Calculate a Celestia sampling miss probability under explicit assumptions
A simple probability calculation helps interpret successful data availability samples, provided the sampling model is stated alongside the result.
Fix the withheld set
Assume that exactly one quarter of positions in an extended square are inaccessible before sampling starts. Draw twelve independent uniformly distributed coordinates with replacement. Each draw misses the withheld set with probability three quarters. Multiplying those probabilities gives about 3.17 percent for missing every unavailable position.
withheld fraction = 1/4
independent samples = 12
probability all miss = (3/4)^12 ≈ 3.17%
Interpret the model before alerting
Repeatedly fetching the same coordinate cannot be counted as new independent evidence. Sampling without replacement uses a different calculation, and a network timeout also has causes unrelated to withholding. Record unique coordinates, header identity and peer failures when investigating actual node results.