Skill and deskilling in an uncertain agricultural world: Introduction to the special issue

Four perspectives

What is skill; what does it mean to be skilled? In everyday usage, ‘skilled’ is synonymous with ‘good’ in the sense of ‘competent’: a good farmer or a good researcher is someone recognised as being capable and effective in their farming or research. But this focus on the successful outcomes of an activity or occupation does not tell us much about skill.

Long-standing debates within sociology, anthropology, education, economics and STS have explored the nature and meaning of skill, the measurement of skill, and trends in and factors affecting the types, levels and distribution of skill. In his paper, ‘What is skill?’, Attewell (1990) distinguished four perspectives, which he termed positivist, ethnomethodological, Neo-Weberian or social constructionist and Marxist.

The positivist perspective portrays skill as an attribute of individuals or tasks, which has an objective character independent of the observer, making it amenable to quantitative measurement. Unfortunately, the positivists’ effort to quantify has been hampered by the diversity of skills and the qualitative differences among them: for example, how can physical skills be compared with mental or social skills? This qualitative diversity complicates the calibration of a ‘common yardstick’ needed to make different skills commensurate, and thus comparable in economic value.

According to Attewell, ‘at the core of the [ethnomethodological] perspective is the idea that all human activity, even the most mundane, is quite complex […] requiring complex coordination of perception, movement, and decision, a myriad of choices, and a multitude of skills’ (Attewell, 1990: 429–430). Advanced skills involve a depth of perception to recognise patterns and confidence to choose or develop appropriate responses, where ‘skill inheres in […] recognizing something new as something old, in acquired or trained “blindness” to uncertainty and uniqueness’ (Attewell, 1990: 433).

The Neo-Weberian or social constructionist perspective ‘tries to understand the conditions under which occupations are socially demarked as skilled and the processes by which some jobs come to command higher standing than others’ (Attewell, 1990: 435). Those processes might be market-based, that is, where the supply of and/or demand for skills are influenced by social action, such as through guilds, apprenticeships, unions, secrecy and so on. The socio-economic status of skilled work is also shaped by ideological claims about the value attributable to different kinds of work, reflected for example in distinctions between ‘professions’ and ‘occupations’, and between men's and women's tasks.

Marxism is Attewell's fourth perspective, and he starts by suggesting that many Marxist scholars treat skill as a ‘common sense’ category requiring no further articulation, while some neo-Marxists might ‘shade into either positivist or social constructionist thinking’ (439). Skills were nonetheless important to the Marxist analysis of commodification (of labour) and alienation (of workers from the products of their labour).

Skill as human capital

The positivist perspective on skill is evident in the economic concept of ‘human capital’, which can be thought of as the stock of habits, knowledge, skills, social and personality attributes (including creativity) embodied in an individual's ability to perform labour so as to produce economic value (Davis et al., 2021). At a macro-level, it is widely accepted among economists that the sum of a nation's human capital is a key determinant of national economic growth and development.

The concept of human capital remains central to mainstream development thought, as illustrated by initiatives such as the World Bank's Human Capital Project and its Human Capital Index (HCI) (The World Bank, 2018). Economists predict that investments in education and training will yield economic dividends in job creation, income generation, technological innovation, economic growth and human wellbeing. The positivist perspective on skill as human capital is reflected in policy rhetoric about modernisation, economic growth and structural transformation, and in talk about increasing human capital via education, training, learning, skill development, capacity building and so on.

Human capital theory assumes that skill is a personal attribute that accumulates within people over time, implicitly measurable as the sum of years of basic, higher and vocational education and training, plus years of on-the-job training and experience. Since skills are embodied within people, they can be carried by workers from one job to another, between organisations, economic sectors and countries. Wages are assumed to be proportionate to human capital, so that higher skills command higher wages as a return on investment in education, training and experience. The human capital of organisations is the sum of the skills of the individuals working for those organisations.

The human capital approach is very much in evidence in a recent FAO and IFPRI report entitled Investing in Farmers: Agriculture Human Capital Investment Strategies (Davis et al., 2021). The authors operationally define human capital ‘as the skills and capabilities of small-scale agricultural producers to successfully manage farming enterprises’ (5). The motivation for focusing on human capital is revealed by the report's emphasis on ‘disruptions’ – from climate change and COVID-19, to droughts, digital agriculture, and ‘advanced technology’ – leading to an expectation that ‘re-skilling and upskilling programmes must be part of the response to address the needs of agricultural workers facing such disruptions’ (9) (also see The World Bank, 2019).

