[EM] No. Condorcet and Hare do not share the same problem with computational complexity and process transparency.

[Closed Limelike Curves]

(Michael, I'd suggest not paying too much attention to rb-j when he gets
like this. He's a Condorcet partisan, and happy to burn down a good reform
just because it's not ideal.)

The issue with IRV isn't that it can't be audited; clearly it can. The
problem is many people won't trust the auditors. And I actually don't think
they should—not because I think the auditors are lying, but because
democracy is only safe so long as we make every effort to safeguard it.

An election without subtotals is effectively closed-source, unscientific.
I can't reproduce it, because I don't have the data. This would be a major
difference compared to today's elections, where I can prove or disprove an
election was fraudulent or miscounted using nothing but public data.

For example, take this great diagram from the Economist:
[image: image.png]
This plot shows precinct-level vote totals for Putin, and it conclusively
proves his election was fraudulent. Polling stations reported numbers
ending in 0 and 5 *far* more often than can be explained by
chance, implying these numbers are just made-up. There's also clear
evidence of ballot-box stuffing: Putin gets huge vote totals in the
elections where "turnout" is close to 100%, suggesting many of those votes
simply weren't real.

This kind of graphical analysis is enough for me to prove that Russian
elections are fraudulent. But the key lesson is I know that even though the
elections *weren't* audited. I know it from the precinct subtotals. This is
what keeps our democracy safe. It's statisticians and political scientists
closely monitoring every set of results, picking apart any inconsistency in
the data and using it to expose fraud. If the last presidential election
had been fraudulent, one of us *would* have figured it out.

Another example is the NC-9 election in 2019
<https://www.brookings.edu/articles/understanding-the-election-scandal-in-north-carolinas-9th-district/>.
Here, statisticians detected voter fraud by noticing bizarrely high
vote shares for candidates in some precincts. Ultimately, it was discovered
that Republican staffers attempted to fraudulently harvest ballots from
elderly voters, pressuring them to sign off on absentee votes that had been
filled out for them. The results were thrown out by a judge and a new
election was held, where the results were reversed. That precinct data was
the key to uncovering the fraud. I find it extremely concerning that with
IRV, we might not be able to prove this had happened.

Reporting first-choices wouldn't be enough, because not all elections are
determined by first-counts (or else IRV would be plurality). We need to be
able to calculate results from what we have publicly-available.

As a rule of thumb, the average precinct has 600 voters. If you have k
candidates, you need maybe 5 * k! voters to be able to release the
ballots—that means you have only about a 20% chance of being able to
identify any particular person from their ballot, assuming any particular
permutation of ballots is equally likely. (If not, identifying voters gets
even easier. In practice, you could probably do it for k=4 by being a bit
clever.) These numbers cross at k=5, which means the results of any
election with 5 or more candidates can't be publicly verified without
exposing voters to the possibility of harassment, intimidation, or bribery.
These aren't a problem for O(k^2) or O(k) methods like Condorcet or score.

I know American elections are safe and secure, because I can prove it
myself, or read the work of scholars who have proven it themselves. You can
try and replace that with audits under IRV, but that just moves the
problem: I have to trust the auditors instead. If you ask me, "How do you
know there wasn't fraud?" I couldn't give you an answer, except saying "I
trust the auditors." For voters who don't trust the system already—and
unfortunately, many Americans don't—that's not going to be enough, and they
wouldn't be totally wrong to worry.

On Wed, Mar 20, 2024 at 12:36 PM Michael Garman <
michael.garman at rankthevote.us> wrote:

