Distraction Free Reading

The Use of Patent Information to Investigate Algorithmic Systems

During my master’s degree, I wanted to study the idea that some social media platforms had tools for guiding users’ behavior in a certain direction. Even though this is by now commonsense among communities interested in studying tech and society, it was a hypothesis that was not easy to confirm with empirical data. That’s when I decided to experiment with patent analysis.

In research I conducted for my master’s degree dissertation in social science in 2019, I analyzed patents from Facebook Inc. (now Meta) with the goal of identifying and categorizing technologies that were capable of modulating human behavior on the company’s social media platform (Machado, 2019). I had already been delving into the study of algorithmic systems and felt I lacked a methodology that would help me understand the workings of these systems, which technology companies strive to obfuscate. I used the patent applications filed with the American patent office, USPTO, as the main source of documents for my research. By reading the patents, I was able to get an in-depth knowledge of the operations and dynamics of the systems that structure the social networking platform. The documents explained how some models that structure the News Feed work, how the platform connects users and objects in the social network, and even what the platform takes into account to decide which search result will appear first.

From time to time, a patent for a new technology that is often perceived as scary is cited in newspapers and technology websites, such as the “workers cages” patent, registered by Amazon (Hamilton, 2018), or the car that would have screens playing content instead of windows, registered by Apple (Chadwick, 2022). On the other hand, the use of patents for research in anthropology and social sciences is not as common as I imagined (Reymond, 2021). Still, studying the information contained in these documents can be very useful for social scientists interested in the development of algorithmic systems and the broader question of how technical objects such as these can shape society.

Cover of a publication by the Experimenter Publishing Company, with the title written in red, "Science and Invention." An illustration of a light-skinned man wearing a blue suit appears below the title text, with the words "Build Your Own Television Receiver." The man appears to be working on a set of equipment that he will use to build his own television receiver.

November 1928 Cover of Science and Invention (via Wikimedia Commons).

Ways to Study Patents

Patents are a set of intellectual property rights granted by the state to an inventor or holder of an invention, which gives them the exclusive right to produce, use, and sell the invention for a set period of time. This right prevents others from producing, selling, or using those same products without authorization from the patent holder. In the US office, for example, every patent must contain a brief summary of the invention, a detailed description of the invention, one or more claims, and a list of other inventions that relate to it (USPTO, 2023).

Patenting digital technologies has had a significant impact on science. For Sergio Amadeu da Silveira (2014), patents are obstructions and their use in software has negative consequences for any entity other than the market. The author states that “granting a monopoly on the use of a logical arrangement for a period of 20 years” (Silveira, 2014: 577) can “block the scientific and technological use of fundamental solutions for the advancement of Science and Technology” (Silveira, 2014: 577). Additionally, this stimulates the race to patent obvious solutions only as a way to protect themselves from lawsuits from other companies and, above all, concentrate the power of research on companies that can invest in the legal structure of patenting technologies.

In the area of innovation, the study of patents is routine. For instance, at my own university, the Universidade Federal do ABC, there is a department dedicated to patenting, with mentors to help use various patent databases and assist students who want to patent their inventions. However, there is little interdisciplinary conversation between it and other social sciences and humanities departments, such as sociology, anthropology, communication, or the public policy sector. Much of the research with patent analysis uses data mining and other methods of automation to group, cluster, and map existing technologies as a way to develop an overview of inventions in different industries and to locate gaps where inventions can be inserted, or where there is a need to develop new technologies that have not yet been patented (Abraham BP, Moitra SD, 2001; Abbas, A., Zhang, L., & Khan, 2014). For researchers interested in using patents for other purposes—in my case, to critically analyze the development of technologies that work to modulate human behavior—these methodologies are not sufficient.

Patents are delicate documents to study. As a technological-legal text, the patent is not written in a manner that aims to ensure easy understanding for everyone. Rather, there are several nuances and jargon specific to the area. The documents can be specific and yet quite vague at the same time. As Aaron Shapiro, a professor of technology studies that uses patent-diagrams to analyze smart city technologies, describes them, these texts are both “precise enough to assert ownership over intellectual property and yet vague enough to capture unanticipated uses” (Shapiro, 2020: 752). The type of language used in the text might make it look like a puzzle of information, where some pieces are there just to prevent it from being put together quickly. The information in a patent often shows itself in layers and requires careful and critical reading. On the other hand, every patent document must follow a specific structure (stipulated by patent offices), which makes it easier to compare and aggregate information from different patents in a systematized way.

