Datatomic Labs
All research directions

05 / Emergence & prediction

Research agenda

Emergence and the limits of prediction

When do local interactions produce a pattern we can reliably anticipate?

01 / MOTIVATING OBSERVATION

A system can be governed by simple local rules and still produce behavior that is difficult to anticipate. A convincing explanation of a past pattern may offer little ability to predict the next one.

02 / STARTING CONTEXT

Nonlinear dynamics and agent-based modeling provide starting tools for examining this gap. The lab's question is which patterns remain useful when observations are incomplete, initial conditions change, or a system adapts.

03 / WORKING HYPOTHESIS

Some aggregate features may be more stable and predictable than individual trajectories. A model should identify the scale and time horizon at which its forecasts remain informative.

04 / PROPOSED INVESTIGATION

  1. Build a small model with explicit local rules and observable aggregate behavior.
  2. Vary initial conditions, measurement noise, and interaction strength.
  3. Compare forecasts across individual and collective scales.
  4. Measure uncertainty and performance on cases held out from model development.

05 / WHAT WOULD CHALLENGE THE IDEA?

If a pattern disappears under small plausible changes, or offers no improvement over a simple baseline, the claimed predictive structure needs to be reconsidered.

STARTING LITERATURE

External work by other researchers provides context for this direction.

DOCUMENTING THE WORK

Papers, simulations, and video explanations will be linked as this investigation develops.

Visit the notebook

Connected questions.

06Emergence & prediction

Seismic models of financial stress

Can the mathematics of stress and failure help us understand financial instability?

Translate a physical analogy into a testable model of market stress, then assess its usefulness against established risk measures and unseen periods.

Research agendaRead the question