I'd say the (random) seek time estimate of 1.2x speedup is overestimating. During the last ca. 15 years, I'd say the whole increase has been not much above 0% in the normal case (7200 rpm hard drives), with the possible exception of later command queueing techniques speeding up even random seeks a bit. I remember 850MB Quantum Fireballs having pretty speedy seek times.
So one way of approximating is that during 1994-2009:
20x faster transfer rate
1.1x faster seek times
And I think application developers still don't generally understand that anything seeking the hard drive is killing performance. Just look at the login times of popular desktop environments.