[Election-Methods] Fwd: [LWVTopics] IRV Voting

[Abd ul-Rahman Lomax]

At 01:15 AM 5/12/2008, Jonathan Lundell wrote:
>On May 11, 2008, at 10:00 PM, Kathy Dopp wrote:
>
>>However, I had not even thought of the need to
>>count all state-level races at the State-level, rather than the county
>>level, or about some of the other issues you mentioned.
>
>I don't see that as a significant problem these days. There has to be
>a roll-up *somewhere*, and it's probably easier for the counties to
>send the ballots, via some electronic method, to a central counting
>facility where they're counted once, rathr than do a bunch of local
>counts and send those results.

It's a problem. It's one thing for an election in a single city, but 
a state-wide election in a large state, that's a lot of ballot 
movement, if it is physical. I agree the problem can be solved by 
moving ballot images instead of ballots.

voting security people have become less than thrilled about automated 
counting, and, I agree, if that is all that is available, it's 
dangerous. If the ballots can be hand-counted in local batches, and 
the ballots are kept locally, it's more secure and more difficult to 
manipulate wholesale.

Ballots are currently hand counted using fairly tedious procedures 
involving clerks and observers and officers. If the counting is moved 
to a central location, all those people need be there, instead of 
being distributed among the polling places

The essential point, though, is that IRV is much more complex to 
count than certain other methods on the table. The complex counting 
makes verification more difficult. The sequential elimination in IRV 
is quirky, each round depends on the round before, so it is not 
enough to look at a sample set of ballots and extrapolate to the 
general results. When many rounds are involved, the method becomes 
sensitive to the far more common ties that take place.

I've argued that if IRV were an excellent method, the extra expense 
might be worth it. But it is not. Most IRV elections are concentrated 
in Australia, where STV is used for proportional representation, and, 
for that purpose, the complexity may well be justified, and then it 
probably seemed natural to use IRV. STV has problems similar to IRV, 
*but they don't become important until the last member being elected 
from a district is being chosen.* Single-winner STV is IRV, and all 
the relatively minor problems of STV become concentrated in IRV.