[EM] Unranked-IRV

Tom Ruen tomruen at itascacg.com
Fri Mar 23 01:28:45 PST 2001


I would like to share perhaps the simplest majority-empowered reform over
our current election problems.

I'll call this system: Unranked-IRV. It works like IRV but doesn't allow
voters to rank preferences, only list them like approval, however unlike
approval votes are divided so each voter only gets one total vote among all
choices.

I've been interested recently in using approval voting to end the spoiler
effect of plurality elections, but I get objections to the approval process,
most strongly from Don Davison, that everyone only should get one vote, and
he's right that is how our current system works. For Don's sake, I
considered split-vote tied-ranks within IRV. Tied ranks in IRV is completely
reasonable and can be supported by dividing a vote by the number of
candidates in the tie.

Now from my analysis of my sample month election I discover somewhat
surprisingly that the IRV process can be run entirely on split-vote approval
ballots! I've known this for a while actually, but I didn't think about the
value of this.

The value is that we don't need those confusing rankings! Instant runoff
rounds have votes change between rounds merely from being split among less
candidates. As candidates are eliminated, support rises for those that
remain.

Well, is this reform enough to help us? How would this work in a real
election?

Imagine the 2000 U.S. presidential election (assuming national voting) had
these approval votes: (Just take this as a semirealistic simplified example,
using only 3 candidates.)
Bush    48%
Gore    41%
Gore&Nader    8%
Nader    3%

A vote for "Gore&Nader" means a voter wants to support both equally. Each
will get 1/2 of the support (half of 8%) until one or the other is
eliminated. Then the remaining candidate will get all of the 8%.

My imagined numbers for the Gore-Nader compromise vote might seem inflated
but remember they are from 2 sources:
1. (Nader-Gore) Nader supporters who fear he can't win and want to support
Gore as a compromise. (Up voters)
2. (Gore-Nader) Gore supporters who are upset with Gore's neglect of Nader's
issues and want to send a message. (Down voters)

These compromises are safe from both sides as long as elimination order is
predictable. Polls will do a good job to allow voters to evaluate the risk.

These approval ballots would become in an IRV process: (split-votes instead
of ranks)

Round 1: Bush 48%, Gore 41+8/2=45%, Nader=3+8/2=7%.

Then Nader would be eliminated as weakest, and the ballots would be
recounted w/o Nader:
Bush    48%
Gore    41%
Gore     8%
NOTA   3%

Round 2: Bush 48%, Gore 49%, NOTA 3%

Well, this doesn't look very different from what happened (assuming a
national vote was counted), right?

However there are 2 fundamental differences:

1. Nader's approval-based support can be correctly determined. (By adding
total votes without dividing splits.)

The approval is: Gore 49%, Bush 48%, Nader 11%. All 3 exceed the 5% needed
for federal funding for parties. (Perhaps even Buchanan would get some
compromise support from Bush supporters and keep the sorry Reform party
rolling!)

2. If Gore loses, it will be because Nader supporters didn't approve of
Gore, not because the system stopped them from supporting him.

A progressive majority in a single vote system only REALLY exists if it is
fully united behind at least one progressive candidate. Otherwise the
majority is an illusion. Unity will always be behind the stronger (more
established) candidate and party. When the stronger faction is in doubt in
polls, the more well established candidate must take seriously the issues of
the newer faction rising in support. Until that strength comes, voters for a
weaker candidate can comfortable vote their bigger progressive compromise
too if they want to influence the election AND support a favorite.

Split-vote approval with instant runoff is simple for voters and avoids all
the controversy over evaluating ranked ballots. Voters are KING in Approval
voting while ALL ranked methods make special assumptions on how voters want
their ranks used. Ranked ballot methods might offer more choice to voters,
but Unranked-IRV seems to do an awful lot with very little.

Ballot design would not change at all, only the counting. Voters may then
consider support more than one favorite. There will no longer be spoiled
"overvotes".

I judge that Unranked-IRV is the simplest reform (closest to existing
system) than can handle the difficulties in our current elections. Spoilers
are optional by voter preference in their willingness to compromise and no
one needs to vote insincerely - we can always vote for ALL of their
favorites as a best defense.

Of course, we can later add rankings as a bigger reform and use
IRV-w/split-vote-tied-ranks or Condorcet-w/tied-ranks later too when we are
ready for a bigger step.

Does this sound like worthy reform? If it can more often produce a majority
winner in a single election, then I say the answer is a clear YES!

Plurality or Unranked-IRV, which method will you choose for unranked
ballots?

Tom Ruen




More information about the Election-Methods mailing list