchange for public disclosure of the invention when the patent is granted. (USPTO, uspto.gov/patents/)
Patents fall into several categories. Approximately 90% of the patent documents issued by the patent office in recent years have been utility patents intended to protect non-obvious, novel and useful inventions belonging to four categories (more on these below): processes (methods), machines, articles of manufacture, composition of matter and improvements in any of these categories. Most patents are for incremental improvements to known technology although some patents have been revolutionary. Design patents protect ornamental features of articles of manufacture. 'New and distinct, invented or discovered asexually reproduced plant including cultivated spores, mutants, hybrids, and newly found seedlings, other than a tuber propagated plant or a plant found in an uncultivated state' are protected by plant patents under the 1930 Plant Patent Act. Reissue patents are granted to correct errors in already issued utility, design, or plant patents. There are items for which patents are never granted including abstract ideas, physical phenomena, and laws of nature. Also excluded are literary, dramatic, musical and artistic works - these can be protected by copyrights. Inventions deemed impossible, not useful, or offensive to public morality cannot be patented.
Lurking in the language for defining patents are seemingly clear words whose meanings required specification. The simpler ones are:
• new and novel: In the USA, to obtain a patent, the invention must never have been made public in any way in any place.
• original and non-obvious: This is assessed through a comparison of what is in a patent application and what is known already. Those making this assessment have a good knowledge of the relevant technical domain(s). The invention must be 'not obvious' in the eyes of relevant expert patent examiners.
• useful: An invention has to be a practical device capable of doing something.
Patents, as protections of intellectual property rights, have value for those holding them. Usually, this is relevant for commercial applications of patents. As such, patents have economic value by excluding other producers from applying the contents of patented items. When others do this anyway, it is a violation of the patent (patent infringement). These violations result in legal cases filed in the US court system (see below). Competing claims for different patents, challenges to existing patents and alleged infringements of patents lead also to court cases. Patents can be challenged by those seeing them as not being legitimately granted, as can rejections of patent applications. When cases are decided in court, especially the US Supreme Court, some terms in patent languages are changed. When relevant below, specific Supreme Court decisions are cited in our discussion of utility patent categories.
Our primary source here is uspto.gov/web/offices/ac/ido/oeip/taf/patdesc.htm (accessed 6/6/2013). All of the Star Wars characters have been patented, as have the distinctive appearance of athletic shoes. See inventors.about.eom/od/inventingl01patents/f/can_be_patented.htm, accessed 6/6/2013.
The exact (but not always unambiguous) nature of the four categories of utility patents was specified further in US Supreme Court decisions:
• Method: Gottschalk v. Benson (409US63, 1972) defined this by '(a) process [method] is a mode of treatment [series of steps] of certain materials to produce a given result. It is an act, or a series of acts, performed upon the subject-matter to be transformed and reduced to a different state or thing (items in square parentheses inserted)'.
• Machine: A machine is a concrete object with combinations of parts. From Corning v. Burden, 56US252, 1854), the specification of a machine 'includes every mechanical device or combination of mechanical powers and devices to perform some function and produce a certain effect or result'. Implicit in this specification is that the parts interact operationally.
• Article of manufacture: In Diamond v. Chakrabarty (447US303, 1980), the Supreme Court relied on a dictionary definition of manufacture (a verb) as 'the production of articles for use from raw or prepared materials by giving to these materials new forms, qualities, properties, or combinations, whether by hand labor or by machinery.' It follows that articles of manufacture resulting from production, in this sense, are tangible items or commodities.
• Composition of matter: 'A composition of matter is an instrument formed by the intermixture of two or more ingredients, and possessing properties which belong to none of these ingredients in their separate state. The intermixture of ingredients in a composition of matter may be produced by mechanical or chemical operations, and its result may be a compound substance resolvable into its constituent elements by mechanical processes, or a new substance which can be destroyed only by chemical analysis (Robinson 1890).' In Diamond v. Chakrabarty, the Court added that this term means 'all compositions of two or more substances and all composite articles, whether they be the results of chemical union, or of mechanical mixture, or whether they be gases, fluids, powders or solids'.
In this case, the Court ruled that, a proposed way for converting numerical information from binary-coded decimal numbers into pure binary numbers, for use in programming conventional general-purpose digital computers, was merely a series of mathematical calculations or mental steps: it was not patentable as a method.
