Actually, I could see it both ways. Maybe PuterTutor is thinking that the server *response* time should be the same, but the latency between client and server should account for the lag in loading pages, thereby not affecting the server lag.
Quick analogy. It'll take me 30 seconds to warm up a pizza when I get home. Now, I could do the speed limit on the way home and it'll take 20 mins, or I could floor it and get home in 10. Either way it should still take me 30 seconds to warm up the pizza, no matter how fast/slow I get home.
I can't see how the generation times would increase based on connection because to me generation times mean
server processing time, not
packet travel time.
Correct me if I'm wrong...
Now if the calculator showing page generation times used packets
timestamped by the client, then it could depend on connection speeds. But is that
really measuring the server, and only the server? Not really because it makes the other variables play such a large role that the server-end page generation times would then be worthless.