The systematization of patent writing can be used to our advantage. In the case of my research, which focused on the patents of Facebook Inc., as much as the patented innovations changed with each document, they were described from the base structures of the platform, such as the systems that work as the skeleton of the News Feed and the search functions.

To use patents as a research source, one must understand their contradictions, and their problems, and remember that they are filed for different reasons. The most obvious would be to protect an invention. Still, patents are also filed as an offensive to competitors—i.e. to prevent another company from developing a certain invention that is profitable (even if the patenting company has no plans to develop it). It is also very common for a company to file a patent just to add market value since even if an invention is not launched, its patent remains part of the company’s portfolio of intangible assets. For Shapiro (2020), a professor of technology studies who uses patent-diagrams to analyze smart city technologies, patents, when accepted and successfully filed, are performative, “confer ownership over future uses by communicating (disclosing) information” (Shapiro, 2020: 754), and are by nature anticipatory in format. The author brings geographer Kinsley (2010) and the concept of the politics of anticipation to assert that the dynamics of patenting innovations are part of “corporate micro-maneuvering to colonize possibility and potential” (Shapiro, 2020: 768).

Patents in the Social Sciences and Humanities

In recent years, from research using patents, it has been possible to understand recurring themes and topics in the development of Palantir‘s surveillance platform (Iliadis & Acker, 2021) and to show the ability to influence some human-machine interactions present in technologies developed for smart cities (Shapiro, 2020). However, the use of patents as documentation for academic research is not yet widely used by social scientists. Why is this so?

There is one relevant piece of information that may help us answer the question: the fact that a patent has been filed does not mean that the technology described in that document has actually been created or is in use. Purely by reading the patent, there is no way to get this information. So how useful would information about technologies that we don’t know exist or not—at least in their materialized form—be in studying their intersections with society?

This is an interesting question to raise. However, it is also in the field of contradictions where the importance of studying these documents can be found. For example: why did a certain company decide to invest money and manpower to register a specific technology? There are reports of professionals in the field who tell us that companies pay them to develop technologies to be patented just to increase their portfolio, without paying much attention to what has been developed (see for example Watkins, 2022). But if by grouping dozens of patents from the same company, we find the same logic, tool, and technology, over and over again, would this be the case? Noting that in recent years a company has increased the number of patents for a particular function of its platform is extremely relevant information as it may indicate the broader directions that are of interest to the development team of the company. Above all, it is essential to point out that patent research, for the social scientist, cannot stop only at document analysis. Documents give clues and show paths that can be followed for a more complete analysis of complex issues.

Observing what differs between the text presented as justification for creating and patenting that invention, and what the document actually proposes as an innovation, was also an interesting analysis as I was able to look for contradictions in the company’s discourse. I could identify, for example, that situations or tools that make the user not interact much on the platform are seen as a problem to be solved by multiple inventions patented by the company. The very act of typing is sometimes described as a “cumbersome input of multiple characters” (Lindsay & Rajaram, 2017: 2), and is shown to be one of the reasons for the invention of a new technology that is able to make assertive text suggestions for the user to choose from while simultaneously hiding the option for the user to use the keyboard to write their own text. An example of the patent text:

“For example, to generate a post using conventional techniques, a user often needs to input a significant number of strokes or button presses to enter the post content. This cumbersome input of multiple characters using Small input devices of a mobile computing device may discourage users from Social networking system use in part due to the inconvenience of typing a significant number of characters” (Lindsay & Rajaram, 2017: 2)

The image consists of an illustration from patent US20170206194A1, Determining phrase objects based on received user input context information (Lindsay & Rajaram, 2017). It demonstrates what an individual's feed might look like, with the individual in the example here being named Andrew Jones. Below the Post button, the text "Eating ice" appears, likely because Andrew had begun typing it. Other suggestive text is listed in the suggestions box, in the following order, "ice cream; chocolate ice cream; vanilla ice cream; delicious ice cream; ice cream sundae; ice cream cake."

