Patent proliferation kills diffusion – Protections fail to account for all of the patents, docs, and users.

Why Patents Might Not Provide Good Fences? In this section we discuss the three conditions introduced above and which are necessary in order to ensure good patent information quality. A first necessary condition is that patent databases must make it possible to easily identify relevant patents. Each user of the patent system must be able to easily find the information that it needs, and that allows him [the user] to take relevant strategic decisions. For example, a firm that is looking for a partner with technological complementarities must be able to browse patent databases in order to identify that partner. In other words, patent databases must make it possible to identify technological neighbors at a minimum cost. Similarly, if one firm aims at launching a new product on the market, it must be able to easily identify potential patents that it could infringe, in order to negotiate licenses ex-ante. However, this is not exactly what patent databases offer. Existing patent classifications, such as the International Patent Classification (IPC) for example, are useful to take strategic decisions about the identification of potential partners or potential future technological avenues (technological forecasting). Based on patent databases, firms increasingly perform patent mapping and patent landscaping, i.e. they organize patent information in such a way as to understand what is happening in a given sector, in a given company, or with a given technology (Lee et al., 2015). But these classifications do not allow the identification of patent neighbors from a legal point of view. They do enable patents to be found in similar technological classes, but they are of little help in performing exhaustive freedom to operate analyses, i.e. identifying patents that a firm might infringe. So far, there is no reliable patent index that allows users to find adjacent inventions. For example, in the case of land ownership, in most developed countries there is a public land registry that allows with certainty, and at a minimum cost, the identification of adjacent neighbors. Investors can easily identify stakeholders with whom they need to negotiate in order to develop real estate projects. There is nothing comparable for patents. This is highly problematic since, as we will see below, the absence of precise patent indexation favors hold-up behavior and patent trolling.

The issue of patent classification and indexation naturally escalates with the existing number of patents (Mulligan, Lee, 2012). The more patents in a sector, the more important it is to have an indexation system that enables users to easily find relevant patents. In the absence of reliable indexation, each user must analyze all the patents in force in the economy. Thus, the more patents there are, the more patents firms have to analyze, and the more costly this is. This explains why patents are so problematic today in sectors where millions of them are sometimes in force, such as ICT. In these sectors, due to the absence of reliable patent indexes, it is highly costly to identify patents that a firm can infringe, thus inducing a proliferation of costly litigations. In the case of software, for example, Mulligan and Lee (2012) performed an original simulation exercise which shows that given the number of software patents in the USA, the number of firms developing software, and the time required to analyze each patent, if each firm wants to perform an exhaustive freedom to operate analysis before it launches new software, it would require more patent lawyers than there are in the US. This simple simulation exercise demonstrates the absurdity of the situation nowadays. In some sectors where huge numbers of patents are in force, it becomes rational for firms not to perform freedom to operate analyses. Since these are costly and hardly reduce the probability of future litigations, it is rational for firms to keep their money aside in preparation for future litigation. Although rational from the firm point of view, these behaviors and the proliferation of patent litigation they induce entail a huge cost for society. The second necessary condition to ensure good patent information quality is that the boundaries provided by patents (the claims) and the description of the information that they contain must be understandable. When reading a patent, users must be able to understand what is protected by this patent (the borders or boundaries), and to understand the description of the invention that is patented. However, this condition does not often hold. Many patents are difficult to read and understood, even by experts. The technical information included in the patent is often of little help to readers. Worse, the perimeter of protection of the patent is often hard to decipher.

This difficulty in reading and understanding patents has many origins. First, it is sometimes difficult to describe with precision technologies that are mostly based on tacit knowledge and know-how. It is difficult to place boundaries on things that are hard to describe. Related to this problem, our language and words often lack the precision to define things. Thus, in sectors where knowledge is more codified, such as chemistry, where a technology can be defined by a mathematical formula, things are clearer and patent information is of better quality. But in sectors where technologies must be defined by words, there is still a significant element of uncertainty due to the ambiguity of these words. This is reinforced by the length of some patent documents, which are sometimes more than 100 pages long (Guellec, van Pottelsbergue de la Potterie, 2007). This length is supposed to increase the precision of the patent by describing each part exhaustively. But in practice it often increases confusion and uncertainty. Furthermore, boundaries also lack precision because, contrary to land, which is fixed, the knowledge space is moving, constantly expanding, and seeing the introduction of new concepts. And, as Menell and Meurer (2013) put it, it is always more difficult to “map terra incognita”. Finally, on top of these problems that depend on the nature of knowledge of technologies, the behaviors of patent users also contribute to blurring patent information. Sometimes borders are voluntarily kept unclear by patent applicants themselves. It can indeed pay off for patent owners to generate uncertainties for their competitors, as illustrated by the example of pre-emptive patents (Guellec et al., 2012).

Consequently, even when relevant patents can be identified (the technological neighbors), it might be difficult (even for experts) to understand their perimeter of protection exactly, and whether or not a given organization infringes these patents, thus largely increasing the cost of freedom to operate analyses. An emblematic illustration of this point is provided by the case of Polaroid versus Kodak in the mid-1970s, concerning instant cameras. Everybody remembers the outcome of this case and how costly it has been for Kodak. But, as mentioned by Menell and Meurer (2013), few remember that Kodak, before putting its technology on the market in order to compete with Polaroid, did hire a top patent lawyer in order to work closely with Kodak’s engineers and to avoid infringing Polaroid’s patents. It was only after lawyers guaranteed that Kodak’s technology was non-infringing that Kodak decided to enter the market, with the dramatic consequences that are now well-known. And ironically, Menell and Meurer also recall that in the ruling of the case the US judge acknowledged this fact and almost congratulated Kodak for its behavior. But at the end, the judge strongly disagreed with Kodak’s lawyers. Therefore, this historical example nicely illustrates the difficulties in defining the perimeter of protection of a patent exactly, and how people can easily disagree on this. And this uncertainty with regard to patents’ perimeter of protection obviously makes it difficult to navigate the patent ticket and develop safe technological collaborations.

Finally, the third necessary condition for good patent information quality is that patent stakeholders (owners, licensees, etc.) are easily identifiable by the users of the system. If a firm wants to negotiate a license, it must be able to identify the owner of the patent. Similarly, in order to use patent information for strategic planning, for example by analyzing the patent portfolio of a competitor or the evolution of patent applications in a given sector, one needs information about who really uses the patents. Yet in most countries today, this information about who are the real users is not easily available. If the applicants and the owners are listed in the patent document and are therefore public, there is no obligation for firms to disclose changes after the deliverance of the title. Information about licenses and even exclusive licenses, for example, can be (and usually are) kept secret. It is the same when a patent holder goes bankrupt and its patents are bought by other companies. And, on top of that, it is also possible for patent owners to rely on shell companies that make it possible to hide the real owner of the patent. In most cases, the real stakeholder of a patent is therefore very difficult to identify. This problem is known in the literature as the “who owns what” problem or “patent recordation failure” (Chien, 2012). In the case of the US, Chien (2012) concluded that “under the current rules, it is impossible to know who owns a particular patent, or what patents a particular entity owns”. To illustrate this result she showed that, based on an analysis of 915 patent litigations in the US, in a third of the cases, “the plaintiff was not the patent owner of record the day the litigation was initiated” (Chien, 2012). Given this situation, it is therefore highly problematic for users of the patent system to identify who are the real stakeholders. This poor information seriously limits the value of patent databases to serve firms’ strategic decisions but, most of all, benefits patent trolls and contributes to provoking litigations, as we will see in the next section.