OISST followed 8-32-day buoy SST changes 0.45 day later in 2022-2025 than in 2016-2019 at 51 moorings

We tested whether the recent loss of OISST response to buoy changes includes a time delay. We used a signed daily archive and complete 128-day windows. The main comparison has 51 moorings in 2016-2019 and 2022-2025. We fitted the cross-spectral phase at periods of 8-32 days. The median added delay was 0.45 day. The spatial-group 95 percent interval was 0.28-0.65 day. A joint year and group test gave 0.22-0.69 day. OISST followed changes later at 46 of 51 stations. A separate seven-day increment test gave an added lag of 0.44 day. The phase-free response at 3-8 days fell to 0.65 of its earlier value. The median response ceiling under any phase correction was 0.79 of the earlier simultaneous response. This limit was stronger at US NDBC stations than at tropical moorings. The phase result held with different windows, filters and season samples. Standard spectral identities show a correction limit. Nonnegative amplitude filters and independent added noise cannot change cross phase at a fixed frequency. A power match therefore cannot by itself restore the timing of buoy-related changes. Buoy assimilation, point-to-grid differences, time-of-day sampling and short eras prevent a true ocean filter or one cause from being identified. This study gives no global delay estimate. Code, annual spectra, aggregate results and exact source hashes are included.