[EM] IRV vs Plurality (Dave Ketchum)

[Juho]

On Jan 16, 2010, at 5:51 AM, Kathy Dopp wrote:

>>> On Jan 14, 2010, at 2:13 AM, Dave Ketchum wrote:
>>>> On Jan 13, 2010, at 4:49 AM, Juho wrote:

>>>>> 2) IRV is easier to count manually. Condorcet gets quite tedious

>>>>> to count manually when the number of candidates and voters goes
>>>>> up. One can use some tricks and shortcuts to speed up manual
>
> Condorcet is SO much easier and quicker and simpler to count than IRV
> if one simply tallies one n x n matrix for each precinct and sums the
> corresponding positions for all the precincts.  No need to wait for
> all the absentee and provisional ballots, no need for centralized
> counting, no need to sort and resort ballots into dozens of piles of
> ballots, or even worse with STV keep track of which portion of which
> ballot goes into which pile (tearing or cutting up the ballots would
> help in that manual counting nightmare.)
>
>
>>>>> Condorcet counting but IRV probably still beats it from this point
>>>>> of view. Manual counting was the only way to count for a long
>>>>> time. Nowadays we have computers and Condorcet tabulation should
>>>>> thus be no problem at all (at least in places where computers are
>>>>> available). But this is one reason why IRV has taken an early  
>>>>> lead.
>>>>
>>>> When an election district has only one polling place, life is  
>>>> simple.
>
> Yes. Another point against IRV/STV is no scalability of manual
> counting. Condorcet is infinitely scalable since it is as simple to
> manually count dozens or hundreds of precincts as it is to count one,
> without moving all the ballots to one central location.

When talking about history and the early lead of IRV I referred to the  
times when both ballot reading and recording as well as summing up the  
results to piles or matrixes and all the way up to declaring the final  
results was manual, i.e. before the time of computers. It seems that I  
assumed that telephone or telegraph was already invented when I  
referred at some point also to the option of local counting of IRV  
results with only centralized control of that counting process  
(=advice from the central counting office on which candidates to  
eliminate next etc.).

IRV has at least a tradition of being counted manually in this way.  
I'm not aware of any similar Condorcet tradition in large elections. I  
believe already Ramon Llull had a simple sequential calculation method  
but one would need some enhancements to that method if one wants to  
cover the most popular Condorcent methods.

The problem that I expected in Condorcet (variants that require more  
info than what the Llull process uses) is the size of the matrix and  
the possible need to have some explicit matrix (or several, e.g. on  
paper) where the results are collected when the ballots are read one  
by one. In IRV it is possible to put the ballots in different piles  
since IRV takes into account only the top (non-eliminated) candidate  
of each ballot at one point in time during the counting process. In  
Condorcet, if one wants to count the full matrix, each ballot  
influences multiple entries in the matrix, so the ballots themselves  
can not be used as tools to mark the number of points in some entry in  
the matrix (unless doing the whole process sequentially, counting one  
part of the matrix at each round).

 From "counting verification" point of view the explicit physical  
ballots may be easier to track during the counting process than  
numeric values (or ticks) in the matrix.

I believe a typical manual IRV counting process uses at most as many  
piles as there are candidates a one point in time (each counter having  
his/her own piles is not a problem since the piles are "physically  
summable") while a Condorcet counting process might use n*n matrix  
entries.

(jumping to another topic) From voting security point of view the  
summability of Condorcet results offers better protection of voter  
privacy since individual votes can not be identified any more after  
they have been once included in the matrix (that contains sufficient  
information for most Condorcet methods to calculate the final results  
in some central location). In IRV it is more practical to store the  
actual ballots, not e.g. the number of votes per each permutation of  
candidates, so the individual votes may be visible to all the  
verifiers, maybe also to all in Internet.

Juho