Gert Jan talks about modelling socio-technical systems

Recently Gert Jan's group at Wageningen started a collaboration with North-West University (NWU) in Johannesburg, South Africa. He collaborates with them in creating agent-based models of socio-technical systems, for instance, about safety culture in industry. During a visit to NWU, Gert Jan was interviewed by Hermien Zaaiman

Perspectives on culture with Prof Gert Jan Hofstede

Comments to the interview

Culture's Causes

You can hear him talk about his overarching research interest, culture's causes. Culture makes certain events and processes more likely than others, and these in turn may perpetuate, or change, culture. In society, culture is slow to change; in companies, it can be set by pioneers but is hard to change after that.

Culture: never the only thing

Why do processes in companies, and in society, develop as they do, and not otherwise? Giving various examples, Gert Jan argues that culture is never the only cause for the unrolling of events, but it is also always one of the drivers. There always is a cultural backdrop. Other, more proximate drivers are for instance individuals, incidents, institutions of all kinds.

Modelling socio-technical systems

It follows that in modelling socio-technical systems in industry, it is important to capture both the technical and the social side. For the social side, one should capture both the proximate and the backdrop factors. This is still a big research challenge, and one that cannot be solved by Big Data alone. A comprehensive understanding of the mutual dependencies between the social and the technical is needed. So is a rich model of the social side.

Agent-based models

Gert Jan very briefly introduces the technique of agent-based modelling. In these models, people can walk around, for instance in a plant, and behave as real people in that situation might. So agent-based models can be intuitive, they can make life happen before your eyes, and they can be easy to grasp for non-specialists.

Verification and validation

Models need to be verified, though: does the model do what was intended?  They also need to be validated: Does the model represent the real-world system it was meant to represent? Models can be used for many purposes. Depending on purpose, they need to be validated more or less thoroughly. Agent-based models are not usually good enough for prediction of what will happen, but they tend to be very useful for showing which developments are likely if people operate in a certain environment under certain incentives.