With less matter for the photons to interact with, I assume the WiFi’s SNR would be improved. If fewer data frames need to be retransmitted at the link layer (WiFi), I figure the apparent bitrate at the IP level might actually be improved!

Actually, I would not be shocked if WiFi itself adapts to conditions, e.g. by sending less data per frame with more error correction bits when SNR is already low.

(Not a networking expert, I am just bullshitting.)