[EM] Two round methods

[Kevin Venzke]

Hi Rob,

Le samedi 2 octobre 2021, 01:55:40 UTC−5, Rob Lanphier <roblan at gmail.com> a écrit :
> Hi Kevin!
> 
> As I'm sure you remember, I've been trying to figure out a system that
> could replace California's "top-two primary" for a while.  Your
> proposal could work, but I'm curious about the value it provides over
> other approaches.
> 
> More inline...
> 
> On Fri, Sep 24, 2021 at 5:59 PM Kevin Venzke <stepjak at yahoo.fr> wrote:
> > I see a need for non-eliminative two-round methods. I like the idea that the first round places
> > a "default winner" or "guy to beat" in the electorate's consciousness, so that, possibly, voters
> > would consider offering more of their preferences in the second round.
> >
> > A difficult thing is to make sure it is worthwhile to seriously compete in the first round, and
> > also, relatedly, that the second round should not simply reverse the outcome of the first round
> > based on small changes in participation rate.
> 
> It seems this is the reason for eliminating candidates in the first
> round.  If one is going to have approval voting in both rounds, then
> requiring a modest approval (e.g. 30% approval) to advance from the
> first round to the second round would be ample motivation, even if
> several candidates are allowed to advance.  I'm still fond of only
> allowing two candidates to advance, so that there's a clear choice
> presented to the voters in the second round.

My concern arises from seeing centrist candidates be unable to compete for a major party's
nomination. My hope is that if you present to the voters that a certain candidate is on
track to be elected, it will inspire them to support more choices (assuming they don't like
the projected winner).

So it is not "one on one" but "any and all vs. one."

I don't have an inherent problem with "one on one," and I don't remember what your complaint
was about the California method. I would just be concerned with what is the best way to pick
the finalists.

> 
> > Here's an idea based on two-round Approval. Let's call it NETRA.
> >
> > 1. If the first round winner ("FRW") has majority approval, he is elected. End.
> > 2. Otherwise hold the second round with all the same candidates. If this new winner has
> > majority approval *and* the FRW does *not* have majority approval, then the new winner is
> > elected. Otherwise the FRW wins.
> 
> I guess I'm too used to the American primary-election/general-election
> split,

Hmm? In the American system you can't win with a majority on the first "round" because only
half the political spectrum participates in a contest. Republican and Democratic are done
separately.

If it were possible to win a majority of the electorate during the primaries, you would
practically have to elect that person. Otherwise you have the potential that small changes
in turnout could reverse the result. That would be really bad.

> so I'm wondering what the value is for allowing a candidate to
> win with a simple majority in the first round.

[in my method]
1. It creates incentive to try to win. If there is no reward for winning the first round,
the first round could return an arbitrary winner when it is supposed to be a strong winner,
one who is likely to prevail in the second round.

2. I also think getting a majority of the voters to approve you is good enough to get the
win, especially if our only tool for conducting this race is approval balloting.

3. If a simple majority doesn't win the first round, then I'm not sure what the second round
is trying to do. My whole goal for the second round was to try to gather a majority when
there had not been one. If we already had one... Why isn't that the winner?

> It seems like a super
> majority (e.g. 60% or higher) should be required to short-circuit a
> second-round standoff with the first-round runner up.

Ok, but you're talking about a different method if you have a "first-round runner up"
concept.

I don't think you should use "top two" approval to pick finalists, because of the
theoretical possibility that the finalists got their support from the same group of voters.
It may not be very likely to happen, but in theory it's not sound, it doesn't accomplish
anything consistent.

> Otherwise, it
> seems likely to imagine something similar to following scenario with
> 1,000,000 votes:
> 
> A (incumbent) - 500,001 approvals
> B (serious challenger) - 499,995 approvals
> C (joke challenger) - 10 approvals
> 
> That's 6 voters in a 1,000,000 vote election who decide B doesn't get
> a head-to-head standoff with A because perhaps some of them got
> distracted by C's messaging.

Six voters? I don't follow. You presumably have at most 4 voters who didn't vote for A or B,
not enough to beat A. C can't even be called a spoiler here.

I'm at a loss as to an alternative: Hold a second round, where B might win by the exact same
bare majority that A already had? That seems awfully wasteful, and likely to be viewed as
arbitrary.

To my mind, the purpose of the second round is not to just give B a second shot at it. The
purpose is to see whether some preferences were left unexpressed due to not knowing who was
going to be in the lead. If some candidate can then reach a majority, that's excellent.