Skill as autonomy

Perhaps the most consequential work on skill from the Marxist perspective is Harry Braverman's book Labor and Monopoly Capital, published in 1974. Braverman, like Marx, used the craft worker as the benchmark for skilled work, as ‘the combination of knowledge of materials and processes with the practiced dexterities required to carry on a specific branch of production’ (Braverman, 1974: 443). In this conception, a craft worker has autonomy and exercises control: s/he ‘decides how to accomplish a particular piece of work, chooses the appropriate tools and procedures, and is self-directed in the work’ (441). The capitalist factory system decomposed the production process into discrete tasks and imposed a division of labour that obliged individual workers to focus on one specific operation. In effect, conception (and thus control) was separated from execution, and management became a separate professional task with a skillset of its own. The linking of skill with autonomy is critical: for some who share this perspective, ‘if workers do not decide on what tools or methods to use to accomplish a task and if they cannot schedule what to do and when, they lack not only control but skill’ (442). In the factory context, the workers’ loss of oversight and control is equated with a ‘deskilling’ of work and workers (a term that Braverman himself did not use).

Form (1987) reminds us that ‘For over two centuries social scientists believed that the mechanization of labour and the factory system speeded up the division of labor, diluted workers’ skills, and increased their unhappiness’ (29). Braverman re-ignited the modern debate about deskilling, arguing that:

…not only does their [workers’] skill fall in an absolute sense (in that they lose craft and traditional abilities without gaining new abilities adequate to compensate the loss), but it falls even more in a relative sense. The more science is incorporated into the labor process, the less the worker understands of the process; the more sophisticated an intellectual product the machine becomes, the less control and comprehension of the machine the worker has. (Braverman, 1998: 294)

It should be noted that Braverman's fundamental observation, and the implications he drew from it, are not universally accepted. Indeed, they are contested with regard to whether, where, when and under what conditions deskilling was and is observed, and if so, the factors driving the deskilling process (for critiques of Braverman's arguments about deskilling see Attewell, 1987; on the deskilling debate more generally see De Pleijt and Weisdorf, 2017; Form, 1987; Myles, 1988).

Skill as status

The ethnomethodological perspective leads to the recognition of an important paradox about skill. As soon as a person can do something fluently and with apparent ease, the skill itself seems to disappear. According to Attewell, ‘skill inheres in the ability to do [an activity] without thinking about it’ (Attewell, 1990: 433). This happens to different degrees for practitioners themselves and for people observing the activity. Most able-bodied adults think nothing of being able to walk down a flight of stairs, nor are they especially impressed to see another adult do the same. But this activity requires advanced motor skills that had to be learned, and with some difficulty; once learned, they became ‘second nature’ and largely vanished from sight. They became embodied within their practitioners, subconscious for the walker themselves and unremarked by most people watching.

More esoteric and unusual skills might remain conspicuous, both for the practitioner and other people. Professional dancers and barristers are well aware of their own advanced skill. This is a function of three factors: the amount of time and effort it took to acquire and sustain the skill at a high level; comparison with others working in the same field; and the existence of institutions that signal appreciation of particular skills with material rewards and cultural recognition. Nevertheless, it is easy to see why skill becoming invisible or buried would be highly problematic if, like the positivists, you want to measure skill.

The Neo-Weberian or social constructionist perspective suggests that professional or occupational status depends more on power and political bargaining than on objectively different levels of skill. A prime example is the relative status typically accorded to managerial work vs. manual work. The higher status of management has less to do with higher levels of skill, and more to do with the carefully created and maintained belief that oversight, supervision and higher responsibility for distributing group-level resources constitute higher levels of skill. In this discourse, responsibility for overseeing and coordinating the disparate parts of a production process is the counterpart of the deskilling that Marxists perceive in the plight of workers consigned to performing discrete tasks within that larger process.

Technography, which is an ethnomethodological approach to studying and analysing practitioners of technology and their skills, offers an alternative view. It sees the unequal distribution of power between management and workforce as an effect of ‘task ordering’, whereby the performance and coordination of tasks in time and space often entails an uneven distribution of information and knowledge among those performing the tasks. This enables power to accumulate to personnel occupying key positions within task groups, such as overseers, managers and the controllers of tools and resources necessary for the task. According to this interpretation, while skill appears to be a product of talent honed by training and experience, it is really an effect of task performance, which bestows privileges and status on actors who perform certain roles (Arora and Glover, 2017; Jansen and Vellema, 2011). In the agricultural realm, technography has been used to analyse a range of cases including for example oil palm processing groups in Ghana (Adjei, 2014) and precision agriculture in The Netherlands (Velden et al., 2024).

