[EM] Approval vs. IRV

[Juho Laatu]

On 27.11.2011, at 18.35, Kathy Dopp wrote:

> Nov 2011 23:05:49 -0700
>> From: matt welland <matt at kiatoa.com>
> 
>> I wasn't clear. I want to hear opinions from the list: Is approval
>> better or worse than IRV and why?
> 
> Approval is a far superior system to FPTP and IRV because approval:
> 
> 1. unlike FPTP and IRV, it solves the spoiler problem of a nonwinning
> candidate altering who would otherwise win;

True. In Approval nonwinning candidates cause no harm. But plausible winners may "spoil" the election.

> 
> 2. unlike IRV, it is precinct-summable and easy to manually
> statistically audit for accuracy and so preserves or allows for
> timely, understandable, election accuracy verification;

True. But it is also possible to record the votes locally in IRV and then send the collected data forward (and based on that data, check the votes later locally).

> 
> 3. unlike IRV, it looks at all the candidate choices of all voters,
> thus treating all voters' votes equally and fairly and is thus
> monotonic;

True. But Approval collects only very little information, and in that sense it does not look at many of the preferences of the voters.

> 
> 4. unlike IRV, it is simple to implement using the same ballot style
> as FPTP, with a fairly simple programming change in the tally program;

In the USA Approval can often be implemented with only small changes to the FPTP procedure. But that is a local and history dependent criterion (i.e. does not apply in some other countries).

> 
> 5. unlike IRV, it preserves the rights of voters to have their votes
> counted fairly and equally with other voters and to participate in the
> final counting rounds (final decision-making process);

The algorithm of IRV is quite weird, but fully ranked votes will participate also in the final rounds. Approval has its own problems. If it turns out that candidates A and B were the most popular candidates, voters who approved both of them or neither of them may feel that their vote was not counted when the final decision was made.

> 
> 6.  unlike IRV, It is very simple for voters to figure out how to best
> strategize (i.e. for single-winner elections, If your favorite
> candidate is one of the top-two most likely vote-getters, bullet vote.
> Otherwise, vote for both your favorite(s) and one of the likely
> top-two vote getters. In other words, the simple principle is that a
> vote for your 2nd choice candidate may cause the 2nd choice, rather
> than your 1st choice candidate to win, so vote for a 2nd, 3rd,...
> choice if you don't mind them winning the contest.)

In Approval finding the best strategy may be very difficult when the number of potential winners is three or higher (that is maybe my biggest concern in Approval). Also in IRV one may easily fail to find the best strategy (and the result of the election may get worse). In IRV sincere voting is however a quite reasonable way to vote. In Approval voters need to identify the best strategy to cast an efficient vote. Sometimes they may face a dilemma where they must either not take position in the key question or risk electing a bad candidate.

> 
> 7,  unlike IRV, it increases the chances of a popular 3rd party
> candidate winning rather than being a system for keeping the smaller
> party candidates from interfering with the 2 major parties (true once
> people figure out how to strategize with IRV by ranking one of the 2
> major party candidates 1st)

It is true that Approval can elect compromise candidates, while IRV clearly favours large parties. I consider methods that can elect compromise candidates to be better general purpose single winner methods. But the choice of method depends here very much on what kind of results the society in question wants to have. In the USA where the tradition is e.g. to elect single winners to form single winner governments, I can understand if some people prefer methods that favour large parties.

Also methods that aim at electing only from the strongest groupings have their place in the family of voting methods. (Such methods, that would also eliminate some of the problems of IRV, have been discussed recently on this mailing list.) I'm not sure if IRV was planned this way or if that property is accidental. But that property may well be one reason why the old parties are happier to support IRV than some of the other methods (i.e. to keep competition out, not necessarily to support the philosophy of electing only strong candidates).

Approval can handle small 3rd parties very well. In a (former) two-party system the problems of Approval are likely to emerge only when those small 3rd parties grow almost as strong as the smaller one of the two old strong parties.

> 
> In other words, approval voting is administratively, educationally,
> and technically simple and equitable, and thus doable - and unlike IRV
> actually makes substantial improvements over FPTP because it solves
> the spoiler problem and does not create nonmonotonicity and huge
> administrative, technical, auditability and other practical problems.

Both methods have some problems - quite different ones.

I'd like to add simplicity of filling the ballot (without errors) to the benefits of Approval.

Juho




> 
> 
> -- 
> 
> Kathy Dopp
> http://electionmathematics.org
> Town of Colonie, NY 12304
> "One of the best ways to keep any conversation civil is to support the
> discussion with true facts."
> "Renewable energy is homeland security."
> 
> Fundamentals of Verifiable Elections
> http://kathydopp.com/wordpress/?p=174
> 
> View some of my research on my SSRN Author page:
> http://ssrn.com/author=1451051
> ----
> Election-Methods mailing list - see http://electorama.com/em for list info