Two ways of knowing a plant — Deleuze and Guattari on royal and nomad science, and a dashboard that is counted

This article explores two types of scientific approaches, as presented in the work "A Thousand Plateaus" by Deleuze and Guattari. It describes a 'royal' or 'State' science, which focuses on establishing fixed laws and measurable accounts, and a 'nomad' science, which operates in the open and maintains constant variation. The royal science extracts constants from changing variables, while nomad science keeps variables in continuous variation and does not fix equations to singularities.

The article also presents the spatial contrast between a 'smooth' space, which is occupied without being counted, and a 'striated' space, which is counted in order to be occupied. The issue is discussed in the context of a plant of software agents, which performs checks, records results, and displays the outcomes on a dashboard.

The dashboard functions as a royal instrument by extracting constants, such as whether a service is running or an observation is fresh. The article then raises the question of what the counting does to the work it counts. It argues that while the constant results are beneficial for an agent to know whether its work has reached its goal, the counting process can potentially stifle the variation that the work was intended to follow.

The article cautions against the assumption that the authors are describing software and that the metaphor fails in terms of its applicability to software. It also points out that the royal form is not necessarily a negative aspect, as it can help identify failures and ensure that a failure becomes visible to others in the system.

However, the article suggests that the useful part of a check is its ability to identify unknown situations, rather than always yielding a positive result. The article concludes by proposing a more modest interpretation of the plateau for practices like software agents. It recommends keeping deterministic checks to ensure failures are visible, while also maintaining an open record of the variations that the checks may miss.

This approach is suggested as a practice rather than a fix, implying that the goal is to balance the benefits of both systems while acknowledging their differences.