Not sure if I understand this. I agree they haven't changed much between, say 2013 and 2014 or 2014 vs 2015. But the jump from 2010 core2duo to 2014 core-i is substantial.
The benchmarks that you and I posted show the 2.6ghz Mini is almost three times faster than the OP's 2010 mini. I get your point about most people not really needing additional speed, but it really depends on what your needs are. For example, ripping a dvd or encoding video that you shot will make good use of the additional geekbench score. These tasks are pretty common for many users. My own tests between a 2012 quad and 2012 base mini showed render times to be proportional to geekbench ratings.
And as I posted earlier, handbrake render times were dramatically improved between a core 2 duo machines and core i machines. I believe this is due to the the turbo mode that was introduced in core i. I saw over 4x render speed improvement between a core 2 duo machine and core i7 machine even though the geekbench score was only 2x faster.
I think you're missing the difference between clock speed and performance. GeekBench is a measure of performance - how much
work can be done, in how much time. Clock speed is only one component of performance. If no change is made to architecture, you would have to increase nominal clock speed by a factor of three in order to triple the speed of execution of the GeekBench tests (which a chip engineer would tell you is impractical). If the clock speed remains constant but the performance triples anyway, changes to other aspects of CPU architecture are responsible for the increase - the CPU accomplishes more work in each tick of the clock. Classic example (though not pertinent here) is a move from 32-bit to 64-bit architecture. Turbo mode is also a change to architecture, as is speed of the RAM bus (and the RAM it addresses). Size and number of on-CPU registers and buffers, dimensions of the die (speed of light)... all factors.
Further, GeekBench is a test of total system performance, not just raw CPU performance. So, matters like speed of mass storage and RAM also factor in.
In your example turbo mode may be responsible for a fair part of the improved speed of execution. Your experience disagrees with GeekBench (4x vs 2x), so one has to examine whether turbo mode is invoked by the test, and if it is invoked, what weight do turbo mode tasks have in the test results.
My guess is that turbo mode is not responsible for much of the Geekbench improvement. Turbo mode is intended for relatively short bursts of compute time - it generates far more heat, which must be dissipated, it draws far more power, which in laptops hurts battery life and in desktops may require larger power supplies. If turbo mode could be sustained at all times, then one could indeed increase the advertised clock speed for the CPU accordingly.
Clock speed is similar to engine RPM, in that it only tells a small part of the picture. Engines of all sizes may run at similar RPM, yet the results in power output can vary dramatically. Number of cylinders, displacement of those cylinders, efficiency of combustion, transmission and drive train design....
