[EM] [CES #8791] Upper-Bucklin naming (was: Median systems, branding....)

Abd ul-Rahman Lomax abd at lomaxdesign.com
Sun Jun 16 18:46:07 PDT 2013

At 09:57 PM 6/15/2013, Jameson Quinn wrote:

>2013/6/15 Abd ul-Rahman Lomax 
><<mailto:abd at lomaxdesign.com>abd at lomaxdesign.com>
>At 07:52 PM 6/14/2013, Jameson Quinn wrote:
>So. Abd and I now agree that a Bucklin system which uses just the 
>above-median votes to break ties is probably the best first step 
>towards median voting.
>Let's stop saying it that way.
>I'd be happy to. What do you propose, in 8 words or less?

A few more words, probably, particularly since you used more.

"Bucklin" is a ranked approval system, where approvals are categories 
into ranks in order of preference. In a modern Bucklin system, voters 
may categorize as many candidates in each rank as they choose, may 
skip ranks, and candidates not voted for explicitly are considered 
not approved for election. Votes are amalgamated by canvassing the 
first rank, checking for a majority, and then proceeding to add in 
the next ranked votes, in sequence, until a majority is found or the 
ranks are exhausted.

This system can produce a multiple majority, and a concern when this 
occurs is that voters may ahve over-enthusiastically added additional 
approvals, not realizing that they were in the majority as to their 
higher preference. Fear of this can discourage adding additional 
approvals, and thus encourage majority failure. Hence, with this 
proposed Bucklin variant, if a multiple majority is found, below the 
first rank, the votes from that rank are removed from the totals and 
the win is awarded to the majority-approved candidate with the most 
votes in the previous-canvassed rank.

(If a majority is found in the first rank, to be explicit, the win 
goes to the candidate with the most votes.)

However, I'm not *entirely* on board this. It violates long-standing 
traditions about multiple majorities. I am willing to *consider* it, 
under the limitation of a deterministic method. I've suggested we 
need more data.

Both "ties" and "median" introduce concepts which are either complex 
or unfamiliar.

More information about the Election-Methods mailing list