Abstract
Lev Manovich’s latest book starts from an indisputable reality: the scale that contemporary cultural production has reached cannot be systematically analyzed with traditional research tools and techniques. To this end, Manovich proposes the adoption of a new hybrid paradigm, capable of combining humanities, social sciences, and data science. A new analytical model applied to cultural parameters through computational science. He calls this field of study “cultural analytics.”
Manovich is an eclectic theorist, a pioneer in the use of data science for the analysis of visual culture, capable of combining computation, media theory, and art history in his work. He is a polymath in an increasingly specialized academic world. The publication of The Language of New Media (Manovich, 2001) made him an influential figure in the field of communication. He offered the first rigorous and systematic theory of what, at the beginning of the current century, were regarded as the new media. The closest precedent to his current lines of research can be found in Software Takes Command (Manovich, 2013), where he discusses how the development of digital interfaces has changed the traditional idea of media. Following this line of thought, Cultural Analytics, published in October 2020, materializes a project he has been developing for years.
In this latest work, Manovich brings together more than a decade of research to present concepts and methods of computational analysis of cultural data, with a particular focus on visual media. Throughout the pages, he shows how cultural analytics, through quantitative methods, complements and extends the development of classic sociological and philosophical theories, such as Pierre Bourdieu’s work on taste, Georg Simmel’s theory of fashion, or Theodor Adorno and Max Horkheimer’s concept of the culture industry.
The book, divided into three parts, devotes its first chapters to discuss the shift from “new media” of the 1990s to “more media” of the 2000s, showcasing the evidence of why computational methods are needed to analyze contemporary culture. Today, cultural production has reached an enormous speed of appearance and reproduction, mainly due to user-generated content. Billions of people share cultural content and interact daily with it on the web, from photos and videos to text and music. The large scale of this data offers researchers the possibility of finding new cultural patterns. However, this challenge requires a new approach, capable of questioning the current concepts and methods of cultural analysis.
Undoubtedly, these techniques based on big data are already very present in our daily lives. They are fundamentally applied by the new cultural industry platforms, such as Netflix or Spotify. Manovich calls these practices “media analytics” (p. 54). They share the same core with cultural analytics: large-scale computational analysis of cultural artifacts and behaviors. However, their goals and motivations are different, as the former are driven by pragmatic business principles, such as targeting recommendations or creating content tailored to users. While humanities and computational social science research have been analyzing relatively small samples and focusing on historical and statistical data sets, the platform industry captures millions of pieces of data that are analyzed in real-time.
The second part of the book outlines a typology of objects of study that researchers can use to examine digital culture in its variants of digital media and artifacts, online, and physical behaviors, interactions, and events. To analyze these objects, we can create large data sets about them, but if we do not narrow them down from samples, we would end up with immeasurable results. Therefore, instead of being guided by criteria based on ideology, tradition, and intuition, Manovich believes that statistical sampling techniques are essential to find cultural patterns. A data set as large as possible will not only ensure that, but it will also allow us to observe the distribution of trends and identify outliers, which may be more interesting. Thus, while computational and statistical methods, as Manovich acknowledges, do not guarantee more objectivity, they can help to confront assumptions, biases, and stereotypes.
Finally, the third part provides the conceptual basis needed to explore cultural data sets using data visualizations, and then it focuses on recently developed methods for exploring image and video collections. Traditionally, information visualization has been able to reveal patterns and structures, but it has paid the price of extreme reduction. Now, the use of algorithms capable of processing very large data sets makes it possible to generate dynamic categories that are not explicitly defined. This implies that new cultural research projects can combine two directions of analysis: either starting from previous categories, or delegating the construction of categories to computational processes: Visualization methods allow us to explore large collections of visual cultural artifacts or samples from a cultural process without measuring them. In other words, we do not have to use either numbers or categories (. . .) Far from being only one of the tools in quantitative cultural analysis, visualization is an alternative analytical paradigm. (p. 184)
Precisely, the final reflection of the book revolves around the question of whether we can think without categories. In his answer, Manovich remains cautious and recognizes that, although we can unravel cultural patterns on a large scale, we must not forget that cultural data can only account for some aspects of cultural artifacts and their reception. Thus, the unique aspects of each cultural object, for the author, are as relevant as those shared characteristics that identify statistical patterns.
We can say that Cultural Analytics is a book about the new methodologies of our era, which intend to conceptualize and theorize about new realities of cultural analysis based on computational techniques. Its greatest virtue is, in my view, the ability to exemplify and refer in each explanation to research works, programs, and analysis companies or cultural databases, which are a reference for carrying out new research. It is an accessible book for those who are new to statistics or data science and will undoubtedly become basic reading for any curriculum that wants to embrace quantitative analysis in the study of cultural phenomena.
