[EM] My CTE post to EM
C.Benham
cbenham at adam.com.au
Sun Oct 8 15:33:53 PDT 2023
Michael,
> What if there are more candidates? What if there’s an additional Bus?
What in your terminology, in this context, is a "Bus"?
> An optional added rule could specify that the 1st elimination is of
> whichever of the 2 candidates with worst base-method count-score is
> pair-beaten by the other.
That doesn't make any difference in the example I gave.
Chris
On 9/10/2023 8:13 am, Michael Ossipoff wrote:
> Oops!! I think I list-replied to an individual e-mail. I appreciate
> the consideration, not posting a proposal-debunk, but I don’t consider
> the proposal debunked.
>
> Anyway, I’d mistakenly assumed that I was replying to a posting, but
> now that I’ve sent my reply to the list, I might as well leave it
> there, when there’s more to add.
>
> Even if randomness determines which of the 3 candidates gets
> eliminated 1st, there’s a 2/3 probability that BF loses.
>
> …backfire is equally likely as success.
>
> But it isn’t random:
>
> You’ve insincerely bottom-ranked CW.
>
> …thereby increasing hir probability of being the 1st eliminated.
>
> …&, with 3 candidates, you’ve upranked Bus from bottom, decreasing hir
> chance of being 1st eliminated.
>
> An optional added rule could specify that the 1st elimination is of
> whichever of the 2 candidates with worst base-method count-score is
> pair-beaten by the other.
>
> What if there are more candidates? What if there’s an additional Bus?
>
> Then BF loses, no matter who is eliminated.
>
> That’s fact obtains when there are more than 1 Bus, but no candidates
> who aren’t CW, BF or a Bus.
>
> I’ll collectively refer to CW, BF & Bus(es) as “the principals”.
>
> Any candidates who aren’t principals, I refer to as “other candidates”.
>
> If there’s an other candidate who pair-beats the Bus, & gets
> eliminated 1st, of course that takes down the Bus. But the same thing
> could happen to any one of the principals.
>
> The possibility the takedown of a principal by 1st elimination of an
> other somewhat dilutes the greater likelihood of CW being the 1st
> eliminated principal, & of Bus not being the 1st eliminated principal.
>
> Because of the burial-thwarting effect of having more than 1 Bus,
> then, with many candidates, the buriers have the problem of burying CW
> to make the strategic cycle, without creating more than 1 Bus.
>
> Could the buriers get lucky & succeed? Sure.
>
> But then you’re raising the philosophical question:
>
> If a strategy doesn’t have a positive result-change expectation, is it
> really a strategy?
>
> Michael Ossipoff
>
>
>
> On Sun, Oct 8, 2023 at 01:14 Michael Ossipoff <email9648742 at gmail.com>
> wrote:
>
> There’s good reason to choose Coombs as the “base-method”, the
> elimination-method to which to apply CT-.
>
> The use of Coombs maximizes the lowering of the base-method
> count-score for the buried CW. (We haven’t discussed fractional
> vs whole, but my first impression is that whole would best deter &
> thwart burial.)
>
>
> …making the CW more likely to get eliminated 1st. …thereby
> taking-down the buriers’ favorite (BF), & electing the less-liked
> candidate whom under whom the buriers buried CW.
>
> (Let’s call that candidate “Bus”, because the buriers are throwing
> CW under the bus.)
>
> That wasn’t the scenario in your example, but BF still lost.
>
> When there’s only one Bus, which of course is the case in all
> 3-candidate cases, then whether or not BF wins, & the burial
> succeeds, depends on who is eliminated 1st.
>
> Could there be a 1-Bus example in which Bus gets eliminated 1st,
> thereby taking down CW & electing BF? Yes, but, with Coombs as
> base-method, the burial of CW is particularly likely to give CW
> the worst base-method count-score. …& thereby the 1st
> elimination. …thereby taking down BF & electing Bus.
>
> So the burial is likely deterred, unless the buriers have very
> detailed & reliable predictive information.
>
> I’m not a fan of LNHa, but when speaking of advantages of STE
> (Successive Topcount Elimination), one of course mentions LNHa
> among them.
>
> But, though we need rank-balloting, & STE is a rank-method, &
> obviously people who enact STE don’t want or intend to bury their
> favorite under a disgusting lesser evil, & though I’d accept STE’s
> top-end defensive strategy-need in order for people to have rank
> balloting so that they won’t elect despicable POS lesser evils…
>
> …I’m not going to defend STE anymore…or vote for its enactment…or
> participate in its enactment campaign…
>
> …because we do such things for principle…
>
> …&, if principle means anything, we don’t condone lying & dishonesty.
>
> FairVote is still using the lie that STE don’t have a spoiler problem.
>
> …using that lie not only with voters, but also to get money from
> donors.
>
> I couldn’t bring myself to help that lie-based project or its
> proposal.
>
> Did you know that FairVote is claiming that it’s Condorcet that
> has the spoiler problem?
>
> I guess, to borrow from a famous statement, that proves that
> there’s no such thing as an unutterably blatant lie.
>
> So, as I said, I won’t defend STE or in any way support its
> enactment campaigns.
