Our own numbers

Our own outreach, real and simulated kept apart.

Reading the system

Every test we are running

Two versions, one measure named before the first send. No verdict before the minimum sample.

Reading the experiments