Swaroop Panda, Shatarupa Roy
Visualizations have pervaded many areas of digital life, such as social media platforms, where graphs, charts, infographics, and other visualizations are used for communicating information, presenting and defending arguments, offering greetings and pleasantries, and multiple other purposes. Often these purposes include not-so-benign activities such as misleading people, sneering, and trolling, among others. These visualizations are often designed and developed by individuals lacking formal training and are extemporized using arbitrary tools and technologies that are easily accessible and cheap. The visualizations may not necessarily represent the personal judgments or choices of the individual who designed them. They might have been hired by an organization to design such visualizations for an ulterior purpose. Or the individual may be developing the visualization for an organization of their own accord.
We characterize this class of visualizations as guerilla visualizations. We borrow this term from Jakob Nielsen's ideas on guerilla HCI . Why guerilla? Nielsen suggests that low-cost, low-resource, and simplified usability testing methods for companies can be a stepping-stone toward adopting more-sophisticated, resource-intensive usability tests. This is because the standard usability evaluation in the HCI research community is far too cost- and resource-hungry for companies, especially small, product-based startups, to complete. There are similar issues with visualizations: Finding it difficult to implement the well-established research methodologies, tools, and frameworks of visualization and interaction design research, individuals resort to low-cost, low-resource frameworks and methods to design and develop these visualizations.
This article aims to characterize guerilla visualizations and highlight the need to formalize research efforts around such visualizations in the research community. Formalizing such visualizations is important because of their impact on basic human communication, which in turn augments and evolves the role of visualization as a technology or medium. Visualization, as a result, isn't restricted merely to those writing research papers, pilots in a cockpit, or doctors examining medical records. Visualization has permeated social media and as such is a part of the digital lives of millions of people. It has been used to teach concepts, debunk fake news, defend political stands, and promote products. A recent example of this permeation is the rise of the slogan flatten the curve, which refers to the many plots, charts, and infographics on the Covid-19 pandemic data. With the rise of social media and the Internet, these ubiquitous visualizations thus make a good subject for study, reflection, and research in the research community.
Guerilla visualizations are found primarily on social media. They are by definition not clearly defined. Guerillas are members of small independent groups who take part in irregular fighting. The guerilla visualizations seen on social media platforms are created by individuals, or individuals representing an organization, whose formal education and backgrounds cannot be precisely captured. Some are more visually literate or statistically sound than others; they can include everyone from professional designers to data scientists to bankers and traders. Their identities are captured only by their respective social media profiles—in many cases they choose to remain anonymous.
Guerilla visualizations include charts, graphs, infographics, data memes, and any other visual presentation created for such unconventional reasons. As mentioned earlier, the purpose of these visualizations could be to communicate information, present or defend arguments, or start a discussion. They can also be used for nefarious activities, including trolling, sneering, and promoting misinformation. These visualizations can be found across the Web in an irregular manner; some of them remain obscure and are ignored, while others get enough attention to provoke a response, often another visualization intended to debunk it or to start a (visual) dialogue. The new visualization elicits the creation of more visualizations that advance the dialogue. The ubiquitous, irregular, and mutating nature of these visualizations is what makes them difficult to define and describe.
A motivation for studying these visualizations is to make visualization and interaction design research more comprehensive.
Guerilla visualizations are designed and developed using tools such as popular graphics editors (often visible from the default watermarks found in the graphs). These editors allow for easy graph drawing using drag-and-drop mechanisms and no coding. The data sources of these visualizations are almost always missing, with no references to or information on the data.
Working toward a more stable definition, guerilla visualizations:
- are found on and shared across social media
- have inconsistent features; they may be technically correct or incorrect and may or may not be visually pleasing
- have a mishmash of objectives that don't include specific goals or tasks
- are not found in scientific research papers, in cockpits, in business presentations, or in any other environment that is controlled and well defined; they are in the wild
- are not necessarily designed by UI/graphic designers, data scientists, or any design team but rather by individuals (who could well be designers or data scientists) on social media.
Nielsen's article  serves as a good introduction to the background and context of guerilla visualizations. Though the ideas in the article ideally should not be directly translated to the visualization, it is worthwhile to analyze and inspect some of them. The article suggests that an intimidation barrier is the reason why usability tests are not used in software development projects outside of academia and R&D. The barrier includes the cost (or perceived cost) of using these techniques—costs include the time taken to conduct the tests, and the remuneration to the study participants—and the perception that the techniques are intimidatingly complex. As an example, Nielsen suggests that a fuzzy logic GOMS model can prove intimidating to software developers. To address these concerns, he proposes the use of "discount usability engineering," the primary idea of which is to provide for good-enough usability frameworks, rather than the best. Nielsen's ideas are a good starting point because he analyzes the difficulties that HCI as a field runs into when transported out of academia to industry—and guerilla visualization, by definition, exists outside of academia.
Another tangential idea in the HCI community is that of conducting research "in the wild" . Research in the wild describes approaches to HCI and user experience research that are not derived from other, lab-based methods. An instance of such a practice is human-computer interactions in museums , away from conventional labs with an on-site focus. Social media, and the Internet in general, is wild, in the sense that it has massive outreach, connecting billions of users and their interactions and therefore cannot be tamed. The large numbers of studies that have been conducted on social media data use a minuscule subset of the Internet. Therefore, ideas from HCI in the wild are a good direction to explore for guerilla visualizations.
