‘Ceci n'est pas une competence’, or, the view from the Shard: some early thoughts on Skills England’s Standard Skills Classification

(The views in this blog are my own, highly provisional, and do not reflect those of the Edge Foundation).

René Magritte (1898-1967)

At the end of April I was invited, in my capacity as the curator of Edge’s Skills Shortages Bulletins, to attend the launch of Skills England’s UK Standard Skills Classification (SSC)[1], hosted at the Shard, London. After the event, I wrote up some internal comments at some length for colleagues and considered drafting into something more presentable for publication somewhere, but I was busy with other priorities and the idea fell by the wayside. Now several months on and with some time over the summer I revisited those notes and thought perhaps now is the time, uninvited and underinformed, for me to wade back into the skills debate.

I was delighted when someone appeared to have read (and wanted indeed to respond!) to my last intervention on HEPI back in February. Never mind that the respondent, Professor Dame Athene Donald, elided the distinction I drew between embodied ‘skills’ and the political discourse of skills, which thankfully Ron Barnett swooped in to clarify in the comments. In my piece, I raised Skills England’s UK SSC as a premier example of the discourse of skills: ‘bureaucratic proxies for specific human capacities’. To emphasise, this policy discourse is what I am talking about when I am talking about skills. This bureaucratic, ontological categorising and ordering schema of ‘skills’ is related to but distinct from the productive, expandable, and socially determined personal qualities concept of a skill that prevails in the literature and work following Francis Green. Green’s widely deployed definition is agreeably embodied, contestable, and squishy, and it therefore has absolutely no capacity to sustain the weight of a centralised accountability bureaucracy of an economising democratic state.  

As an early Government Social Research paper by the SSC’s eventual architects put it in 2023, ‘the measurement and assessment of skills in the UK remains fragmented and deficient, and the language used to describe skills is inconsistent and unnecessarily complicated’. This paper was succeeding work begun under as early as 2021 and under the Unit for Future Skills. It proposed a standardised taxonomy: a structured system for describing, organising and classifying individual skills using a common language across the labour market. This would mean ‘a ‘skill shortage’ can be both defined and identified with greater precision, enabling qualification and training providers to plan for the provision of specific skills’. This is, in my reading, enables a way of governing where national resources can be more deliberately allocated to increase the supply or demand of any given ‘skill’ in pursuit of any given social or political end. This is part of a broad dissatisfaction with the capacity of market allocation systems to achieve these priorities. As Linkedin commenters and the Economist delight in pointing out, as if there hasn’t been over a century of debate on this point, from a market perspective skills shortages do not exist – simply that the unit price of the required labour exceeds that which an employer or the state is prepared to pay. This argument is unhelpful if you have a democratic mandate to achieve a given end which requires a given supply of skills but only limited resources to achieve that goal. You might be better off using your resources to stimulate the education and training ecosystem to provide the skills you need. As the paper argued, ‘the SSC can be part of the answer to facilitating a more efficient labour market, simplify reskilling, disseminate innovation and, ultimately, improve individual agency and workforce productivity’. Skills England’s Driving Growth and Widening Opportunities shared this diagnosis, and the University of Warwick's Institute for Employment Research, the University of Sheffield, and Omnifolio (a non-profit psychometric profiling ‘services setup’ established by consultant Neil Bachelor as far as I can see around 2010) were commissioned to develop the classification. My point is that this is explicitly about accountability in education and training a democratic society with strong systems of numerated accountability structures. It is a bureaucratic system for the definition and linkage of units of public and private expenditure with units of human capacities. As I will explore I think that these linkages are essentially arbitrary and that this elision between embodied skills and the SSC is both its main strength and weakness. The signifier is not the signified, image of fork found in kitchen is not fork.

Fast-forward to April 2026 and the publication of the classification at the Shard, on a sunny afternoon with stunning views from Warwick Business School’s conference and teaching rooms. I won’t run though the programme blow-by-blow, and to avoid editorialising my initial thoughts too much, I will proceed by themes. These thoughts are intended to be an articulation of my current, limited, thinking which I hope will be able to be developed and informed by further reading and colleagues’ thoughts, and not at all intended to be taken as a coherent thesis.

AI and judgment

I was struck by just how much the use of AI was embedded in the methodology and design of the classification: it was attributed to decreasing the time required to produce the classification from 4 years for the NOS to 2 years for the SSC, and from ‘10’ years for the American classification. AI’s primary use on this seemed to be synthesis of job description data (though presenters were candid regarding difficulties in wielding ChatGPT3.5 for this purpose and the guard-rails that were needed). This is arguably no worse than using manual researcher judgment which I presume is the original way this sort of work was done and is likely to be as variable and bias in its own way. However, since LLMs were likely trained on widely available data it is likely that some residual assumptions about what constitutes a valid summary and skill has been reapplied without reflection into the judgements and summaries, and therefore associations, that the project has made. The text embedded clusters methodology to demonstrate the relative proximity of different skills/tasks is pretty powerful, however, as well as the extent of the apparent interoperability with other skills taxonomies such as Standard Occupation Classification (SOC) by 2030.[2] Particularly too the capacity of the tool to determine the skills associated with qualifications was interesting – but again, qualifications are not the same as curriculums (and not even learning).

I thought it was interesting that the idea that the classification had indicated that, in one example, that the skills in a history degree were close to a data analyst and that this was important for diversifying recruitment pathways for businesses. This is of course not surprising for anyone (or any provider) that has looked at the history of higher education and the strength of generalist degrees/liberal education. It will be interesting to see whether this classification scheme validates that argument, and whether those areas of political discourse that have very narrow understandings of the relationship between qualifications and work-place capacities will be convinced.  

