Metric Glossary

A reference table for analysis fields and endpoint names. Use it together with:

Metric ID	Definition	Source	Unit	Direction
`turnout_pct_t`	global participation rate at step `t`	`steps.turnout`	`0..100`	higher = more participation
`gini_assets_t`	inequality over agent assets at step `t`	`steps.gini_index`	`0..100`	higher = more inequality
`mean_dissatisfaction_t`	mean `agents.dissatisfaction_value` at step `t`	derived from `agents.parquet`	`0..1`	higher = worse
`gini_dissatisfaction_t`	Gini over `agents.dissatisfaction_value` at step `t`	derived from `agents.parquet`	`0..100`	higher = more inequality
`quality_distance_t`	eligible-weighted quality-gate distance at step `t` (mode-aware source)	derived from `area_steps.parquet`	`0..1`	lower = better

Aggregation semantics:

turnout_pct_t = 100 * sum_a participants(a,t) / sum_a area_num_agents(a); if denominator is zero, value is 0.
quality_distance_t = sum_a quality_distance(a,t) * eligible_voters(a,t) / sum_a eligible_voters(a,t); if denominator is zero, value is NaN.
quality_distance(a,t) source:
puzzle mode: area_steps.puzzle_distance
reality mode: area_steps.dist_to_reality
dist_to_reality_t remains a secondary grid-alignment diagnostic (not the primary gate metric in puzzle mode).

Metric ID	Definition	Source	Unit	Direction
`mean_altruism_t`	mean altruistic-vote propensity over agents at step `t`	`steps.mean_altruism`	`0..1`	descriptive only
`diversity_first_choice_entropy_t`	normalized entropy of first-choice ballot IDs	`votes.parquet`	`0..1`	higher = more diverse
`dist_to_ref_utilitarian`	distance to utilitarian benchmark trajectory	analysis output	`0..1`	lower = closer
`dist_to_ref_nash`	distance to nash benchmark trajectory	analysis output	`0..1`	lower = closer
`dist_to_ref_egalitarian`	distance to egalitarian benchmark (`lambda=1`)	analysis output	`0..1`	lower = closer
`dist_to_ref_rawlsian`	distance to rawlsian benchmark trajectory	analysis output	`0..1`	lower = closer
`dist_to_ref_egalitarian_lam025`	egalitarian sensitivity (`lambda=0.25`)	analysis output	`0..1`	lower = closer
`dist_to_ref_egalitarian_lam400`	egalitarian sensitivity (`lambda=4.0`)	analysis output	`0..1`	lower = closer

Metric ID	Definition
`turnout_mean`	mean of `turnout_pct_t` over all recorded steps
`turnout_final`	final observed `turnout_pct_t` value
`turnout_volatility`	adjacent-step L1 volatility endpoint
`gini_assets_mean`	mean of `gini_assets_t`
`gini_assets_final`	final observed `gini_assets_t` value
`gini_assets_volatility`	adjacent-step L1 volatility endpoint
`gini_dissatisfaction_mean`	mean of `gini_dissatisfaction_t`
`gini_dissatisfaction_final`	final observed `gini_dissatisfaction_t` value
`gini_dissatisfaction_volatility`	adjacent-step L1 volatility endpoint
`mean_dissatisfaction_mean`	mean of `mean_dissatisfaction_t`
`mean_dissatisfaction_final`	final observed `mean_dissatisfaction_t` value
`quality_distance_mean`	mean of `quality_distance_t`
`quality_distance_final`	final observed `quality_distance_t` value
`quality_distance_volatility`	adjacent-step L1 volatility endpoint
`diversity_entropy_mean`	mean of diversity entropy over time
`diversity_entropy_final`	final observed diversity entropy