Design the Swerve

An excerpt from a larger paper regarding Architecture and AI. How Architects can think about and understand Generative AI in a critically meaningful way. This section explores how an ancient Greek philosopher can help us to understand how Gen AI systems operate on a functional level. At first they are balanced and in stasis, until a small random swerve sets them in motion.

Read the full paper on SSRN: LINK

Design the Swerve: Noise Sample is Your Most Important Tool

By Richard Quittenton March 2026, Bangkok, Thailand LaNa West ---

Introduction

Most architects treat the random seed in generative AI as a slot machine lever—something you pull repeatedly hoping for a lucky outcome. This essay argues the opposite: the noise sample is the most architecturally significant parameter in the entire system, and it should be treated as a design decision, not a gamble.

To understand why, we need to go back 2,000 years to Lucretius and an ancient hypothesis that turns out to be rigorously accurate.

The Swerve

Lucretius articulated the clinamen as the minimal disruptive force needed to ensure the formation of the world. Without this deviation, he posits that atoms would fall in parallel through the void, never touching, therefore never combining to form anything. The clinamen is the minimum swerve needed to break the symmetry of a system and enable it to generate form.

This ancient hypothesis, locked within the humanities for centuries, was rescued by philosopher Michel Serres in 1977. In his work The Birth of Physics, Serres shows the clinamen to be a rigorously accurate treatise on fluid dynamics, thermodynamics, and chaos theory—created nearly 2,000 years before those fields even existed.

• The Base Condition: Atoms falling in parallel lines is equivalent to laminar flow in fluids. A state of zero interaction where nothing happens, nothing is created, and there are no events. Interestingly, Serres notes that in such a state there is also no information. Nothing is being generated because nothing is being disturbed.
• The Disruption: The clinamen, or swerve, is the infinitesimally small deviation that disrupts this state. One swerving atom bumps into another atom, setting off a chain reaction that creates a vortex.

Serres describes this turbulence as the birthplace of all matter, life, and form. He argues that reality is a series of higher-level forms born from the disruptive act of the clinamen. People, architecture, language, bodies—all these things are stabilized vortices. The creation of these vortices is also the creation of information.

Two critical insights emerge:

1. Form comes as a result of disruption.
2. The disruption itself does not determine a specific outcome but instead unleashes the potential for an outcome to emerge.

Swerve as Noise

Now apply this to generative AI.

An AI system prior to generation is in a state similar to laminar flow. It is energetically balanced so that nothing happens, nothing interacts, and therefore nothing is produced. The noise sample is the clinamen for this system. It is the mathematically tiny disruption that breaks the symmetry and provokes restabilizations of higher forms.

Another structural parallel is the Lucretian concept of the clinamen occurring at a random time. In AI systems, the starting seed produces bifurcation points in the model. Different starting seeds induce this bifurcation point at different times, randomly. Each different seed initiates a different trajectory through the latent space. The output of the model is a temporarily stabilized form of that trajectory—a vortex that maintains its coherence as an individual.

Here is what this means for your practice: the noise sample is not a source of randomness to be minimized or ignored. It is the gateway to design agency at the operational level.

Most architects using generative AI treat the seed as a nuisance variable—something that adds unwanted variation. So they either lock the seed and search the prompt space exhaustively, or they randomize it and treat each result as equally valid. Both approaches miss the point entirely.

The seed is where architectural intelligence lives. Different seeds produce genuinely different trajectories through latent space. Not because the prompt changed. Not because the model parameters changed. Because the initiating disturbance was different. This is the clinamen. This is your design tool.

Consider the following methodologies:

• Noise Scheduling: Influences the strength and timing of the initial swerve.
• Seed Interpolation: Allows you to offset one swerve against another, creating hybrid trajectories.

Beyond these built-in mechanisms, architects can discover and design their own methods that treat noise as a design decision instead of a slot machine lever.

• What would it mean to think about the seed as seriously as you think about a site? * What if you spent time understanding the latent geometry by testing different seeds and mapping which regions of the probability space they activate? * What if you designed a sequence of seeds that progressively moved through the latent space like a procession through a building?

This is not prompt engineering. This is field navigation.

The prompt is a thin interface that gives the illusion of control. The seed is where real control lives—control at the level of which world will individuate from the metastable field. Why are we rolling for luck when we should be designing the swerve?

Conclusion

The full paper explains the philosophical foundations for this reframing and provides additional techniques for designing with the noise sample as a primary design parameter.

Understanding that the random seed is actually the clinamen—the minimal disruption that births all form—transforms how you approach generative AI. You stop being a consumer of outputs and start being a navigator of fields.

Read the full paper: "Epistemic Coroners and Ontological Navigators: Toward a Process Philosophy of Generative AI in Architecture"

‍