> Do “general trends” explain the discrepancy in levels of representation
> between IRV and non-IRV cities with similar characteristics?
>
> IRV is compatible with risk limiting audits. Some people will always cry
> fraud when they lose, but IRV can be audited rigorously.
>
>
> On Wed, Mar 20, 2024 at 8:25 PM Michael Garman <
> michael.garman at rankthevote.us> wrote:
>
>> My goodness, you’re disingenuous! Or perhaps you’re ignorant — should I
>> give you the benefit of the doubt? The Alaska counting delay was due to a
>> law requiring them to count mail ballots received up to two weeks after the
>> election. It had nothing to do with IRV!
>>
>>
>> On Wed, Mar 20, 2024 at 8:23 PM robert bristow-johnson <
>> rbj at audioimagination.com> wrote:
>>
>>>
>>>
>>> > On 03/20/2024 2:04 PM EDT Michael Garman <
>>> michael.garman at rankthevote.us> wrote:
>>> >
>>> >
>>> > Also, regarding the Otis interview, here's the transcript (
>>> https://www.wbur.org/onpoint/2024/03/19/could-ranked-choice-voting-take-the-poison-out-of-politics).
>>> This is her description of the tabulation process. No lies detected.
>>> >
>>>
>>> My goodness, you're disingenuous.  Or, perhaps, you just don't know what
>>> you're talking about.  Should I give you that benefit of doubt?
>>>
>>> ...
>>> OTIS: Voters in Alaska will not have to do the strategic voting. They
>>> just get to ranked honestly. Voters in some other states are going to have
>>> to do some tough math.
>>> ...
>>>
>>> Problem is that in Alaska in August 2022 and in Burlington Vermont in
>>> 2009, voters would have to do even tougher math to avoid being burned in
>>> those two elections.
>>>
>>> Voters for the loser in the final round would have avoided causing the
>>> election of their least favorite candidate by voting tactically.  But
>>> because they vote sincerely, they got burned.
>>>
>>> CHAKRABARTI: What's your best argument for why you think moving to this
>>> system is better for American democracy?
>>>
>>> OTIS: Oh, there are so many reasons. I think the biggest one is just
>>> getting majority winners.
>>> ...
>>>
>>> That's a lie.  And FV knows it.
>>>
>>> CHAKRABARTI: What's your best argument for why you think moving to this
>>> system is better for American democracy?
>>>
>>> OTIS: Oh, there are so many reasons. I think the biggest one is just
>>> getting majority winners.
>>> ...
>>>
>>> Another lie.
>>>
>>> ...
>>> (OTIS cont.): And ranked choice has been shown to improve representation
>>> for women and people of color.
>>> ...
>>>
>>> That has also not been shown.  There is no statistical significance that
>>> shows that RCV other than trends happening in the 21st century has improved
>>> representation of women and POC.
>>>
>>> ...
>>> OTIS: The ranked choice voting is like an instant runoff. The votes are
>>> tabulated in a matter of a couple of seconds.
>>> ...
>>>
>>> What's disingenuous here is that, for statewide RCV, it takes ***days***
>>> to centralize the ballot data before the IRV tabulation can begin.  Alaska
>>> took 15 days in November 2022.  Maine took 10 days in 2018.
>>>
>>> Computers are fast.  Big deal.  But that does not make IRV "Instant".
>>>
>>> Without Precinct Summability, we lose a critical part of Process
>>> Transparency we need to keep some Trumpist election official from "just
>>> find[ing], uh, 11780, uh votes."  Fortunately Raffensberger was not Jeffery
>>> Clark.  But if he *was* corrupt, the first thing that prevents a corrupt
>>> official from padding the numbers is Precinct Summability and the fact that
>>> polling places had already published the necessary data for us to know how
>>> the election will turn out.  We have that with FPTP and we lose it with IRV.
>>>
>>> ...
>>> OTIS:  And ranked choice voting is the best way to take the temperature
>>> of the electorate and elect someone who has the broadest support and can
>>> best represent the voters.
>>> ...
>>>
>>> Again a falsehood in both Alaska 2022 and Burlington 2009.  And FV is
>>> aware of it.  The candidate with broadest support (by *any* mathematical
>>> measure that is not defined by the IRV procedure) in those two elections
>>> was the candidate eliminated in the semifinal round.
>>>
>>>
>>> --
>>>
>>> r b-j . _ . _ . _ . _ rbj at audioimagination.com
>>>
>>> "Imagination is more important than knowledge."
>>>
>>> .
>>> .
>>> .
>>> ----
>>> Election-Methods mailing list - see https://electorama.com/em for list
>>> info
>>>
>> ----
> Election-Methods mailing list - see https://electorama.com/em for list
> info
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20240320/a176b191/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 1639865 bytes
Desc: not available
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20240320/a176b191/attachment-0001.png>