The technographic perspective on skill in small-farm contexts emphasises the real-time decision-making and performance skills of farmers and labourers, including a focus on improvisation and coping in the face of environmental, market and other uncertainties. These performance skills are embodied, practical and tacit rather than articulate. A good outcome (i.e. a skilful performance) depends on adjusting nimbly and effectively to a dynamic context. This kind of skilful improvisation combines a sensitivity to environmental signals with an ability to choose and enact a repertoire of practised techniques and routines to suit the circumstances (Flachs and Richards, 2018; Glover, 2018; Richards, 1986, 1989). Emery Roe has compared the improvisational performance skills of pastoralists with the skills of professional managers of complex industrial and infrastructural systems. In both cases, achieving successful outcomes depends on being able to vary management processes considerably to cope with varying conditions, in order to achieve outcomes that are as stable and reliable as needed (Roe et al., 1998).

Four cases

Here we refer to four cases that illustrate some of the ways that the notion of deskilling has been brought into the analysis of agricultural change. The cases comprise hybrid maize in the USA, contract farming of bananas in the Eastern Caribbean, contract broiler chicken production in the USA, and transgenic cotton in India.

Over a 10-year period from the mid-1930s, hybrids replaced open-pollinated maize varieties throughout the US corn belt. Fitzgerald (1993: 327) argued that ‘hybrid corn was an agent by which farmers were effectively deskilled’. Significantly, Fitzgerald's focus was not on the replacement of manual skill by machines, nor the simplification of complex, skilled tasks by fragmentation of the labour process. Rather, she was concerned with ‘mental skill’, specifically ‘The knowledge farmers had created over the years – regarding the differences between various lines of corn, their growing range, maturing rates, value as livestock feed or on the open market, susceptibility to bugs or disease, and in particular the visible features that good corn should have for their particular farms’ (343). Fitzgerald argued that hybrid varieties, as well as the manner in which they were marketed, made that kind of know-how obsolete for farmers, as their ‘authority and knowledge were thereby delegated to geneticists and seed dealers’ (343). This left Midwestern corn farmers unable to understand some aspects of their own operations without expert help. Notably, while Fitzgerald felt that this was clearly a case of deskilling, she pointed out that, unlike many industrial examples, farmers did not struggle against the introduction of hybrid corn technology, and that corn growers generally benefited from growing the new varieties. For Fitzgerald, the value of deskilling is as a heuristic device for understanding work, which can be useful ‘even when shorn of its political charge, and can reasonably be extended to work sites off the shop floor’ (325), such as farms.

Grossman's study of the deskilling of banana contract farmers in the Eastern Caribbean (Grossman, 1998) engages more directly with the central concerns of Labour and Monopoly Capital. Grossman wrote that the study's ‘labor question’ was to investigate ‘the extent and import of control exercised by capital and the state over the peasant labor process’ (3) as a result of contract farming. The separation of conception from implementation is one of the indicators identified by Attewell (1990) as a signifier of control moving from craft producer to capitalist master. Contract farming, in general, had already been identified as an important context for the study of agricultural deskilling through its impacts on producer autonomy and control (e.g. Mann, 1990; Watts, 1994), although the category of contract farming is very diverse and heterogeneous, and must be approached with care.

Grossman found limited evidence of deskilling. First, the technology employed in banana production in the Eastern Caribbean became more complex over time, which required increasing levels of technical skill on the part of banana growers. Second, capital and the state did not control all aspects of the production process uniformly: apart from harvesting and packing, peasants retained considerable autonomy over production. Third, farmers had been obliged to readjust how and when labour was used on the farm, which was ‘an element of conception ignored in the deskilling debate in relation to contract farming’ (125).

Grossman concluded that ‘caution is warranted in uncritically adoption the concept [of deskilling]’ (25). ‘The basic problem in applying perspectives developed to analyze changes in industry to the realm of agriculture and contract farming is that the labour processes in industry and agriculture are fundamentally different, a function of the environmental rootedness of farming (209). And it is because of that environmental rootedness that ‘farming requires a much more complex set of responses from growers than assembly line processes do from deskilled factory workers; farming is not a repetitive motion exercise’ (125).

