[EM] Tideman data results

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

[EM] Tideman data results

John
[Not subscribed; please CC me on responses.]

Does anyone have Smith Set, Schwartz Set, IRV, and STV (2, 3, 4, 5, etc.) results for the HIL files here:

https://rangevoting.org/TidemanData.html

I get the following for A1 with Meek-STV (precision = 9, seats = 3):

Tabulation completion data:
  0.0000000000 Candidate  2 defeated
  156.7058824400000000000 Candidate  9 elected
  0.0000000000 Candidate  7 defeated
  135.9776274220000000000 Candidate  1 elected
  0.0000000000 Candidate  5 defeated
  0.0000000000 Candidate  6 defeated
  0.0000000000 Candidate  8 defeated
  131.1851762120000000000 Candidate  3 elected
  0.0000000000 Candidate  4 defeated
  0.0000000000 Candidate 10 defeated
  Notes:

The next step is to follow the debugger step by step and straighten out the logic (it gives an odd report with 2 candidates), and then clean up the remaining refactoring artifacts (the constructor for a tabulator has number of seats as an argument—so does the Tabulate() call).

Right now, the tabulator computes IRV (broken by vote splitting, not recommended), Tideman's Alternative (recommended), and Meek-STV (recommended).  It does not compute:

 - MNTV (non-representative, not recommended)
 - Approval (non-representative, inherently tactical and voter-risk-behavior based, not recommended)
 - Range voting methods (subject to an amplification of the Approval voting problem, but can be computed as ranked by flattening)
 - Plurality (broken by vote splitting, not recommended)
 - STV rules without surplus pass-through (broken by Woodall free riding, not recommended)
 - Schulze STV (resists Hylland free riding, recommended)


This is pre-alpha work.

Thanks in advance.
—John

----
Election-Methods mailing list - see https://electorama.com/em for list info
Reply | Threaded
Open this post in threaded view
|

Re: [EM] Tideman data results

John
Well then.


"smith set" is the smith set.

"schwartz set" is valid when the smith set and schwartz set differ.

When the smith or schwartz set are 1, a tabulator can test itself as "condorcet-smith" or "condorcet-schwartz"

Other valid types of results are "tideman's alternative" (if schwartz = 1, elect; else eliminate non-smith, runoff 1, repeat); "tideman's alternative smith" (elects by smith set = 1); "tideman's alternative schwartz" (schwartz set elimination); and "meek-stv".

This is an extremly basic outcome format.  The json format.. I can't parse yet.

On Mon, May 13, 2019 at 1:34 PM John <[hidden email]> wrote:
[Not subscribed; please CC me on responses.]

Does anyone have Smith Set, Schwartz Set, IRV, and STV (2, 3, 4, 5, etc.) results for the HIL files here:

https://rangevoting.org/TidemanData.html

I get the following for A1 with Meek-STV (precision = 9, seats = 3):

Tabulation completion data:
  0.0000000000 Candidate  2 defeated
  156.7058824400000000000 Candidate  9 elected
  0.0000000000 Candidate  7 defeated
  135.9776274220000000000 Candidate  1 elected
  0.0000000000 Candidate  5 defeated
  0.0000000000 Candidate  6 defeated
  0.0000000000 Candidate  8 defeated
  131.1851762120000000000 Candidate  3 elected
  0.0000000000 Candidate  4 defeated
  0.0000000000 Candidate 10 defeated
  Notes:

The next step is to follow the debugger step by step and straighten out the logic (it gives an odd report with 2 candidates), and then clean up the remaining refactoring artifacts (the constructor for a tabulator has number of seats as an argument—so does the Tabulate() call).

Right now, the tabulator computes IRV (broken by vote splitting, not recommended), Tideman's Alternative (recommended), and Meek-STV (recommended).  It does not compute:

 - MNTV (non-representative, not recommended)
 - Approval (non-representative, inherently tactical and voter-risk-behavior based, not recommended)
 - Range voting methods (subject to an amplification of the Approval voting problem, but can be computed as ranked by flattening)
 - Plurality (broken by vote splitting, not recommended)
 - STV rules without surplus pass-through (broken by Woodall free riding, not recommended)
 - Schulze STV (resists Hylland free riding, recommended)


This is pre-alpha work.

Thanks in advance.
—John

----
Election-Methods mailing list - see https://electorama.com/em for list info