[Election-Methods] Dopp: 8. “Difficult and inefficient to manually audit”

[Abd ul-Rahman Lomax]

>8. Dopp: “Difficult and inefficient to manually audit
”
>
>  IRV can be manually audited just as well as 
> vote-for-one elections, although it does take 
> more effort (since voters must be allowed to 
> express more information on their ballot). A 
> manual audit can either be done using a random 
> sample of ballots from all jurisdictions, or a 
> random sample of ballots from a random sample 
> of voting machines, or by a complete re-tally 
> from a random sample of voting machines. A 
> complete re-tally of all ballots (a recount) 
> is, of course, possible but unnecessary unless a court recount is ordered.

Notice the slipperiness. Dopp says it's difficult 
and inefficient, FairVote says it can be manually 
audited "just as well" but "it does take more 
effort." Those are contradictory. Taking more 
effort is not "just as well." Dopp did not say it 
was impossible to manually audit. This is typical 
FairVote argument, Richie is unfazeable. You can 
tell him his pants are on fire and he'll say that 
they feel no hotter than usual. Which might be 
true *and* his pants are on fire.

The counting process is far more complex, so the 
auditing is far more complex. Ordinarily, with an 
audit, one can pick a sample precinct and count 
it. Period. But with IRV, one has to know what 
round, what candidates were eliminated, before 
the counting will match what was reported. You 
have to go through the same process, depending on 
the same instructions that were given from the 
central office. I'd estimate that, in an election 
that went through many rounds, the difficulty 
could be ten times as great as with simpler, 
precinct-summable methods. With a precinct 
summable method, you just count what is on the 
ballots, you don't need to know anything from 
other precincts. And then you can compare this 
with what was reported officially from the precinct.

Note that IRV and other Later-no-Harm methods 
can't be precinct-summed. This is actually 
unusual, most election methods can be summed. 
Approval is obvious: you just count all the 
votes. Determining a Condorcet winner (candidate 
who would beat all candidates in a pairwise 
election with them) is a matter of summing the 
votes in all pairwise elections, what the 
precinct reports is a matrix of results, and the 
matrixes are then summed. So, though it takes 
more data, Condorcet methods are precinct-summable.

Continued with:
>9. Dopp: “Could necessitate counting all 
>presidential votes in Washington, D.C.
”