Those are the values for insertion between a zero
impedance source and an infinite impedance load. If your
impedances are different, won't the attenuation also
differ?

Of course, but we're not engineering the input attenuator for a Fluke 4 or
5 digit meter.

This isn't for measurements, just a useful stepped
attenuator between a notionally low and high impedance.
The values won't be dead right, but near enough for
government work. The choice of 10k as a working impedance
is a reasonable loading compromise between typical input
and output impedances.

Agreed, and consistent with accepted professional practice for recording.


If you use three resistor in a Pi Attenuator arrangement you can set the
input and output impedances as well as attenuation (though of course the
laws of physics apply so you can't whatever you want).

http://chemandy.com/calculators/matc...calculator.htm

Is a handy online calc to determine values.