The primary reaction to such visualizations by many designers, researchers, and experts has been outright dismissal because they aren't well defined, lack credibility (perhaps because they emerge from anonymous social media handles), and are technically and ethically dubious. But these are also the exact reasons why researchers and experts need to study guerilla visualizations.
A motivation for studying these visualizations is to make visualization and interaction design research more comprehensive. The bulk of visualizations appear to be moving away from research papers and business meetings into social media. For the research community, each such visualization is a data point that deserves study, research, and reflection. The deluge of visualizations on social media is just another kind of big data. Ignorance of such a large quantity of data would necessarily prevent the research community from gaining a larger perspective on and understanding of the impact of visualizations. Another, parallel reason is that this flood of visualizations on social media could result in them becoming a primary mode of communication. This is more plausible owing to the rise of data-driven and "show me the data" thinking: You want me to change my views? Can you please show me the data? Visualizations are well equipped to deal with the rise of such appeals.
The study of fallacious visualizations is also an emerging research theme. A lot of work has been done on calling out the incorrect use of axes, the lack of accommodation for uncertainty, and the imposition of patterns onto rather arbitrary data points. Within the visualization design community, Alberto Cairo's How Charts Lie  presents many of the fallacies that creep into visualizations across the media. Cairo provides a description of such anomalies, prescriptions for avoiding them, and some limited examples (e.g., visualizations that are influential or emerge from important social media handles and thus have considerable impact), which closely resemble guerilla visualizations. Another important work is Michael Correll and Jeffrey Heer's "Black Hat Visualization" , in which they discuss the different ways that people lie and mislead using visualizations, including manipulating or obfuscating data. Such visualizations not only communicate incorrect information but also can potentially cause other damage due to erroneous inferences. The research, study, and practice of guerilla visualizations would then ideally attempt to address the above concerns by both studying the maladies contained within the visualizations and looking at the individual designer, the context, and the background of the visualizations' development.
Other reasons for pursuing research on guerilla visualizations could include ethical implications, doing social good with visualizations, or conducting visualization research in the wild where no standard rules apply.
The foremost challenge in guerrilla visualization research is the collection of data points for analysis. As stated earlier, the data is unevenly and irregularly distributed on the Web across several social media platforms. Study of the impact of these visualizations would entail studying the number of people they have reached and the subsequent response they have elicited. As a start, we can borrow existing data-collection strategies from social computing research. Researchers working with social computing regularly scrape data from social media sites and observe and extract patterns and influence, among other things. Data privacy remains an important consideration in such approaches; thus, ethics must also remain a focal point.
The visualization research community could also facilitate the creation of tools that aid in creating these visualizations. This does not include creating software that requires less or no coding. Such software is already freely accessible, but it can be as complex to use as writing code. Thus, design researchers may strive for the creation of tools for individuals, whose formal identities go no further than their social media handles. User-friendly tools coupled with social media platforms are one example. For instance, Twitter has a text box that allows the user to type 280 characters. What about introducing an interface (a feature) that allows the user to draw a chart on Twitter? Such tools could incorporate guidelines and best practices for designing, publishing, and evaluating visualizations.
Finally, there will be handles who knowingly publish visualizations to propagate an agenda—visualizations will continue to mislead and lie. The research and study of such visualizations should lead to improved guidelines for visualization literacy, which is an existing theme of study. Visualization and design literacy could act as insulation against visualizations that cheat and lie. Interesting questions to start with are how a particular visualization managed to convince people of an argument, or how a visualization managed to counter the opposition party's claims.
The above is a non-exhaustive set of research topics that can be pursued in guerilla visualizations. But these visualizations, by their own characteristics, can include many other tangential questions that could be pursued by the research community.
Ironically, formalizing research and study on guerilla visualization leads to its democratization. It legitimizes these visualizations within the research community, which means different types of inconsistent, in-the-wild visualizations gain acceptance and recognition. Further research on guerilla visualizations would give rise to new platforms and frameworks along with greater accessibilities because of the inherent nature of these visualizations (e.g., shared on social media, designed by various individuals, a mishmash of objectives).
Our goal in this article was to motivate the visualization community to study and research guerilla visualizations. We characterized guerilla visualizations, analyzed their properties, and described their connections to existing research. We discussed the motivation and challenges of instituting guerilla visualization in visualization research and concluded by proposing specific areas of research that can be pursued.
Swaroop Panda is a Ph.D. candidate in the interdisciplinary design program at the Indian Institute of Technology Kanpur, India. He primarily works with visualization, interaction design, and design fiction. firstname.lastname@example.org
Shatarupa Thakurta Roy is an assistant professor in the design program and the Department of Humanities and Social Sciences at the Indian Institute of Technology Kanpur. Her research interests include visual culture, Indian folk and minor art, graphic design, and design theory. email@example.com
©2022 ACM 1072-5520/22/01 $15.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2022 ACM, Inc.