The harm state vector

For each person or cohort $i$ at time $t$, the model tracks an 8-dimensional harm state:

where $L$ is life-years lost, $Q$ is morbidity burden, $M$ is material extraction, $A$ is agency loss, $P$ is political voice dilution, $E$ is ecological damage, $I$ is epistemic degradation, and $T$ is tail risk. A higher-level capability map $c_i(t) = \Phi(x_i(t))$ transforms this into what the person can realistically do, become, and avoid.

The damage functional

The full functional integrates harm over time, across populations, with explicit treatment of catastrophic risk and institutional fragility:

Here $\beta$ is the temporal discount factor, $\omega_i$ are moral weights, $\ell$ compares actual to counterfactual capability, $\mathcal{R}_a$ is catastrophic risk, and $\mathcal{F}_a$ is institutional fragility. A scalar index is derived from this functional only after vector aggregation, repair adjustment, and obstruction correction.

Moral weights and incomparability

There is no morally neutral weighting. The framework makes this explicit: different ethical commitments yield different weights. Rather than forcing all harms into $\mathbb{R}$, the full theory uses a partially ordered harm space $Q$ where some harms dominate others, some are incomparable, and the residual $x \Rightarrow y$ represents the minimum repair needed.

Open questions

The hardest problems are not technical. What is the baseline — no actor, a regulated version, a cooperative alternative? Should future generations get equal weight? Is consent meaningful under structural dependency? How do you model epistemic harm mathematically? These are the true foundations, and the playground makes them adjustable rather than hiding them.