>
>
>
> On Sat, Oct 7, 2023 at 23:16 C.Benham <cbenham at adam.com.au> wrote:
>
>
> Michael,
>
>> ... there's no offensive strategy for changing the winner from the CW
>> to one's own favorite,
>
> Say sincere is
>
> 46 A
> 44 B
> 10 C
>
> A is the CW (as well as being every other type of winner).
>
> Now suppose the B voters all decide to bury against A by
> ranking C second:
>
> 46 A
> 44 B>C
> 10 C
>
> Now C has the lowest Borda score and the lowest "top count",
> so if we are using
> IRV/RCV or Baldwin C is eliminated, "taking down" A and
> leaving B elected.
>
> The "offensive strategy" has succeeded. This is the same
> outrageous failure of
> the Later-no-Help and Plurality criteria that we see with Margins.
>
> I'm not clear how Coombs (or your preferred version of it)
> handles truncation.
>
> Is the "bottom count" Fractional (in other words based on the
> symmetrically completed ballots)
> or Whole (so that truncated ballots give a whole vote to each
> of the bottom-counts of the truncated
> candidates)?
>
> In this case neither version eliminates C, but I'd be
> surprised if examples couldn't be made of
> them also failing those criteria.
>
> You implied that you are a fan of Later-no-Harm. If that is
> the case I don't think you will find a better
> method than plain IRV/RCV
>
> Chris B.
>
>> Here are 3 elimination-methods that eliminate 1 candidate at a time:
>>
>> …
>>
>> RCV/IRV: Eliminates lowest topcount
>>
>> …
>>
>> Coombs: Eliminates highest bottomcount
>>
>> …
>>
>> Baldwin; Eliminates lowest Borda-score.
>>
>> …
>>
>> Any one of those can be the “base-method” .
>
>
>
>
>
>> *Michael Ossipoff*email9648742 at gmail.com
>> <mailto:election-methods%40lists.electorama.com?Subject=Re%3A%20%5BEM%5D%20My%20CTE%20post%20to%20EM&In-Reply-To=%3CCAOKDY5CU-zc3Y%3DC%3DrFWagy3zVw5Px6KPaULqW89gHdxshRZh8A%40mail.gmail.com%3E>
>> /Fri Oct 6 14:33:40 PDT 2023/
>>
>>
>> ------------------------------------------------------------------------
>> When I said:
>>
>> "There’s no offensive strategy for changing the CW to one’s own favorite."
>>
>> I meant, there's no offensive strategy for changing the winner from the CW
>> to one's own favorite.
>>
>> On Fri, Oct 6, 2023 at 5:21 PM Michael Ossipoff <email9648742 at gmail.com <http://lists.electorama.com/listinfo.cgi/election-methods-electorama.com>>
>> wrote:
>>
>> >/Name of Method: />//>/… />//>/CW,Takedown-Elimination (CTE) />//>/… />//>/or />//>/… />//>/Simmons-Ossipoff Method />//>/… />//>/Okay yes, I like the 2nd one. />//>/… />//>/It adds an enhancement to any of several already-existing />/elimination-methods. />//>/… />//>/Here are 3 elimination-methods that eliminate 1 candidate at
>> a time: />//>/… />//>/RCV/IRV: Eliminates lowest topcount />//>/… />//>/Coombs: Eliminates highest bottomcount />//>/… />//>/Baldwin; Eliminates lowest Borda-score. />//>/… />//>/Any one of those can be the “base-method” . />//>/… />//>/Method rule: />//>/… />//>/Ranked balloting. Equal-rankng & truncation allowed. />//>/… />//>/1) Check for a CW & elect hir. />//>/… />//>/2) If none, do the base-method. />//>/… />//>/3) During the doing of the base-method: />//>/… />//>/When the base-method’s rule eliminates a candidate,
>> eliminate additionally />/anyone who is pair-beaten by that candidate. …& additionally
>> any />/candidate beaten by that 2nd candidate. />//>/… />//>/That’s takedown & secondary takedown. />//>/… />//>/4) If anyone becomes un-pairbeaten due to elimination of who
>> beats hir, />/s/he wins. />//>/… />//>/5) Continue till only one candidate remains uneliminated. />//>/… />//>/[end of count-rule definition] />//>/… />//>/Though it was Forest who introduced the unprecedentedly
>> gamechanging />/Takedown, & applied it to Coombs & Baldwin, the bombast in
>> this post is all />/mine. />//>/… />//>/Obviously a CW wins if voting is sincere. />//>/… />//>/There’s no offensive strategy for changing the CW to one’s
>> own favorite. />//>/… />//>/There’s no need for defensive strategy to protect the win of
>> a CW. />//>/… />//>/While MinMax(wv), Schulze, RP(wv) & Smith//MinMax(wv)
>> require defensive />/truncation to deter burial strategy against the CW, the
>> above-defined />/method requires no such defensive strategy, & voters can
>> rank all the way />/down to the bottom if they want to. />//>//>/PS. I added a statement that, for the purposes of takedown &
>> secondary />/takedown, "pair-beaten" should probably mean "pairbeaten
>> according to the />/rankings before any eliminations." />//>/… />//>/Michael Ossipoff/
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20231009/02ff8038/attachment-0001.htm>
More information about the Election-Methods
mailing list