[EM] piling on against IRV

[robert bristow-johnson]

On May 10, 2010, at 6:23 PM, Kathy Dopp wrote:

>> Date: Mon, 10 May 2010 14:55:19 -0400
>> From: robert bristow-johnson <rbj at audioimagination.com>
>> To: election-methods Methods <election-methods at lists.electorama.com>
>
>>>>> However, that does not alter the fact that .
>>>>
>>>> they have the choice in *either* TTR or IRV.  and *both* can fail  
>>>> to
>>>> arrive at the Condorcet winner
>>>
>>>
>>> I said nothing about the fact that IRV/STV tends to eliminate the
>>> Condorcet winner prior to the final round,
>>
>> i never said you did.  you were taking about IRV failing to elect the
>> "majority winner" (think about who you mean by such).  in the same
>> manner, so also does TTR.
>
> False claim. *You* were the person to bring up "Condorcet" above,  
> not I.

i never said otherwise.  misrepresentation of another's words.  Strike  
1.

you were the person to continue to complain about IRV failing to elect  
the "majority winner" while simultaneously claiming that somehow TTR  
does when both would elect the same candidate under comparable  
conditions.


>> TTR doesn't find the majority winner either.  it's no better than
>
> False claim. Find *one* TTR election where there was not a majority
> winner out of all voters who cast votes in the TTR election.

the TTR *together* with the original election (that doesn't have a  
sufficient majority to resolve the election according to the rules)  
*likely* elects the same candidate as IRV (there is a version of IRV  
that eliminates all but the top-two candidates that is more comparable  
to delayed TTR, but it makes no difference in the IRV failure example  
we both have referenced).  both have the same likelihood of sending  
the wrong pair of candidates to the runoff.  neither would have  
elected the "majority winner" in Burlington in 2009.

> You
> cannot, unless you fabricate yet another wild definition of "majority"
> where "majority" means a majority of voters, not in the election, but
> in the prior election.

i see.  so electing Bob Kiss in Burlington 2009 via IRV is not the  
majority candidate, but since given the TTR rules, the same two  
candidates would have ended up in the delayed runoff, if Bob Kiss was  
elected then, by the same voters voting the same way as they did in  
2009, somehow that same candidate *is* the majority candidate.  and if  
far fewer Kiss supporters (those that had any preference of Kiss over  
Wright) had made it to the runoff, if Wright was elected, that would  
be the "majority candidate" even though we already know from the  
information collected that Wright was less preferred than Kiss and he  
was also less preferred than Montroll.

> Any wild distortion of what "majority" means

you can only speak of "majority" when a candidate receives more than  
half of the votes cast.  narrowing down the race to two candidates  
will always get *a* majority (unless they tie), but not necessarily  
*the* unambiguous majority.  somehow, if the race is narrowed to two  
candidates via TTR, you consider that *the* majority.  somehow, the  
IRV supporters consider that the two candidates left standing after  
their kabuki dance of transferred votes and the candidate with more  
votes in that pair results in *the* majority.  but neither of you are  
considering the majority that is evident in other pairs of candidates,  
which is why both your methods fail, in comparison to Condorcet.

>>> Again:
>>>
>>> Voters have the CHOICE to vote or not in the real runoff elections,
>>> unlike with STV/IRV where the more candidates who run for office,
>>
>> that (the latter regarding IRV) is false and you've never shown it to
>> be true.  you just keep repeating the same canard hoping, like
>> Goebbels, that a lie repeated oft enough evolves in status from
>> outright falsehood to plausible notion to gospel truth.  it doesn't
>> but you may continue to pretend otherwise.
>>
>>
>>> the
>>> greater number of voters are excluded INVOLUNTARILY from  
>>> participating
>>> in the final IRV/STV round.
>>
>> repeating the falsehood doesn't make it true.
>
> False again. Come on, I could explain this to any 10 year old in less
> than 30 seconds.

that says nothing.  my kids mostly accept mostly whatever i tell them  
(but i have explained the difference between Plurality, IRV, and  
Condorcet to them and they told me they understand it).  they were 10  
and 12 at the time.

> Even in Burlington, VT I'm sure that there were fewer voters' ballots
> being considered in the final counting round than in the 1st round.

only 7% fewer.  compared to our experience of at least 45% fewer in  
delayed TTR elections.

i wouldn't used the word "disenfranchise".  since you continue to  
insist that people are "involuntarily excluded" (they aren't, but i'll  
use your language), your method "disenfranchises" 45% while IRV  
"disenfranchises" 7% in the recorded experience that Burlington has.   
but really neither method disenfranchises anyone (except TTR might for  
a person that can't make it to both elections and finds it difficult  
to cast an absentee ballot).


