- let me give some wider bandwidth answer without all the math.
Imagine, if you will, that you can see "any" mismatch with a Time Domain Reflectometer (TDR) within a defined SNR range. So even short reflections can be seen over a long matched cable. You understand these reflections add or subtract noise on the signal by under or over the nominal impedance and long delays with possible small mismatches at either end repeating themselves. This can be measured by SNR or distortion levels or even phase noise in dB down from the carrier.
The 10% wavelength rule of thumb is based on a breakpoint in percentage of the worst attenuation of carrier loss due to mismatched short feeders. 5% is a better quality standard for narrow band RF.
But for baseband or extreme wideband Network Analyzers that Rule of Thumb gets reduced significantly both on length and tolerance of mismatch for SNR reasons. A short mismatched feed on signal or ground will have significant quality effects on baseband signals for measurement errors or or if you expect high SNR quality such as baseband broadcast video with ghosting effects.
So the BW expectations of the channel must be much higher than the signal for wideband high SNR signals. This is not necessarily true for Logic Signals except near the transition with high gain. This is when signal integrity matters with digital crosstalk on analog signals, it is best to use twisted pair magnet wire if coax is not possible in order to prevent loose jumpers on breadboards wreaking havoc to lower the impedance and possible add a source R to match the link to at least prevent reflections at the source even if the load is mismatched open circuit in logic.
For tolerable losses the breakpoint in narrow band carrier signals is 10% wavelength.
Time domain & FFT Simulation
A 10 ps sample time was used to plot the red log. spectrum of from 10 MHz to 1 GHz and there is signal over this entire range with some dB level down.
So the 100 ps 3cm twisted pair wire (220 ohm ) visibly gave "good square results" at the end of the long line, but not perfect. Imagine how bad it would look it it was inductive only with very high impedance yet still only a short wire delay! It took a low pass filter on D to restore the peak to peak expected Voltage of 1vpp or in other words, reduce the wideband receiver to match the channel (Ideal Receiver theory) of the signal fundamental due to a tiny mismatched jumper wire.
Here is the frequency domain of the same
Notice the 100 ps is the inverse of f=10GHz and the attentuation goes from 6 to 6.6 dB at 5% f.