On voting, see SCIENTIFIC AMERICAN

If I didn't already understand what they were talking about, that article wouldn't have made much sense. Most of you have heard about IRV and how it is supposed to improve things. Well IRV (by itself) can producee anomalies - including some likely to lead to civil strife. The cause of this defect is that low total candidates are eliminated one at a time -- making the order of their elimination VERY important in deciding the final outcome. It can result in a candidate losing who was the one who could beat any of the others one to on (the "Condorcet candidate"). More seriously, it can result result in an overall leftward (rightward) shift causing a rightward (leftward) shift in the winning candidate! << we'd be fighting in the streets if THAT ever happened >>

The idea proposed by the authors of the article is to FIRST use RCV. Rank choice voting means that for each candidate you determine whether he/she beats each other candidate -- ignoring "by how many places" (how many candidates are in between). The if for some candidate they beat every other candidate by a majority, they are considered the winner (there can be only one such cadidate).

Of course there may not be a "condorcet victor" in which case recount using IRV rules (which DO condsider "order in which candidates are elimianted").

The voting public does not expect perfection but a reasonable degree of accuracy.

The SCIENTIFIC AMERICAN is a psuedo-scientific nationalist propaganda rag.

I could hire a community college electronic tech to design a better voting machine than the expensive ones currently being used, for less money, and within a reasonable timeframe. I can also assure that it's accuracy would be unsurpassed by any currently available machine.

See the Patriot softwareless voting machine.

I think it was aristotle who said "do not strive for accuracy where only and estimation of the truth is possible"

In the case of a voting machine. 99.99 percent accuracy would be acceptable and a realistic expectation.

In Mathematics, the politics of whoever is presenting a proof is totally irrelevant. This is not an area of thought where we are expected to "believe" the presenter. All that matters is if we ourselves can FOLLOW the proof when presented to us (if not, we do NOT accept it). So whether "Scientific American" (or the presenters) are left, right, etc. is rather irrelevant EXCEPT with regard to what they choose or not choose to present.

The politics of Kenneth Aroow (who originally proved that NO democratic system will work over ALL possible distirbutions of preferences) is unknown to me. AND IRRELEVANT (you either can follow and thus accept the proof or you can't). Same with these authors -- their politcis is NOT a reason to ignore what they are saying.

Since there is much talk (here) about the advisability of replacing our current method of deciding voting with IRV in terms of its benefits we should NOT ignore the problems with IRV. Thye are actually rather serious. I agree with you Albert, voting need only be an approximation -- but the point being made is that under certain circumstances IRV can result in a very BAD approximation. Worse, these are cuircumstances under which people would be unlikely to be willing to accept that bad result.

Maybe it would help if I gave a concrete example? One that you all should be able to follow. Suppose the society is divided into three factions, "fascists", "moderates", and "socialists". Suppose we assume that under IRV the "moderates" would be the second choice of both the "fascists" and the "socialists" but that forced to choose a second choice the "moderates" would split evenly left and right <remember -- you ARE forced to choose>

election time one:
fascists 40% moderates 35% socialists 25%
----- under IRV the moderate candidate wins

election time two: Four years later, the sentiments in the society are rapidly shifting leftward. Both the fascist and moderate party are losing support and the socialisrts gaining. The election results
fascists 36% moderates 30% socialists 34%
----- under IRV the FASCIST candidate wins

Albert -- are you willing to accept THIS as an "approximation"? And those of you who imagine that pure IRV eliminates the "lesser of evils" problem, do you not see that it has reappeared, and worse, instead of being when your first choice is irrelevant, comes when your first choice is competitive and has a real chance of winning!

What the authors of the article suggest is that this other scheme of counting votes should be applied BEFORE using IRV to prevent this sort of thing from happening. There may not be a candidate who would have majority support against any other opponent (whose followers chose to take to the streets) -- but if there is one, best to decalre him/her the winner.