The 400 ms threshold cited for the Doherty threshold is from a 1982 research paper. I imagine that in 1982, 400 ms response times felt pretty darn good.

Also, the nature of the tasks that users performed in 1982 may have meant that there was no benefit to <400ms response times. (If it takes twenty seconds to read and understand the output, and another ten to key in your next command, then a 400 ms delay is not a meaningful productivity drag.)

Today, I would guess that the threshold is closer to 50 ms.

(Also, "time spent by the computer waiting for the user" makes sense in the context of a 1982 research paper. That has changed in the last 28 years, to say the least.)

All of this is a long way off saying that if you are making UX decisions based on 28 year old research, you are on shaky ground!