This article considers the Frankfurt School’s critique of capitalism--that technological rationality corrodes critical agency--from today’s perspective. There are three sections. First, I explain what instrumental reason, technological rationality and critical agency are. Second, I apply these ideas to today's political climate. Finally, I consider the limitations of the traditional methods of critical theory when analysing this climate.
1.) Instrumentalisation, Critical Agency and Technological Rationality
In any human action, there is an intention, which specifies the goal (end) of an action, and there is a method, a means of achieving the end. The question of what ends to pursue is one that requires a moral rationality—a system of thought able to assign goodness and badness (value judgements) to things in a logically consistent manner. The question of what specific means to follow requires instrumental reason—a system of thought able to understand and manipulate the facts of the world. The Frankfurt School argued that knowledge in capitalist society is biased towards instrumental reason over moral rationality. Though the technical knowledge (natural science, engineering etc.) needed to instrumentalise the world for any given ends is highly developed, the moral knowledge needed to determine appropriate ends is in a state of decline. Those disciplines and areas of culture which traditionally allowed for the development of moral rationality, such as philosophy and the arts, have become instrumentalised.
This process of instrumentalization could be seen in the philosophy of the Frankfurt School’s contemporaries, the logical positivists and their ‘rejection of metaphysics’. This rejection is based on the principle of verificationism, which argued that statements are meaningful if and only if they can be empirically verified, either through direct sensory observation or by experiment. Empirically unverifiable statements, including, amongst other things, morality, are nonsense. Moral statements, which are statements about goodness and badness, are unverifiable because goodness and badness are not things which can be observed. One can only observe that murder happens, not that murder is a bad thing. Horkheimer argues that the positivists’ verificationism is the consequence of a gradual separation of description (statements which merely describe/facts—such statements are empirically verifiable) from prescription (ought statements which imply value judgments). Older variants of philosophical rationalism, inheriting the classical tradition of Plato and Aristotle, tried to fit both description and prescription into a continuous, rational whole. Empiricist philosophy of the 17th and 18th centuries put a spanner to this project, with Hume demonstrating that there are no value judgements that may be validly deduced from a set of purely factual premises. Value judgements can only be derived from valued premises. The logical positivists complete the instrumentalising movement from a fact/value rational whole, to the separation of fact from value, to the erasure of value altogether as unverifiable nonsense.
The Frankfurt School saw the decay of moral reason as the decay of critical agency—the capacity of people to think and act with a system of thought in which both instrumental and moral reason are united. Critical agency is based on a ‘dialectical’ view of human agency, in which subject (person) and object (environment—both natural and social) are locked together in a mutually determining relationship. Critical agency consists of the struggle to reshape the environment to better suit rational ends. This struggle consists of being able to, a.) recognize the facts of the environment (the environment’s ‘established factual order’) through an instrumental reason able to explain, predict and manipulate those facts, b.) criticize the inadequacy of the established factual order via a moral rationality, and c.) transcend the established factual order to establish a new factual order better suited to rational human ends, using instrumental and moral reason as a unified system. If morality, and any talk about right and wrong ends, is nonsense, then critical agency becomes unthinkable, since rational transcendence needs moral reason. The instrumentalization of knowledge makes alternatives to capitalism unthinkable—capitalism is the established order of facts within which instrumental reason, without moral reason, is forever contained.
The development of technology under the constraints of market imperatives is also part of the same process of instrumentalization. Marcuse argued that the machine process of manufacture had extended beyond the economic base of society, restructuring every aspect of society in its image. The machine process refers to the total system of machinery and workers involved in production. The development of this system was caused by the pressure exerted on firms to become more and more productive—to produce more for less. This process, which lead to the industrial revolutions of the 18th and 19th centuries, started through specialisation, in which a good’s construction process was atomized into many constituent parts, where each part was carried out by a single specialized worker. Previously a single good would have been constructed by just one craftsperson. Specialisation made production more efficient. The limited number of tools used by a craftsperson exploded, with each constituent part of a good’s construction process demanding its own specialized tool. Eventually machines were invented to automate the specialized job of each tool, and each machine, along with human workers, was organized into a production line, forming the total machine process. Both Marx and Veblen noted that in the machine process, both machine and worker are of equal status as specialized instruments in a system whose only function is to make as much as possible as quickly as possible. Veblen noted that the worker is forced to act as a pure instrument, and this affects the worker’s thought. The worker thinks in the strictly descriptive, quantitative terms required by the machine process. This thought is also present in the consumer, who must adjust personal tastes to the standardized form of the machine process’ products. Veblen feared that this kind of thinking, which Marcuse called technological rationality, would corrode traditional modes of thought such as morality and spirituality. For the Frankfurt School, it corrodes critical agency, and technological rationality was spread throughout society through a vast bureaucratic network that extended from the managers and shareholders in private firms to the administration of the state—the management structure one might find in an office repeats the instrumentalization of people and efficiency distinctive of the machine process.
