So, are you agreeing or disagreeing with clock speed being a poor choice of gauging CPU performance without regard to differences in architecture? Hell, the original post that this is for was comparing the Eniac to a modern mobile phone, I'm pretty sure that is old Celeron vs new Xeon, times about a billion.
Given identical chips, one at 2.0 Ghz and another at 4.0 Ghz [lets assume the latter has enough cooling to handle the excess heat generated, sure CPU clock speed is a good measure], but the reality is that when the CPUs are even one generation apart, let alone several or dozens, clock speed is virtually meaningless.
I suppose if you test is following a link and timing how long until a plain text HTML page loads, the Xeon and Celeron CPUs, certainly won't be separated by a factor of 56... but I'm pretty sure the Xeon will still destroy the Celeron in that basic task.