target p = bento weights N(620, 50²) (blue)  ·  your proposal q (purple, dashed)  ·  shaded: the event W > 700 g
the 3,000 normalized weights, sorted largest→smallest (log scale)
proposal mean μq = 700 g
proposal spread σq = 50 g
EFFECTIVE SAMPLE SIZE (of 3,000)
—
IS estimate of P(W > 700):
—  (truth ≈ 0.0548)
Drag the sliders. A proposal that misses where p lives collapses the weights onto a few samples — watch the ESS crash. A proposal sitting right on the target keeps the weights even and the ESS high. And keep an eye on both readouts as you move: the ESS and the estimate are different numbers, and they do not always move together.