2.) The Acceleration of Technological Rationality
Three features of today show us that the expansion of technological rationality into every aspect of life is accelerating; a.) data harvesting b.) the destruction of the public institution and c.) the casualization of work.
2a.) Data Harvesting
The collection of data is central to the way profit is made today. Facebook and Google, for example, collect data from users to determine which audiences are likely to be interested in some advertisement. Advertisers then pay to be able to target their adverts to the most appropriate audience—known as buying targeted ad-space.
Platform companies provide a space equipped with a standardized set of tools. Any action done inside the space is done through these standardized tools. Uber provides drivers and passengers a common space, via the Uber smartphone app, which contains all the tools required for a functioning taxi service. Facebook provides a common space which contains all the tools necessary for people to share content with each other and for advertisers to use this content to target ads. Platforms are a means for companies to gather huge amounts of people into a controllable space where information can be harvested.
This harvested information can be used to make all kinds of inferences about the platform’s users—inferences can be made about their ethnicity, their sex, age, education, income, personality, political opinions and so on. Data harvesting means technological rationality is able to instrumentalise more aspects of individual existence than ever before.
2b.) The Destruction of the Public Institution
The governing principles of public institutions (such as those of the welfare state, e.g. the NHS) can be reduced to two sorts of moral principles. Firstly, there is the utilitarian principle of the greatest possible good for the greatest possible number. The amount of good, or utility, an individual has can be measured by his income, health, level of education and so on. The good of a society is taken to be the sum of individual utility, and the aim of a utilitarian public institution is to maximise the total utility in a society. The kind of institution which offers universal, free at the point of use, access to whatever service the institution provides broadly follows this utilitarian principle.
Secondly, there is the difference principle, in which the only unequal distributions of utility in a society permitted are those which benefit the worst off in society, provided that fair equality of opportunity (in which opportunities must be open to all individuals) is not violated in the pursuit of the difference principle. This kind of morality is behind welfare programmes aimed at providing services for people disadvantaged in some way, whether that disadvantage is in terms of illness, disability, unemployment, and so on.
Public institutions are supported by the state rather than the market. This is because the private firms which compete in the market follow the principle of maximizing profit, and there is always the danger that profit maximization will conflict with the utilitarian and difference principles—sometimes it is more profitable to ignore those principles rather than follow them. Crucial for the preservation of those institutions which follow the utilitarian and difference principles is the boundary, between the state-run public sector and the free market private sector, that prevents their mixture with profit maximizing companies.
Today this boundary doesn’t exist. For example, the NHS’s Clinical Commissioning Groups (CCGs), the bodies responsible for deciding what sort of healthcare services are needed by some local population, can buy services from the private health sector through a bidding process—companies compete to get a contract offered by a CCG. The NHS is no longer a purely public institution, it is somewhere in between private and public, and multiple reports that private service providers are not being properly regulated, that CCGs are restricting services to reduce costs, suggest that profit maximization is eroding commitment to the utilitarian principles the NHS was founded upon.
The public institution was one of the few barriers against the spread of technological rationality into every area of life. Those services which were protected by the public sector are now open to instrumentalization.
2c.) The Casualisation of Work
The recovery from the 2008 crash has come with a deluge of casual work—low paid, flexible, unpredictable and unguaranteed work hours with none of the privileges associated with traditional post-war waged work such as paid holidays, sick leave and so on. Srnicek and Williams argue that the casualization of work is the result of increased surplus populations—people unable to find paid work. The larger the surplus population, the more competition there is for work, the more people are forced to take on low quality casual work. The production of surplus populations is a routine part of the capitalist economy, rising and falling in tandem with cycles of demand. However, Srnicek and Williams argue that automation is one of the causes, independent of cycles of demand, of today’s surplus population. Technological improvements mean companies can manufacture more or provide more services without relying on lots of human labour. While unemployment isn’t a necessary consequence automation—for example the introduction of automation could allow a company to expand in size and so create more jobs—it is argued that the long term trend of jobless recoveries since the 1990s and the unemployment traps created by deindustrialization show that automation is a cause of today’s surplus populations.
The casualization of work is therefore a consequence of an augmented technological rationality which relies less and less on instrumentalized human workers.
