What evidence is there that voluntary voting doesn't just add noise to the selection process?

That's a serious question: voluntary voting means that a higher percentage of the voters are in a blue-vs-green mindset (since they are more likely to vote than someone who has weak preferences), while compulsory voting gives a more accurate picture of the feelings of the entire population, even if that involves people who donkey-vote etc.

(That's not to say your point isn't valid, just that the sword cuts both ways.)