> Moreover, since this is approval voting
> we're talking about here, it's also possible for this outcome in a
> 1,000,000 ballot first-round election:
> 
> A (incumbent) - 500,100 approvals
> B (serious challenger) - 500,050 approvals
> C (joke challenger) - 10 approvals
> 
> Now, both A and B have slim majorities for their respective approval
> ratings.  Perhaps people approved "A" and "B" because "C" seemed like
> a real lunatic, and "A" chose to legitimize "C" and ignore "B", and
> scaremonger people who would have voted exclusively for "B" into
> voting for "A" as well.

Within my method:
1. I can view it as a reward for taking the first round seriously, that the larger majority
wins.
2. Again my only purpose for the second round was to try to find an elusive majority. Here
we already have one.

> It seems to me that it shouldn't be too hard to get candidates and
> voters to take the first round seriously.  

If you're picking finalists, then absolutely, the first round matters.

> A small enough
> supermajority (e.g. 60%) would seem achievable for a competent
> incumbent

I'm surprised you feel that way, but maybe it's a common situation in your area.

>  -- which would cause the rest of the field of candidates to
> motivate their base to go to the polls if one candidate getting a
> supermajority means  "no second round".
> 
> It also seems to me that perhaps the "short-circuit" threshold should
> be a large margin (e.g. a 10% lead over the next candidate)

I don't think the margin has much meaning on approval ballots. I don't totally hate it, but
I wouldn't apply it in cases with a majority.

> rather
> than an absolute value (like a 60% supermajority approval).  If two
> candidates get over 50% approval in the primary election, it seems
> like BOTH candidates should advance to the general election regardless
> of the margin, since (by definition) there's a contingent who felt
> that both candidates would be acceptable, but (to quote Sean Connery
> in Highlander) "there can be only one".  .  If three candidates get
> over 50% approval ... well, it sucks to be third place in a primary
> where the rules state "only two shall advance".  What seems most
> likely in a crowded first round is that no candidates get a majority
> 
> Now to reply to your other points:
> 
> > So in the second round the FRW is given a couple of perks. If he raises his vote to a majority,
> > or if (again) no one can get a majority, he doesn't have to beat anyone. He can prevail based
> > on his first victory.
> >
> > This has two purposes: It should be desirable to be the FRW. And the FRW's advantage
> > should increase the electorate's assessment of the FRW's odds, so that hopefully the second
> > round serves to tell us whether there is a candidate preferred by the voters to the FRW. (It
> > is an approval ballot in form, but what I'm actually after is the FRW's pairwise contests.)
> >
> > Main concerns:
> > 1. If candidates were to conclude that being the FRW is overall a bad situation. I'm not
> > sure how likely this is, given that a first-round majority is an auto-win.
> 
> This seems like an unlikely strategy.  In an approval election, there
> seems to be very little strategic reason to not try to get first
> place, even if a majority first-place doesn't seem likely.

The possibility here is that candidates find that being FRW is too powerful at uniting opposing
voters against them. Or that being FRW in the second round is an unfair position.

On paper the second round seems a little prejudicial against the FRW. We're checking if any
candidate can defeat the FRW, when the FRW is posited as a likely winner. But any such candidate
will not, in turn, be subjected to this same check. They just win. If such a check were to occur,
or if a different candidate were the FRW, we might find that the original FRW could actually get
a majority. But he has no opportunity to show this.

That's why I'm trying to grant any reasonable perks I can to the FRW, to balance this out.

> > 2. It's still quite possible that a majority opposing the FRW in the second round would fail
> > to agree to get behind any single candidate. Then a lack of an approval majority in the first
> > round would simply be repeated in the second, for no benefit. (There is, I suppose, the
> > additional confirmation, for all to see, that there is no majority willing to unite against the
> > FRW... It also confirms that the winner couldn't attract a majority.)
> 
> I think this is the benefit right here.  A tepid response in the
> second round would be embarrassing for the FRW, and make it incredibly
> difficult to claim a "mandate" on radical policy changes.  It would
> also send a message to the opposition that they need to unite in order
> to win.  Borrowing an old bit from the Monty Python movie "Life of
> Brian", the Judean People's Front and the People's Front of Judea need
> to unite against the Romans (so to speak):
> <https://youtu.be/WboggjN_G-4>

One hopes.

Kevin