Elizabeth Miller (2018) used Braverman's work to analyse contract farming of broiler chickens under monopoly and monopsony conditions in the USA. She undertook detailed ethnographic research on two ‘industrialized contract broiler farms’, which she described as:

a model where farmers agree to raise chickens for meat for a set amount of time, at a rate of pay based on the ratio of feed to chicken weight at slaughter. Farmers invest in the built infrastructure to execute this process, but the company they contract for [the integrator] is mostly in control of the upstream and downstream supply and processing chains that depend on the production of the broiler chicken for their continued functioning. (iii)

Miller identified several factors that contributed to deskilling, including the introduction of off-farm inputs and experts, corporate concentration and changes in economies of scale. She found that the integrators actually wanted farmers to be deskilled, as reducing the variability of farmers’ labour processes helped to create more certainty in the production system. Deskilling would allow them to reduce farmers’ pay because ‘so much relative value is created outside of the farm, on grandparent flock farms, in laboratories, at the further processing plant, in the marketing department, and so on’ (166).

Glenn Stone analysed the case of transgenic, insect-resistant Bt cotton in India (Stone, 2007, 2011; Stone et al., 2014). He observed that cotton farmers in different villages of Andhra Pradesh (in a locality that is now part of Telangana) were exhibiting remarkable crowding behaviour in their seed choices. In a context where many new Bt varieties were being marketed, with relatively little information about their characteristics or relative performance, highly localised seed fads saw farmers in villages just a few kilometres distant from one another planting different cotton varieties in the same season, despite very similar growing conditions. Farmers were also rapidly switching from one seed brand and variety to another, and not, as might be expected, converging over the space of a few seasons towards a common seed choice. This seemed to contradict the expectation that farmers would make rational economic calculations based on the expected agronomic performance of the available seeds. Further, Stone observed that the farmers were being bombarded with marketing and advertising, that many of them relied on seed dealers to recommend which seeds to plant, and that many farmers were mistaken about the genetics of the cotton varieties they were ‘choosing’.

From these observations, Stone concluded that in this particularly dynamic context farmers were being deskilled because of shifts in the way they were learning about the new cotton varieties (see next section on Deskilling and learning). However, he noted that ‘crop genetic modification is not inherently deskilling’ (Stone, 2007: 67). The links between learning, decision making and technology, have also been explored for other contexts and crops (Flachs and Stone, 2019; Glover et al., 2020), and are further developed in Stone's contribution to this Special Issue.

Deskilling and learning

Two developments in relation to agricultural knowledge production and learning are relevant. First, at an organisational level, ideas around innovation systems and knowledge systems emphasising interaction, networks and links among diverse actors have been brought squarely into agricultural development (Hall et al., 2003; Roling and Engel, 1991; Sumberg, 2005). This perspective highlights the presence of multiple sources of information, knowledge and innovation (Biggs, 1990), including farmers, the private sector, universities, NGOs, research institutions and extension.

Second, at the farm level, Stone (2016) set out the foundations of what he called a general theory of agricultural learning, including three modalities of learning. Stone uses environmental learning to refer to adaptive experiment-based learning, which involves ‘observing and basing decisions on empirical payoff information’ (6). This is the kind of learning that is highlighted in the literature on farmers’ experiments, but as Stone argues, there are inherent problems with environmental learning that limit its contribution to knowledge generation. Social learning involves changes to practice based on evaluation of someone's social status, rather than on payoff signals from that person's farming practices. One's choice of an individual or group from which to learn might be affected by prestige bias, conformist bias or aversion bias, and Stone draws attention to the potential for social learning on its own to be maladaptive.

While there are potential limitations to both environmental and social learning, they are nevertheless ‘relatively free of external interests’ (Stone, 2016: 10). On the other hand there is didactic learning, which Stone uses to refer to change in farmer practices resulting from the intervention of ‘agricultural didacts’. These didacts are external parties such as public agricultural extension, commercial input suppliers, NGOs and the like who, in Stone's view, act ‘primarily out of their own interests while claiming to act in the farmer's interests’ (10). The tools of didactic learning include instruction, demonstration, exhortation, advertising, regulation, coercion, adulation, and shaming (10). Agricultural didacts have been active in many rural areas in the Global South for a very long time, starting with colonisers, missionaries and early traders in commodities wanted by the metropole.

One of Stone's key propositions is that in some situations environmental learning becomes increasingly costly or inaccurate. When this happens, as in the transgenic cotton example discussed above, farmers will increasingly rely on social and/or didactic learning, with potentially significant negative consequences. Autonomy and control on the part of farmers require learning modalities that develop both the skill required to make decisions (e.g. where when, and how to do something), and the skill required to implement them.