I was trying to make the point that the average forced voter might be little better than random at actually identifying the candidate that would lead to the greatest fulfillment of his preferences.
You are right that the difference doesn’t matter in the end and I would certainly extend my reply to cases where “little better than random” voters are classed as ‘noise’. Adding “little better than random” voters is (practically) no problem, adding worse than random voters would be a problem. The latter is actually a possibility worth considering as at least arguable for some demographics. As dbaupp said, there are considerations along those lines that go either way.
I note that even assuming the “average” additional voter is noise (and that the mean, median, mode and the ones denoting each border of the interquartile range are too) doesn’t result in a plausible “just add noise” picture. I go as far as to say that if one in twenty of the new voters have a clue and the rest are “little better than random” the voting system wins. For the additional voters en masse to “just add noise” it would require none (or close to none) of the new voters to make meaningful better-than-random votes. This is unlikely and in fact isn’t compatible with the historical data we have on how compulsory voting actually occurs in practice.
(This just affirms your observation that the differences in our positions aren’t merely the result of different usages of the term ‘noise’.)
You are right that the difference doesn’t matter in the end and I would certainly extend my reply to cases where “little better than random” voters are classed as ‘noise’. Adding “little better than random” voters is (practically) no problem, adding worse than random voters would be a problem. The latter is actually a possibility worth considering as at least arguable for some demographics. As dbaupp said, there are considerations along those lines that go either way.
I note that even assuming the “average” additional voter is noise (and that the mean, median, mode and the ones denoting each border of the interquartile range are too) doesn’t result in a plausible “just add noise” picture. I go as far as to say that if one in twenty of the new voters have a clue and the rest are “little better than random” the voting system wins. For the additional voters en masse to “just add noise” it would require none (or close to none) of the new voters to make meaningful better-than-random votes. This is unlikely and in fact isn’t compatible with the historical data we have on how compulsory voting actually occurs in practice.
(This just affirms your observation that the differences in our positions aren’t merely the result of different usages of the term ‘noise’.)