Illustration from patent US20170206194A1, Determining phrase objects based on received user input context information (Lindsay & Rajaram, 2017). Creative Commons.

Additionally, as an example, multiple tracking systems to analyze a user’s emotion and location data are also suggested as part of the algorithmic system that has to be rearranged so that this patent’s technology works assertively. The patent text further states that what the company understands as a successful and valuable suggestion— that is, one that will lead to user interaction with the suggested buttons—takes into account the company, third parties, and advertisers’ goals.

The brief analysis of this patent information already gives us an idea of how we might study the ways that the company can—and does—use users’ alleged desires for efficiency and speed to develop technologies that increase the use of tracking tools that make it easier to direct users’ behavior in a way that is most valuable for the company and its advertisers.

Challenges of Studying Patents

There are multiple possibilities of information that can be absorbed from a patent. The contradictions present in studying them from the social sciences, in research that is not market-oriented, are also diverse. Shapiro (2020: 769) writes, “If patents are a weapon in corporate struggles to colonize the urban future, then they can also be used against their holders to articulate precisely where we should not be headed.” As I learned when I first began studying patents, the use of patent information is a way of taking ownership of a kind of information and knowledge that is mostly used for the benefit of the market, and further using it as both a source and a tool to develop a sharper understanding of what companies are proposing as possibilities of technological futures.


Abbas, A., Zhang, L., & Khan, S. U. (2014). A literature review on the state-of-the-art in patent analysis. World Patent Information, 37, 3-13.

Abraham BP, Moitra SD (2001) Innovation assessment through patent analysis. Technovation 21(4): 245–252.

Chadwick, J. (2022). Apple Car could be windowless, patent suggests | Daily Mail Online. https://www.dailymail.co.uk/sciencetech/article-10836423/Apple-Car-windowless-patent-suggests.html

França, R. O. (1997). Patente como fonte de informação tecnológica. Perspectivas Em Ciência Da Informação, 2(2). http://portaldeperiodicos.eci.ufmg.br/index.php/pci/article/view/636

Hamilton, I. A. (2018). People are horrified by an Amazon patent that puts workers in cages—But an Amazon exec said even “bad ideas” get submitted. Business Insider. https://www.businessinsider.com/amazon-defends-worker-cage-patent-that-was-recently-unearthed-2018-9

Iliadis, A., & Acker, A. (2021). The Seer and the Seen: Surveying Palantir’s Surveillance Platform. https://doi.org/10.2139/ssrn.4129583

Kinsley, S. (2010). Representing ‘Things to Come’: Feeling the Visions of Future Technologies. Environment and Planning A: Economy and Space, 42(11), 2771–2790. https://doi.org/10.1068/a42371

Lindsay, R. T., & Rajaram, G. (2017). Determining phrase objects based on received user input context information (United States Patent No. US20170206194A1). https://patents.google.com/patent/US20170206194A1/en?oq=+United+States+Patent+Application+20170206194+

Machado, D. F. (2019). MODULAÇÕES ALGORÍTMICAS: uma análise das tecnologias de orientação de comportamento a partir das patentes do Facebook [Dissertação (Mestrado)]. Universidade Federal do ABC.

Reymond, D. (2021). Patents information for humanities research: Could there be something? Iberoamerican Journal of Science Measurement and Communication, 1(1), Article 1. https://doi.org/10.47909/ijsmc.02

Shapiro, A. (2020). ‘Embodiments of the invention’: Patents and urban diagrammatics in the smart city. Convergence: The International Journal of Research into New Media Technologies, 26(4), 751–774. https://doi.org/10.1177/1354856520941801

Silveira, S. A. da. (2014). Entre Software e Genes: A resistência ao paradigma do conhecimento patenteado │ Between software and genes: Resisting the patented-knowledge paradigm. Liinc em Revista, 10(2), Article 2. https://doi.org/10.18617/liinc.v10i2.753

USPTO. General information concerning patents. https://www.uspto.gov/patents/basics/general-information-patents

Watkins, N. (2022, March 15). Inside Big Tech’s Race to Patent Everything. Wired. https://www.wired.com/story/big-tech-patent-intellectual-property/

Leave a Reply

Your email address will not be published. Required fields are marked *