Q/A AILiner panel discussion

Q1. If we look 50 years ahead, how might sewer networks change? Will we still need the same networks we have today?

My discipline teaches a simple lesson: infrastructure almost never gets replaced, it gets either actively maintained or it is ruined by the lack of maintenance. Infrastructures are built on an installed base (Star and Ruhleder 1996). The question “will we still need the same networks?” is usually asked about the visible layer, such as treatment plants, sensors, digital twins. But the buried network, the pipes in the ground, is the part that outlives every innovative vision. Thus, if decentralized treatment and blue-green infrastructure is implemented then it will be added, not substituted. This will result in hybrid systems, with the conventional network as the backbone that everything else is built on. What will change most is not the pipes but the regime around them: climate stress, shrinking and ageing cities that have to decide which districts to keep served and which to retire. This retirement decision is then where it gets extremely political. But the trunk lines remain essential.

Q2. How can we make sewer networks more resilient and adaptable to uncertain future conditions?

In my discipline we resist firmly the idea that adaptability is a design property you build in once. It is a special case of what we call the design fallacy (Stewart and Williams 2005), the idea that emergent properties can be designed in from the beginning. Adaptability is such an emergent property. This is because it is a practice, not a feature. This practice keeps the network legible, accessible, and repairable. Practice theory teaches us that stable practices depend on skills, norms, and material tools supporting the practice (Shove, Pantzar, and Watson 2012). Thus, there has to be systems of skill transfer and preservation, a political will (which allocates resources), and there are also physical properties that support flexibility and adaptability. A network designed for maintenance - with access, redundancy where it’s cheap, and documentation that survives staff turnover - adapts better than a network marketed as “smart.” Important in this context is that the material network, the competences, and the normative commitments have to change together. Change only one - e.g., modernise the material base without keeping competences or resource allocation in sync - and the network loses the property of resilience and adaptability.

Q3. Should utilities continue prioritizing the rehabilitation of existing assets, or increasingly invest in adapting and transforming their networks?

This question contains a false binary as long as it insists on transformation being the opposite of rehabilitation. In reality both the rehabilitation without transformation and “greenfield” transformation are the extreme points of a continuum. Similarly, preservation of infrastructure value and preparation for uncertain futures do not have to be polar opposites. Not investing in existing infrastructure comes with cost beyond the material value. Writing off infrastructure is cheap for the people who decide it - because it has been paid for before - and expensive for those who have to learn to live without this infrastructure. Obsolescence claims should therefore always be treated as unproven claims with costs (for someone) attached, not as predictions of the future.

Q4. What are the possible futures for urban infrastructure management, and which one is most desirable?

Of the three scenarios, in line with what was said above, I am critical towards claims of radical innovation. “Self-healing networks” is a metaphor that hides labour because nothing heals magically by itself, in one way or another you have to add work and energy to it to heal, entropy makes sure of this. Digital twins and AI optimization have a real potential to be useful, but our research on how digitalisation of infrastructures is performed in everyday life (in our case remote operation and maintenance of a building portfolio), makes me warn about what could be called its “counterfeit” version: a dashboard or app that performs what Haraway (1988) has called the “God’s eye trick”, by promising a total view of all that happens, which gives god-like control. In reality, the maintenance of these tools sucks up resources that would be needed to care for the maintainer and to strengthen the relation between the maintainer and the maintained assets. The desirable future would instead be a future where maintenance is visible, funded, and skilled, with digital tools in service of that relation rather than as its replacement.

Q5. How much data do we really need to improve asset management decisions?

I will leave this for the experts.

Q6. Beside technology, what will drive the transformation of urban infrastructure management and what are the main barriers preventing it?

Here I would like to focus on the human dimension. If the current demographic development continues we will see rising retirement rates in an ageing society. And with the workforce, tacit knowledge about the network retires which is an erosion of maintenance competence one simply cannot magically reverse through digitisation. This is at least true as long as budgets prioritise novelty and neglect upkeep. Maintenance is invisible until something breaks, which is why it loses budget fights against something politicians can claim as their doing. Here we need policy, indeed. We need maintenance accounting that makes the real cost of backlogs visible on the balance sheet.

Q7. Let’s dream for a minute. You are the mayor of your city and the head of the water utility and you have the power to change things.

Three moves: First, structurally protect maintenance funding so it cannot be traded against glamour projects. Second, implement additional support for the competence pipeline: apprenticeships, documentation practices, and time for maintainers to hand over what they know. This would mean to treat knowledge maintenance as seriously as pipe maintenance. Third, for every new digital system we buy, account at purchase for the true costs including expertise needed to run and maintain it, and the expertise that is destroyed by it.

Q8. What are the concrete takeaway messages for AILINER and the consortium; what do you think is key to consider for the project? Any last message before we go home?

Three recommendations: First, the project’s outcomes will and should be judged not by their demo performance but by what happens to them after the funding stops. Who will use, maintain, improve them? Think about this now, carefully! Secondly, be aware of the knowledge dimension: the AI must keep the humans’ judgment in the loop and augment it, not replace it. Third, a slogan: an asset-management system is not an asset but a relation between things, between people, and between things and people that has to be maintained. If these relations are not maintained, the assets turn more or less slowly into ruins.

References