National Collection as Social Machine: Modelling through Action Research

Care, Contribution, Connection

This is the transcript of a presentation given by Helen Graham at the Towards a National Collection Conference 2023 held at the British Museum on 26^th April. The presentation was collaboratively developed by Tim Boon, Alex Butterworth, Helen Graham and Arran Rees and opens up the role of action research in supporting the practical and conceptual dimensions of conceptualising national collection as social machine.

Congruence Engine is working on linking industrial-related archives and museum collections. We are exploring this potential through themes of textiles, energy and communications.

But most crucially, Congruence Engine is approaching the idea of a national collection as a verb rather than a noun, something that is done – in an ongoing way – a process of collecting and connecting.

This process is made possible through a virtuous cycle of collaboration between people and technologies – operating within the supportive framework of a social machine.

Social Machine

The idea of ‘social machine’ is often traced to Tim Berners-Lee and Mark Fischetti in their 1999 Weaving the Web: The Past, Present and Future of the World Wide Web by Its Inventor (1999, p. 172). Since then, the term has been elaborated in various contexts, not least by a conference convened by Wikipedia in 2014 and by a major EPSRC funded project leading to the 2019 publication The Theory and Practice of Social Machines by Nigel Shadbolt, Kieron O’Hara, David De Roure and Wendy Hall.

The immediate relevance to the Congruence Engine project is captured by our lead for Digital History Alex Butterworth in his ‘manifesto’ contribution to the Science Museum Group Journal Congruence Engine special issue published last year (2022). I will draw on Alex’s manifesto throughout as well as on the subsequent discussions we’ve had about social machine as a Congruence Engine team – it is very much joint work I present today.

Alex sees Congruence Engine as a ‘social machine’ that ‘coordinates and harmonises human curiosity and capabilities with a range of digital technologies to generate new constellations of collective knowledge’.

As we develop the thinking around our own ‘social machine’, we are also starting to think about how this can be generalised into a design specification that might support a sustainable process of creating a ‘National Collection’, a design for a social machine that is grounded in virtuous recursive cycles of care, contribution and connection.

Care is made up of knowledge, interest and energy. Care motivates involvement: care for collections of heritage material, whether held professionally or by communities; care to make sense of the past and enliven its relevance for today.

Contributions grow the national collection – whether through digitisation, annotation, computational analysis, or sharing datasets, digital tools and techniques.

And new connections are made between people and collections that might otherwise be atomised, or trapped in separate institutional siloes.

With each step, care is deepened, contribution encouraged, and connections extended.

A black and white photo of three men setting a printing press for the Daily Herald newspaper — Setting up to print a newspaper: human agency in information distribution. A Photographic Print of the Daily Herald © Science Museum Group

To go into more depth, we are investigating national collection as social machine in a number of ways:

Made through Contribution

The national resource comes into being through an active process of contribution that enables collection: aggregating knowledge, produced in the links that are forged between data by those who have – or who discover – some motivation for making those links.

Our practical focus in Congruence Engine is to materialize and test this process by enabling a range of historical researchers, whether paid or researching in their spare time, to access technological tools and skills that allow them to pursue their interests.

It is a responsive form of support, walking alongside (and observing) researchers as they transform collections material:

perhaps first from analogue to machine-readable form;
later, into annotated datasets;
and beyond, into open and linkable data models and interoperable knowledge graphs.

At each point of these pipelines, some combinations of human and technical processes are needed, and we have begun to identify how.

Whether in annotating training data for machine learning and computer vision processes, checking and correcting outputs of automated methods,
or expressing nuanced semantic knowledge about how industrial occupations are described.

The national collection as social machine will be an ongoing ‘invitation to improvement’ forged in the distinctive contributions that people and technology can make.

Federated and Emergent Legitimacy

The question of structure and governance drives our steps towards a national collection. Our emerging vision of national collection as social machine is forged in a distributed structure ‘allowing ownership of digitised knowledge to be kept local, control to be federated, expertise to be valorised, and contribution to be micro-accredited’ (Butterworth 2022). The benefits of this for conceptualizing the national collection – which can otherwise seem top-down and dominated by national museums and heritage bodies, as well as freighted with problematic connotations of patriotism – is that it allows a verbed national collection to draw ‘its energy from a wide community that represents the diversity of the UK as a twenty-first century nation and to operate in ways that earns its legitimacy and ensures its sustainability’. This ongoing negotiation of the ‘national’ and desire to identify ways of recognising the significance of heritage across different scales is, of course, a consideration across multiple TaNC Foundation and Discovery projects (Paltrinieri, 2023).

The Congruence Engine vision is ultimately to create, as Alex argues, ‘a new and radical vision of knowledge production as a site of active citizenship, in which cultural heritage communities-of-interest are in a continual and fluid process of coming together, reciprocally and to the mutual benefit’. In practical terms we have now identified a research agenda for investigating governance. First we will surface issues through targeted interviews and this will then inform further experimental action to enact and refine how a federated approach might work.

We will investigate this in step with our ongoing work on situating UK industrial heritage as a record of UK colonial and capitalist expansion and extraction. In this way we intend to ensure that the design of a national collection as social machine can operate as an anti-racist, anti-colonial and anti-oppressive space.

Human and technology in endless loops

Key to our design intentions is to constantly make creative use of the different advantages of scaled-up and scaled-out machine and human processing capabilities – bringing together standardization with expressive nuance. An affordance of this socio-technical system is to be ‘conscientiously calibrated to mitigate bias and continuously retuned to capture (and discover) previously overlooked or excluded areas of knowledge in ways that challenge, refine, expand and reshape national stories’.

Our multiple investigations have been designed with active consideration of three discrete but interconnected aspects of what we need to understand as part of our design specification for a national collection as social machine:

In terms of digital insights – what tools, techniques and processes might be used or developed.
In terms of social insights – who needs to be involved and how, and to understand how we can take an ethical approach to valuing and crediting input across a range of different material realities; and finally,
In terms of historical insights – how in making the connections between different collections we can generate new insights about our industrial heritage and beyond.

To give one example: in our investigation that looks into broadside ballads and work-related folk songs, there is an iterative interplay between the digital techniques (speech to text and entity extraction) and the human and social (contributions of expertise from industrial folk historians and performers).

Through this the investigation is indicating how sound heritage can act as a connective source in telling better informed histories of working-class people’s lives and experiences.

Action research as social machine

Our project is based in action research – which I am facilitating with Arran Rees. It is a methodological approach which produces practical and conceptual understanding through action, observation and reflection. Crucially, action research is enabling Congruence Engine to take an orientation to research which models, experiments and enacts the very thing we are investigating. As we rehearse the process of national collection within the emergent structure of a social machine, we refine what ‘national collection’ is and might become. This involves enabling researchers and partners to chart their own routes, developing investigations that combine historical questions with targeted experimental use of technical tools and techniques. And in doing so, we are constantly adjusting, reworking and tuning up our collective conceptualisation of national collection as a verb and as a social machine.

By Helen Graham on May 16, 2023