> But surely if you didn't have Rshunt, then 100% of the signal would be present at the output, or am I missing something really fundamenal here?
Define "100%".
For the case of a 10 input mixer, 9 inputs fed "silence" (effectively grounded) and one input fed a signal, the 9 resistors on the 9 dead inputs shunt the one live input. Gain for one input to output is 1/10th.
If you have 10 different signals, they can combine in phase, but will usually sum as average power. Then the output rises 3db for each doubling of equal but different inputs: 14% for 2 in, 20% for 4 in, 28% for 4 in, 40% for 10 in, about 50% for 10 in.
The only way to get 100% is to feed ALL the inputs the exact same signal. In that case, why do you need a mixer?
Since even with a 10-man band, there may be a solo that should be at full output, that's about your overall loss. One solo at max comes out at 10%, you mix the 10-mike passages to output 10% or 20% at most.
Another problem. If you input say 1V, the level on the mix-bus is 0.1V or 0.2V. This will leak back through all the mix resistors and appear on the source device's output connector. That probably won't hurt the source. And if the source is zero impedance, the leak-back will be zero at the source. But the source is often padded up to a few hundred ohms. If the mix resistor is 10K and the source is 500 ohms, there will be about 20mV of the full mix at the source's output connector. If you tap that point to monitor "just that source", it is contaminated with the full mix at about 40dB down. This may be negligible or annoying. Shunting-down the mix bus reduces level on the bus and back-leakage.
But the real reason is: amplification is cheap, and the cheapest and/or most-flavorful amp around the studio is often a Mike Amp. Mike amps do not want to see line level, or 1/10th of line level. They usually want 1/100 or less of line level. We WANT around 100:1 of loss in the mixer to avoid overloading the mike amp input. The example shows that (0.01, except the exact number he got was 0.099, same thing).
We also want a low and nominal source impedance for the mike amp's input. Most want to see 100-300 ohms for best response and noise. Ten 10K+10K mix legs would be a 2K source, too high impedance and too high level. A hundred such legs would be a 200 ohm source and 100 down in level, a happy match; but we may not want a 100-in mixer today. So we pencil a 100-in mixer, build as many inputs as we need, then calculate the equivalent of all the arms we didn't build, and throw in a dummy resistor. For any practical number of inputs, this will be 220 ohms. If you build a 64-in passive monster mixer, you can do the math and figure a slightly higher value.
That works because the Mike Amp is "spare" at mix-down. If you don't use a mike amp, sure: you omit the dummy load and see where you are at, see if you have an amp to suit. For a reasonable number of 20K mix-legs, the output will be a low line level and a few-K source. Many line inputs have enough reserve gain to cover the mixer loss, and a few-K signal can be run many feet without cable loading. For 2 to 4 inputs, we often mixed right into the tape recorder's line input; a 2-in mixer was built-in on some recorders. With more inputs, you may need some booster amp. You can always roll your own.
