[Election-Methods] Dopp: 7. “Difficult and time-consuming to manually count”

[Abd ul-Rahman Lomax]

>7. Dopp: “Difficult and time-consuming to manually count
”
>
>Manual counts can take slightly longer than 
>vote-for-one elections, but aren't difficult, 
>unless many different races on a ballot need to 
>go to a runoff count. As cited earlier, Irish 
>election administrators can count more than a 
>million ballots by hand in hotly contested 
>presidential elections in one standard workday.

"Slightly"? Most hand-counted elections are done 
in a few hours, I think, because they can be 
counted precinct by precinct, and the totals 
added up. That summation doesn't work with IRV 
when it goes into runoff rounds, because the 
results of one round then control how the next 
round is counted. So all precincts must wait for 
a central facility, collecting the results from 
all precints or counting stations, to complete 
the first round and report eliminations, before 
they count the next round. If an error is made 
that affects an elimination, all counts done 
depending on the incorrect elimination must be 
redone. An error anywhere can affect the next round counting in all precincts.

Now, about that Irish Presidential election. 
FairVote misrepresented it. The last contested 
presidential election was in 1997, and it took 
two days to count it, not one. And it is 
Contingent Vote, though they call it Alternative 
Vote, I think, which is easier to count than most 
IRV proposals for the U.S. (Another example of 
the slipperiness in the name "Instant-runoff 
voting.) This method only has two rounds, all 
eliminations are done in a batch, leaving the top 
two for the last round. I wrote the following in 
another response to these arguments which had 
been forwarded to a list that Kathy Dopp participates in:

Um, hotly contested? See 
http://en.wikipedia.org/wiki/Irish_presidential_election,_2004, which has:

The Irish presidential election of 2004 was set 
for <http://en.wikipedia.org/wiki/October_22>22 
October <http://en.wikipedia.org/wiki/2004>2004. 
However, nominations closed at noon on 
<http://en.wikipedia.org/wiki/October_1>1 October 
and the incumbent 
<http://en.wikipedia.org/wiki/President_of_Ireland>president, 
<http://en.wikipedia.org/wiki/Mary_McAleese>Mary 
McAleese, who had nominated herself in accordance 
with the provisions of the 
<http://en.wikipedia.org/wiki/Constitution_of_Ireland>Constitution, 
was the only person nominated. Accordingly she 
was re-elected for a second seven-year term of 
office without the need to hold a poll. This is 
the third time a president has been returned 
unopposed, following 
<http://en.wikipedia.org/wiki/Sean_T._O%27Kelly>President 
O'Kelly in 
<http://en.wikipedia.org/wiki/1952>1952 and 
<http://en.wikipedia.org/wiki/Patrick_Hillery>President 
Hillery in 
<http://en.wikipedia.org/wiki/1983>1983. 
President McAleese's re-inauguration took place 
on <http://en.wikipedia.org/wiki/November_11>11 November.

Okay, what about the previous Irish Presidential election?
http://en.wikipedia.org/wiki/Irish_presidential_election%2C_1997

McAleese won that one with a reported 45.24% of 
the first preference vote, with her closest rival 
having 29.30% of that vote. From my experience 
examining IRV elections, there is about zero 
chance that the rival could come back and win, 
with McAleese having less than 5% to go. If we 
assume that the closest rival stays the 
runner-up, what generally happens is that the 
vote transfers later go roughly according to the 
early preference ratios, so, without knowing the 
final results after transfers, other things being 
equal, I'd predict that McAleese would get 
(45.24%)*(100% - 45.24% - 29.30%) of the vote in 
the last round (neglecting exhausted ballots 
entirely), and would thus win an additional 
11.52% of the vote from transfers, giving her a 
total of 56.76% of the vote. It will be 
interesting to see the actual results when I have 
a chance. What was "hotly contested about this is 
that there were five candidates altogether, which 
is the most that have ever been seen. But it certainly was not close.

According to one source, McAleese ended up, in 
the second round, with 58.7% on second count. 
Pretty close to prediction. With an election that 
is truly "hotly contested," there would have been 
three rounds, not two, so the counting would have 
taken almost half again as long. If no mistakes 
were made. [This was incorrect, because at that 
point I had not discovered that the method was 
Contingent Vote.] (Fewer rounds occur when more 
than one candidate gets such a low vote count 
that coming back to survive is impossible, this 
is called "batch elimination.") In San Francisco, 
there were *19* rounds of counting in one 
election! (There were 22 candidates. Both top-two 
runoff and IRV seem to encourage larger numbers 
of candidates to run, since they can do so 
without spoiling the election. Is that a good 
thing? Probably. But it's also expensive, and 
explains why San Francisco had so many runoff elections before IRV.)

However, other things aren't necessarily equal, 
and this is a partisan election, so, in 1990, 
there was indeed a "comeback" election, the first 
preference leader lost to the runner-up. As we 
might expect, there appears to have been 
vote-splitting between the two runner-up parties, 
resolved by the STV method. Without knowing more 
about Irish politics -- I know practically 
nothing -- I think chances are good that this was a proper result.

Now, how long did it take to count these 
elections? Well, 2004 was pretty easy! No 
ballots. 1997 was two rounds. See 
http://www.cnn.com/WORLD/9710/31/ireland.elex/index.html.

"Voters cast ballots Thursday, and when the first 
round of counting was finished Friday, McAleese 
had 45.2 percent of the vote, easily besting 
second-place finisher Mary Banotti, who had 29.3 
percent. Three other candidates trailed far behind.

Under the Irish election system, voters selected 
both a first and a second choice in the 
presidential race. After the first count, the 
three trailing candidates were eliminated and a 
second round of counting took place, with their 
votes redistributed to voters' second 
preferences. A candidate had to win a majority to claim the presidency."

Continued with
Dopp: 8. “Difficult and inefficient to manually audit
”