3.) The Effects of Accelerated Technological Rationality: Echoes not Consensus
The Frankfurt School’s thesis is one of homogenization, in which everyone’s critical agency is erased by an instrumental agency whose logic repeats the logic of technological rationality. This means everyone is only able to think about the world in the functional terms of technological rationality, creating a cultural consensus where everyone agrees alternative rationalities unconstrained by market imperatives are unthinkable. The key assumption of this thesis of consensus is that the logic of technological rationality is repeated in the individual’s psychology—following this, one might expect that the current acceleration of technological rationality will result in the deepening of a consensus that doesn’t see beyond the present state of things. But polarization, rather than consensus, describes today’s political culture.
The acceleration of technological rationality, through the casualization of work and the destruction of the public institution, has lowered living standards. This has caused a lot of resentment against the current neoliberal factual order; hence Corbyn’s defeat of New Labour, Podemos, Syriza, Trump’s victory over Clinton, the shadow of fascism in Europe with National Front, Alternative for Germany, Golden Dawn and so on. The trend appears to be that urbanized, young, middle-income people tend to polarize towards the left, while lower-income people outside major city centres, especially in those places which were once centres of manufacture but have since been abandoned by state and capital since deindustrialization, drift further right. This has been accompanied by a whole host of other polarisations, as nationalist right-wing narratives lure the poorest in society by explaining their problems by scapegoating immigrants and foreigners, intensifying pre-existing ethnic tensions, while the left-wing urbanites respond with an essentialist identity politics, also intensifying pre-existing ethnic tensions. The fact much of the information about the world is mediated by social media platforms have exacerbated these polarisations. Information about the world is targeted according to inferred characteristics about the user, meaning urban middle-income people get exposed to only one kind of news coverage designed specifically for urban middle-income people, while rural lower-income people are exposed only to coverage designed for rural lower-income people.
Today, the instrumental reason of technological rationality is not simply repeated in individual psychology. The acceleration of technological rationality is causing an echo chamber effect, in which culture splits up into several distinct chambers of thought, where each chamber has its own common-sense understanding of the world. Each chamber develops a distinct language which immerses its members completely, so that they are surrounded by a world structured by the premises of the chamber’s common sense. Instrumentalisation splits culture into pieces, and individual psychology repeats the logic of the fragment of culture that surrounds it—much to the detriment of critical agency.
4.) The Effects of Echoes upon Critical Agency
Critical agency must always begin with a proper understanding of the established factual order of the social and natural totality that contains everyone, before it can apply the rationality of ends and produce a new factual order. The capacity of each person to gain knowledge about the world is constrained by the localized scope of her existence. The facts of everyday habit, language, routine, class, gender and ethnic background, one’s physical abilities etc., which only capture an extremely small portion of everyone’s natural and social totality, determine and limit one’s knowledge of the world. This idea is captured in Bourdieu’s notion of habitus, which is
…necessity internalized and converted into a disposition that generates meaningful practices and meaning-given perceptions; it is a general, transposable disposition which carries out a systematic, universal application—beyond the limits of what has been directly learnt—of the necessity inherent in the learning conditions.
For a person to loosen the constraints upon her knowledge imposed by her habitus, he must be exposed to experiences from outside his habitus. To maximize the exposure of each person to non-local experiences, communication must take place between all parts of society, so that different local experiences may be passed around everyone as quickly as possible—this allows the area outside one’s habitus to become known to those inside. With echo chambers, the area outside the habitus disappears, and this kind of communication becomes very difficult. The common-sense of the habitus restructures any potential communication about the outside according to its own consensual logic, transforming any potential genuine knowledge about the outside into a self-confirming repetition of the habitus disguised as outside-knowledge. To the person inside the echo chamber, the chamber appears infinite.
5.) The need to go beyond the methodological echo chamber of critical theory to understand echo chambers
The traditional methods of critical theory generally gather a small amount of data[1] (usually objects of culture such as texts, artefacts, face-to-face interviews, and so on) and then process it through highly detailed interpretations, which are descriptions of the possible shared meanings ascribed in everyday life to a cultural object. These interpretations are often informed by basic theoretical assumptions—an interpretation could assume the arguments of Marxism, poststructuralism, the Frankfurt School, and so on—so that making sure the interpretation stays relevant to whatever question is being pursued is convenient. A question about depictions of the working class in film is likely to assume Marxist arguments when interpreting the films. It is possible to deduce the existence and properties of macrosocial phenomena from interpretive methods despite small sample sizes, especially if interpretation is based on data gathered from a broad variety of walks of life, such as a study based on a combination of literary texts, newspaper articles, face-to-face interviews and online blogs.
