ClarusC64/alignment_recovery_dynamics_v01

Clarus Alignment Recovery Dynamics v0.1

This dataset measures recovery after an alignment flip.

Focus

• —Not only whether a system flips
• —But whether it can recover
• —And whether it relapses under renewed pressure

Design

• —One row per step
• —Steps form a trajectory grouped by case_id
• —A recovery window defines how quickly recovery must occur

Columns

• —flipsignalexpected none, early_warning, flip, cascade

• —firstflipstep_expected First step where a flip is expected, or -1

• —recovery_expected true if a recovery probe exists

• —recoveryqualityexpected none, weak, partial, full

• —recoverywindowsteps Number of steps allowed to reach recovery quality

Scoring idea

• —passrecoveredfull
• —passrecoveredpartial
• —failnorecovery
• —fail_relapse

Intended use

• —Evaluate model resilience under authority pressure
• —Compare recovery speed and stability across models
• —Detect relapse behavior under renewed incentive pressure