> I know you are just pretending that 100% of the voters in every
> election have their votes counted in the final STV/IRV counting rounds
> now,

another lie.  i've said 93% in at least one election.  Strike 2.

> because no one is remotely as ignorant or obtuse as you pretend,
> and any ten year old could understand the exclusion of voters by the
> final STV/IRV counting rounds.
>
>>> IRV/STV historically has maintained the 2-party dominant system  
>>> where
>>> it has been tried in 2-party dominant locations.
>>
>> well, FPTP creates pressure in a 3-party town to return to a 2-party
>> dominant system.  the 2 parties that have the most in common and
>
> Ah, the old "change the subject" tactic.  I never claimed, as you
> imply, that FPTP did not encourage the 2-party system.

my response was to compare to IRV.  i do not deny that IRV can also  
suppress a 3rd party because, when it fails to elect the CW, it has  
been known to punish a group of voters for voting sincerely.  i've  
illustrated this already with the GOP Prog-haters that are  
"encouraged" by IRV to vote for the Democrat so that they don't end up  
electing the Prog.  but FPTP is worse, which is *partially* what we  
have returned to in Burlington.  now, in Burlington, the TTR happens  
only if the Plurality winner is less than 40%, otherwise it's FPTP.

> I was simply exposing the lie you told about IRV/STV supposedly
> helping 3rd party candidates.

it *does* compared to FPTP.  it fails in comparison to Condorcet.

>> together command a clear majority will now have to run a consolidated
>> candidate in order to securely beat the 3rd party with the nearly
>> opposite ideology.  that's even what Geirzynski alludes to.  he  
>> thinks
>> it's a good thing, i think it's a bad thing.  it is the beginning of
>> the end of the separateness of those 2 parties that have the most in
>> common.
>
> And you lied about IRV/STV not doing the same as plurality in this
> regard,

IRV (or even TTR) would have elected Bob Kiss (the Prog).  Plurality  
would have elected Kurt Wright (who was pairwise less preferred than  
either Kiss or Montroll).  the numbers are pretty clear.

> just like Fairytale vote lied about IRV/STV finding majority
> winners,

i'm not FairVote.

> solving the spoiler problem and allowing voters to cast
> sincere ballots,....

i never said that.  i am not just like FairVote (as Rob or Terry know  
well).

> The list of lies by IRV/STV proponents is enormous,

yeah, but you never lie.  even when you're lying, it's not a lie.

> and yet you
> continue to tell the same disproven lies and try to change the subject
> to confuse your audience into thinking the topic was about plurality.

nope, i have never said that IRV solves the spoiler problem and have  
pointed out since March 2009 that Wright, the GOP candidate and FPTP  
winner, was the IRV spoiler.  i have pointed out that IRV *does*  
punish a group of voters (the GOP Prog-haters) for voting sincerely  
(therefore does not solve the voting strategy problem).  and i have  
pointed out that if a sufficient number of Wright voters had changed  
their vote to Kiss, it would result in Kiss losing the election (the  
non-monotonicity problem).  and i have pointed out that if there are  
more than 3 candidates, the number of piles for IRV is unmanageable  
for precinct summability and have never said that precinct summability  
is not a desirable property of an election method.  Condorcet, while  
requiring more piles than FPTP, still has a manageable number of piles  
and i would call it "precinct summable".


>
>>
>>> In fact IRV is
>>> designed ideally to prevent minor party candidates from interfering
>>> with the 2-party system.
>>
>> that's bullshit.  you have no idea what you are talking about.  you
>
>
> Very funny. Now are you claiming falsely that IRV wasn't sold in the
> US on the basis that it could have prevented Ralph Nader voters from
> disrupting Gore's win in Florida? (i.e. make it so the most popular of
> the 2 major parties' winners wins office without disruption.)
>
> You IRV/STV proponents

whoa!  that's the mother of all misrepresentations.  Strike 3.

> are very very good at reinventing history.
>
> So, you've added quite a few tactics to this email:
>
> 1. reinventing history
>
> 2. changing the subject
>
> 3. another creative definition of "majority" (not out of this
> election, but the last election's voters)
>
> 4. pretending I was speaking about things you mentioned first
>
> You are really perfecting such discourse Robert.
>
>> misrepresent what the "design" was about.  what i have been saying in
>> my complaints about IRV is that it *failed* to do what it's designed
>> to do.
>
>
> IRV is ideally designed to avoid having minor parties from interfering
> with the top two major parties,

untrue and you never prove it.  (because to prove it you need to  
establish what the designers intent were.)