However, the nature of culture today is that its online presence, on social media platforms, the circulation of articles, pictures and videos, and the ocean of comments from users, is one of culture’s most visible and influential parts. A lot of the references points made in standard, face-to-face interactions originate in online life. It’s the same with echo chambers—much of their content is spread online in the circulation of Youtube videos and comments. This presents a difficulty for the traditional methods of critical theory, as a lot of this online information is unsuited to analysis via the interpretive methods so typical of critical theorists. It is difficult, for example, to produce a highly detailed interpretation of the online comments one might find on Youtube or Facebook because the contents of such comments are usually too sparse, or too ambiguous in intent, or only representative of a fraction of a second of someone’s habitus.
For these kinds of information, one must gather and process a great quantity before it can be used to deduce things about social reality, which can only be done through ‘quantitative’ methods of data analysis. It is not surprising that most studies of online communication and echo-chambering use machine learning and statistical modelling to process tweets and follows to infer things about culture. It is therefore in the interest of critical theory to collaborate with disciplines which study and apply the various methods of data analysis and seek a methodological fusion between those methods and interpretive methods.
There is a prejudice within critical theory against using methods of data analysis for critical projects. Much of the prejudice is targeted at statistical modelling; before continuing I briefly explain the gist of statistical modelling. Statistical modelling is a method of uncovering patterns in data so that predictions about whatever the data is measuring can be made according to whatever patterns have been uncovered. Statistical modelling first isolates the different variables being described by the data (e.g. perhaps one has two lists of people’s ages and heights—the variables would be height and age). Certain assumptions about the relationship between the variables are made (one might assume that though age is not determined by height, height is determined by age), and then the data is organized in a way that can test whether those assumptions are plausible or not (one may decide to list the ages in ascending order in regular classes of five years, and list the heights for each class—if there is a clear pattern in this organization of data, e.g. if heights increase at a constant rate as one moves up the ascending classes of age—a positive correlation—then the assumptions made about the relationship between variables are plausible). If the test is passed, then these assumptions are used to make predictions (one could predict that someone aged between 20-25 will be taller than someone aged between 15-20 given a positive correlation between age and height under the assumption age is independent of height). If the test is failed, then different assumptions about the relationship between variables are made and tested continuously until the data supports the assumptions.
There are good reasons to be critical of statistical modelling—there is suspicion of the rhetorical usage of statistical results by governments and the media, and there are reservations about the inherent limitations of statistical modelling. Regarding the first point, there is within common sense thinking a tendency to be uncritical of statistical results. Saying ‘93% of x is y’ sounds authoritative, and this authority makes the user of statistical rhetoric able to get away with unclear definitions, selection bias, and ignorance of the assumptions made in any statistical analysis, which are not infallible despite the routine testing. This makes it easy to give questionable ideologies a believable image. Regarding the second point, there is the worry that the assumptions made during statistical modelling are not sociological in character—they are not assumptions about subjects in social interaction, they are assumptions about the mathematical relations between variables which do not necessarily represent social relationships well. The premise of this objection is that there is an excessive reduction of the subject in the conceptualization of variables—to support this it is often pointed out in statistical modelling variables are assumed to be independent or straightforwardly dependent (x depends on y but y does not depend on x), whereas in social interaction subjects are locked together in a mutually determinative, or ‘dialectical’, relationships (x depends on y and y depends on x). This is, to an extent, true. Though there are ways of statistically modelling mutually dependent variables (see ‘mutual information’), these methods are difficult and not often used in social research. However, none of this means statistical modelling can’t be used in conjunction with interpretive methods, and there are rigorously thought out ways of combining different methodologies (see ‘methodological triangulation’) to make up for the weaknesses in any single methodological approach.
Moreover, these sensible critiques of statistical modelling are often deployed as a justification for an uglier prejudice against any kind of method that requires ‘positivist’ formal rigour, even though these kinds of methods are needed to analyse much of the online presence of culture. The result is that all ‘mathsy’ methodologies, potentially of great use to cultural research and which could make up for the weaknesses of the traditional interpretive methods, get conflated and dismissed under the buzzword ‘positivist’. There is within critical theory a methodological echo-chamber, supported by the consensus that interpretive/qualitative = good and positivist/quantitative = bad. These deceptive equivalences are so engrained in the common-sense of critical theory that simply to speak its language risks reproducing the consensus. Any discussion of different methodologies from inside critical theory seems to circle back to these equivalences, such that it is impossible for the critical theorist to think about outside methodologies without the instinct that qualitative = good and quantitative = bad.
[1] ‘Data’ refers to information that has been organized in a way that makes it easy for people to process to start inferring things about reality. In data analysis (i.e. statistical modelling and machine learning) this could be a list of people’s height measurements organized in ascending order or according to age—unlike data analysis, in interpretive methods there is often no need for someone to organize information into data, since texts, paintings, and so on are already in a form designed to be meaningful to people.