This broad classification of utility patents is complemented by a detailed classification of technological content. There are more than 400 three-digit patent classes devised by USPTO with over 120,000 patent subclasses. Hall et al. (2002) (p. 415) state 'This system is continuously being updated, reflecting the rapid changes in the technologies themselves, with new patent classes being added and others being reclassified and discarded.' A footnote discusses granted patents being reclassified. As a practical matter, they developed a two-digit classification with 36 categories before constructing six main categories:
1. Chemical (with six two-digit subcategories: Agriculture; Food; Textiles; Coating; Gas; Organic Compounds; Resins and Miscellaneous Chemical);
2. Computers and Communications (with four two-digit subcategories: Communications; Computer Hardware and Software; Computer Peripherals and Information Storage);
3. Drugs and Medical (with four two-digit subcategories: Drugs; Surgery and Medical Instruments; Biotechnology and Miscellaneous Drugs and Medical);
4. Electrical and Electronic (with seven two-digit subcategories: Electrical Devices; Electrical Lighting; Measuring and Testing; Nuclear and X-rays; Power Systems; Semiconductor Devices and Miscellaneous Electrical and Electronic);
5. Mechanical (with six two-digit subcategories: Materials, Handling and Processing; Metal Working; Motors, Engines and Parts; Optics; Transportation and Miscellaneous Mechanical) and
6. Others (with nine two-digit subcategories: Agriculture, Husbandry and Food; Amusement Devices; Apparel and Textile; Earth Working and Wells; Furniture, House Fixtures; Heating; Pipes and Joints; Receptacles and Miscellaneous Others).
These classifications of patent classes and technological domains (with the detailed sub-domains) make it clear, purely in terms of intrinsic content, that patents and patent applications are parts of a complex intellectual arena. It is also one that can become very conflictual, given the potential economic stakes resting on the exploitation of patented inventions. Conflicts over patents arising in the USA frequently end up in the courts. Patent cases have been on the docket of the Supreme Court, as noted above. There is a whole area of Patent Law with specialist lawyers. Inventors seeking patents have to enter a multifaceted domain with features going well beyond the details of the inventions.
Patent applications are submitted to patent office's having authority to evaluate them in the context of prior patents (and knowledge). Patent officers issue final rulings within a patent office regarding the patentability of inventions. The entire process of applying for patents, assessing these patent applications and granting patents (or not) directly affects the form of patent citation networks. As described above, patents are sought for genuinely new inventions falling within one or more of the technological domains listed above. For all technological areas, their prior knowledge bases are the foundations for evaluating patents. Patent applications must be very precise documents stating clearly 1) how potentially patentable items build on prior knowledge, 2) how they are differentiated from this knowledge, 3) the exact nature of their novelty, and 4) the potential usefulness of inventions.
In contrast to scientific citation networks (Chapter 4), for which authors cite whatever earlier scientific papers they select, or the Supreme Court citation network (Chapter 6), where Justices can choose any prior Court decision they wish to cite (while ignoring other decisions with potential value as precedent), patent applicants are severely constrained. They must cite exactly the relevant prior patents and only those patents (directly relevant prior knowledge). The role of patent officers with special relevant knowledge is crucial. They interact with applicants and can insist on modifications to patent applications if relevant knowledge is omitted or extraneous knowledge is included. The resulting patent citations networks are cleaner and sparser. A strong case can be made for such citation data being more accurate than the other types of citation networks because patent officers are enforcers of content, especially citations. Allowing for some inventions not being patented, the resulting network is one where the network boundary has been well established. See Section 1.3.1 for a description of the network boundary problem and how this is solved for the patent citation data.
-  The patent citation dataset constructed by Hall et al. (2001) for patents granted between 1963 and 1999 created an important treasure trove for work in this area.
-  Chakrabarty, a genetic engineer who worked for General Electric (GE), developed a bacterium designed to break down crude oil. GE's patent application for this was rejected by a patent examiner. At that time, living things were not patentable. This case reached the Supreme Court. It ruled (5-4) a living, man-made micro-organism is patentable subject matter as an article of manufacture or composition of matter within the meaning of the 1952 Patent Act.
-  This can get quite subtle. American Fruit Growers, Inc. v. Brogdex Co (283US1, 1931) ruled 'the addition of borax to the rind of natural fruit does not produce from the raw material an article for use which possesses a new or distinctive form, quality, or property. The added substance only protects the natural article against deterioration by inhibiting development of extraneous spores upon the rind. There is no change in the name, appearance, or general character of the fruit'. It declared the invention of this patent application unpatentable and voided it.
-  The rationale for dividing utility patents into four categories was driven by a need to place inventions in them, perhaps uniquely. This particular statement by the Court serves to muddy the waters. Items such as fiberglass hulls for boats can be both an article of manufacture and a composition of matter.
-  When Googling terms such as 'patents', 'patent applications' and related topics, the results include a plethora of advertisements for lawyers, consultants, and organizations offering their services to inventors. Most often, these services are expensive.
-  Patent applications include statements of the technical problems solved and the operational details for using the invention.
-  There are no formal constraints on Supreme Court Justices on what they cite, beyond making compelling arguments for (or against) Court decisions. As shown in Section 6.6, ignoring precedent is possible (indeed, it occurs often). For academic journal citation networks, the interactions involve authors, reviewers, and editors. While these process can lead to adding citations (seen as relevant by reviewers) there is little pressure to remove tangential or irrelevant citations.