Erbar-Maas Singular Causal Interventions
An interactive laboratory demonstrating singular limits on Continuous-Time Markov Chains.
Causal Inference
From Boolean Verdicts to Quantitative Witnesses: Why DAG Topology Needs a Trail Semantics
Since Pearl (2009), causal inference on DAGs has crystallized around a powerful but austere toolkit: boolean d-separation and do-calculus. Does evidence flow? Full stop. Does it flow after an intervention? Full stop. This framework is sufficient for causal identification—determining whether an effect is estimable from observed data. But it is curiously silent on a question that seems equally natural: how much flows, through which channels, and with what residual structure? ...