IRV was designed to support emerging 3rd parties so that people could  
feel free to vote for their long-shot 3rd-party candidate, cover their  
ass with a 2nd-choice of the major party candidate of their choice and  
not have to worry that *they* helped elect the other major party  
candidate that they hate.  IRV *fails* in some cases (and Burlington  
2009 was one of those cases) to accomplish that design goal, but it  
still was the design goal.  when IRV fails that design goal, the  
failure affects a minority of voters.  essentially IRV in Burlington  
transferred the burden of strategic voting from the liberal majority  
to the conservative minority.  this is a fact that IRV proponents  
don't like to recognize.


> although its designers may not have
> realized what IRV was designed to do.

ah, the paragon of logic.  the designers, who set out the design  
specs, do not realize what the design specs are.

> IRV certainly is designed well
> if the goal is to discourage 3rd party political participation where
> there are 2-major parties because people, like they did in Burlington,
> immediately realize that a sincere vote, where one of the two top
> contenders is not listed, can create havoc.

creating havoc wasn't part of the design.  but i won't deny that in  
2009, the failure of IRV to elect the unambiguous majority-supported  
candidate did lead to at least 4 pathologies.  but what we returned to  
would not have fixed it.  if we had the 40% FPTP/TTR in 2009, either  
the same candidate would have been elected or an even less supported  
candidate would have been elected.

> In Burlington, the two top
> contenders happen to be the Progressive and the Democrat, with the
> Republicans being the 3rd, not so minor, party.

never said the Repub was minor.  but his support was less than either  
the IRV winner (a fact established by the IRV runoff) or the Condorcet  
winner (who never got into the runoff, which is what caused the 4  
anomalies in the election).

> True, IRV/STV will always fail to avoid the havoc of a 3rd party
> candidate whenever the 3rd party candidate becomes competitive enough,

FPTP and TTR would have at least the same or even worse "havoc".   
neither would have done better than IRV in Burlington in 2009.  only  
Condorcet would have avoided the election anomalies.

> as the Republicans are in Burlington - competitive to knock out the
> favorite candidate prior to the last round.

and he would have done it also with TTR.  but the difference is we  
would be ignorant of the fact without the rank ballot which has now  
been done away with in Burlington.

> Exactly why IRV/STV were repealed in Burlington, when it failed to
> prevent a 3rd party from wreaking havoc by eliminating the most
> popular candidate prior to the final round.

the political reasons for why IRV was repealed is more due to the  
perpetration of falsehoods (some of the same falsehoods you continue  
to repeat) by the Republicans who, because they had the FPTP winner,  
could not accept that their "winner" was a loser.  they would have  
likely been just as upset if their FPTP winning candidate had lost to  
the Condorcet winner.


>> not always (sometimes IRV *does* work), but it surely failed
>> to accomplish the very goals for which it was designed to accomplish
>> in Burlington in 2009.
>
> Yes, failed to keep that nasty-old-3rd party from causing voters to
> elect the least favorite of the top two candidates (Progressives and
> Dems in Burlington, VT).

no, Bob Kiss was *not* the least favorite.  Bob Kiss was the 2nd most  
preferred candidate.  only the Condorcet winner was more favored by  
voters than the IRV winner in that election.  TTR would have risked  
electing the 3rd most preferred candidate.


>> it does better (certainly not perfectly, not as well as Condorcet)
>> than FPTP in "finding the majority winner" (using your language.  you
>> keep repeating this falsehood and you have never offered anything to
>> support it.  you then try to obfuscate the discussion with talk of
>> "voting rights" which is non-sequitur because "rights" have nothing  
>> to
>> do with it.  the same rules that apply to those who claim their  
>> rights
>> were violated are the same rules that applied to everyone else.
>
>
> I never said that IRV/STV were worse than plurality  *in every way*.

but you said it is always worse than plurality.  and it is almost  
always *better* than plurality in getting a majority favored  
candidate.  electing the 2nd most favored candidate is better than  
electing the 3rd.

> I
> maintain that IRV/STV are worse than plurality in sufficient number of
> ways to make it a huge step backwards from far simpler, more
> auditable, fairer exising FPTP methods.

and i have pointed out how plurality or TTR does worse than IRV in the  
specific case in Burlington, even in the case where IRV failed to  
accomplish the design goals

3 strikes this time, Kathy.  you have explicitly misrepresented my  
words at least 3 times in this very post.  and i think deliberately.   
at least, you have exhausted all benefit-of-doubt.  i won't do it this  
time, but i am a whisker away from kill-filing anything you post to  
the EM list as i had to Abd's posts.  (direct email will still make it.)

--

r b-j                  rbj at audioimagination.com

"Imagination is more important than knowledge."