Method, since the number deserves scrutiny: the dataset is every bird on my site where both dates are documented, the last confirmed record (specimen, photo, recording, accepted sighting, or death of the last captive) and the first formal extinction declaration by a recognised authority. That gives 13 species and subspecies. Median gap 36 years, mean 47.5, range 3 to 124.
Two rules matter for reading it. Where a bird was declared twice (IUCN and later the US delisting in October 2023), the earlier declaration counts; using the US dates would inflate most figures. And the sample is small and weighted toward Hawai'i, so the median describes these birds, not a published constant.
The pattern that seemed most robust to me: the three short gaps are all birds whose last individual was watched dying, e.g. the dusky seaside sparrow in captivity in 1987, declared in 1990. Where the record just stops, nobody can date the stopping, and the declaration takes decades.
Happy to go into the sources or any individual case.