The definition of a skill provided (actually, an orthodox typology, of several thousand occupational skills and 12 core skills) I thought was further evidence that ‘skill’ has very little to do with the embodied capacities of worker-citizens. Indeed, workers and those particularly the unemployed were portrayed as needing ‘help’ to articulate their skills (which does not provide good evidence that their ‘skills’ are the same as their ‘embodied’ skills). Instead from this perspective skills resemble a lens of analysis, a way for government and employers to realise these capacities visible and manipulable in the domain of spreadsheets and bureaucracies for ‘accountability’. The use of AI in establishing the validity of different skills is actually I think further evidence that the relationship between ‘skills’ and embodied skills is fundamentally arbitrary for the purposes of the classification. Instead, what matters is the epistemic confidence by which the reality of the relationships is asserted.

Accountability and education

My concern is that this is a facade of accountability, a ‘good enough’ proxy for the real complexity of the labour market and individuals’ capacities within it that it will become itself the measure against which providers and employers are expected to adhere to, rather than complicated material reality. While labour market movement will appear frictionless in the self-referential and reduced internal classification, in reality misalignments are inherent and there will be continuous drift even with instrument continuously operating to catch and address this. Accepting that perfect is often the enemy of good (it’s not ‘beyond the wit of man’ to make some sense of the complexity, Frank Bowley, from the Department of Education, said) nonetheless I see this as a rhetorical excuse which will enable political justifications of the distribution of resources to certain areas without reference to the material consequences of that distribution. This is necessary in democratic societies, but these forms of seeing are never politically neutral. It blinds policymakers as much as it illuminates by narrowing what form of evidence is considered valid. While it was claimed there were no plans to use the classification as a ‘metric’ for higher education (‘We will not use this a metric’, Bowley said explicitly so we can hold them to that), nonetheless there were stated capacities for the classification to be able to be used to target investment to skills priority areas, and the providers/courses/qualifications that are deemed to provide such skills (and likely a cost-benefit analysis wouldn’t be too hard to consider based on the information the classification can provide; would any Treasury input bring in this sort of lens?). In particular and as usual, what is valued is the relationship of identified skills to workplace tasks to the exclusion of broader purposes of education. This will be a richer tool for a clearer articulation of democratically agreed national needs and their supply outside of market signals, but elements of fundamental and long-term importance are still excluded. A worker with the skill of empathy is not necessarily empowered to deploy that empathy if they work in an exploitative and abusive customer service environment. This needs thought. 

Adoption and success

I thought concrete details on the effort to champion the use of the classification was light on detail. This is primarily a supply-side classification? I got the strong sense that the primary lens of the project leads was that shortages were a consequence of supply-side failures and behaviours rather than reckoning with broader demand-side limitations. While the event did explore some ways that employers might use this tool and savings employers could make on maximising employer skills and training budgets, I didn’t get the sense this was going to be an effective instrument for stimulating greater employer investment, for example, or changing cultural meanings of training and social partnerships. There are, as far as I understand, a variety of international comparisons and indeed predecessors of the SSC; both America and Singapore in particular seemed to be a strong comparison point.[3] I don’t know as to whether these types of classifications have been used with success, what level of success, and how success is defined.

I’ve been quite critical so far but I do think it comes across as a powerful tool. The personability and precision in generating clear links between occupations, capacities, and the gesture to education feels far stronger, for example, than the useless careers advice and guidance tools I used at school in the 2000s. I am particularly interested in its capacity to evidence the contribution of generalist subjects. I suspect universities will be drawn on to use this for their own marketing purposes, and it will be easy to spin a story that obscures as much as it illuminates (I further suspect it will be very easy to make a clear case for most of the existing offer of providers already, and neuter some of the complaints from employers regarding the lack of applicability of skills to workplace applications – though this will do nothing to challenge the basis on which this complaint exists which is a matter of squeezed resources for training). I won’t pretend I fully understood this, but there were some interesting and possibly concerning possible applications of such taxonomies particularly regarding ideas of ‘skills-based pay and reward’ raised in questions, which while seen as not yet viable have ramifications for collective wage-bargaining and equal pay. The further atomisation and alienation of individuals from their capacities (0-hour contracts, etc) under the guise of greater freedom has historically not improved workers’ conditions.

My concern then, as I’ve repeated several times, is that what looks like accountability ends up continuing to obscure worthwhile educational and training goals and the nature of the relationship between central government decision-making and its outcomes. The grounds of greater efficiency has a temporal dimension and it is a political decision where that line is drawn. I don’t doubt that there is potential for some valuable and important developments and for informing the distribution of resources in the short-medium term. But this may be at the expense of other, local forms of accountability and professional judgment. I am not anticipating a huge audience for this piece but if you are reading this and you worked on the SSC or similar systems, I would be delighted to learn more about the research and classification and understand your perspective on some of the issues I’ve raised here – please drop me an email.

[1] I had been reading about SSCs (Sector Skills Councils) at the same time, and having the same acronym pop up again gave me a headache and I think indicative that initialisms are just out of control man.

[2] Existing sector classifications were expected to augment the SSC beneath providing granularity at the sector level. 

[3] Prior taxonomies include the Dictionary of Occupational Titles by the US Department of Labor in 1939, its predecessor in the 1990s O*NET, ESCO for the EU, and Singapore’s Skills Framework.