stv-discuss Mailing List for OpenSTV -- STV software and IRV software
Brought to you by:
jco8
Menu
▾
▴
stv-discuss — General purpose list for OpenSTV
You can subscribe to this list here.
| 2004 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
(1) |
Sep
(10) |
Oct
|
Nov
(6) |
Dec
(1) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2005 |
Jan
(6) |
Feb
(15) |
Mar
|
Apr
(8) |
May
|
Jun
(1) |
Jul
|
Aug
(2) |
Sep
(1) |
Oct
(1) |
Nov
|
Dec
|
| 2006 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
(1) |
Jul
|
Aug
(8) |
Sep
|
Oct
(2) |
Nov
|
Dec
|
| 2007 |
Jan
|
Feb
|
Mar
(1) |
Apr
|
May
(6) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|
From: Jonathan L. <jlu...@po...> - 2007-05-28 16:54:16
|
On May 28, 2007, at 1:54 PM, Jeffrey O'Neill wrote: > The Hare system or (Hare-Clark system) is sometimes meant to refer to > STV in general and not with regards to a Hare quota. I think that is > what Craig meant. In that case, my clarification was unnecessary (though still the case wrt OpenSTV). Craig had asked, > Curious why Candidate Bob wins even though he doesn't make the > threshold (not enough surplus votes). > If the Hare system assumes the seat is filled despite not meeting > the threshold one might want the option to disable this. We could certainly universalize that approach as an option; it's used by the GPCA2000 rules now. I'm not entirely convinced that it's desirable (even though we use it for GPCA elections), but then I'm not convinced that many of the OpenSTV options are desirable. GPCA uses a mandatory threshold as a kind of substitute, within STV, for NOTA. The idea is that we'd prefer to leave a seat open rather than have a marginal candidate elected simply because we have an uncompetitive election. But the dynamics of this approach are open to question. I counted an election this weekend (79 voters, 37 seats, 37 candidates) that excluded two candidates because of the mandatory threshold, even though those two candidates had relatively high approvals. Perhaps a better method would be to impose a minimum approval, and exclude candidates not meeting the approval threshold before doing the STV count. That kind of thing no doubt messes IIA, though by way of compensation it tends (slightly) to avoid premature exclusion. I'd like to read (and write, as a last resort) a paper on the subject of adapting NOTA and recall to STV, as well as preserving proportionality when filling vacancies. I have a few ideas, but not enough to offer a general solution. > >> But a mandatory Hare threshold is going to lead to trouble. If any >> ballot >> gets exhausted, you're guaranteed to leave at least one seat >> empty. And if >> you prohibit ballot truncation (another bad idea, IMO), there's no >> need for >> a static threshold, since ballots will never be exhausted. >> >> One last note. You mention (above) "the Hare system". In the >> method you >> used, the algorithm is Meek's. "Hare" specifies only that a >> threshold of >> votes/seats will be used (as opposed to a Droop threshold of >> votes/(seats+1)); it implies nothing more about the counting method. |
|
From: Jeffrey O'N. <jc...@co...> - 2007-05-28 13:55:20
|
The Hare system or (Hare-Clark system) is sometimes meant to refer to STV in general and not with regards to a Hare quota. I think that is what Craig meant. > But a mandatory Hare threshold is going to lead to trouble. If any ballot > gets exhausted, you're guaranteed to leave at least one seat empty. And if > you prohibit ballot truncation (another bad idea, IMO), there's no need for > a static threshold, since ballots will never be exhausted. > > One last note. You mention (above) "the Hare system". In the method you > used, the algorithm is Meek's. "Hare" specifies only that a threshold of > votes/seats will be used (as opposed to a Droop threshold of > votes/(seats+1)); it implies nothing more about the counting method. |
|
From: Jonathan L. <jlu...@po...> - 2007-05-28 12:29:05
|
On May 26, 2007, at 10:59 AM, Craig Seeman wrote: > Sending this to the list again. > > Curious why Candidate Bob wins even though he doesn't make the > threshold (not enough surplus votes). > If the Hare system assumes the seat is filled despite not meeting > the threshold one might want the option to disable this. > > I can imagine this situation with Droop Static where "bullet" > voting results in a high number of exhausted ballots although I > haven't attempted to simulate that yet. Meek's method (like most STV methods, I think) will fill all seats if there are enough candidates, regardless of the threshold. If you're going to use Meek (which is a fine idea), you should by all means specify a dynamic Droop threshold; otherwise it's not really Meek's method. The only OpenSTV STV method that enforces a threshold is GPCA2000, which I added for our GP California rules, but it's based on a variation of the BC rules, not Meek. My Perl-based counter <http:// lobitos.net/voting> will do a Meek count with a mandatory threshold. You can run it online if you like, and it will accept a .blt file with a couple of lines added to specify the election method. WRT your question, you may be confused by OpenSTV's distinction between a static and dynamic threshold; I know I was. "Static" prevents the threshold from being adjusted downward as ballots are exhausted, but it can still elect with less that a full quota of votes. However, you can use the results below and enforce the threshold manually, simply by not electing the third seat (in this election). But a mandatory Hare threshold is going to lead to trouble. If any ballot gets exhausted, you're guaranteed to leave at least one seat empty. And if you prohibit ballot truncation (another bad idea, IMO), there's no need for a static threshold, since ballots will never be exhausted. One last note. You mention (above) "the Hare system". In the method you used, the algorithm is Meek's. "Hare" specifies only that a threshold of votes/seats will be used (as opposed to a Droop threshold of votes/(seats+1)); it implies nothing more about the counting method. > > Craig Seeman > Sec > GPNYS > > At 4:57 PM -0400 5/23/07, Jeffrey O'Neill wrote: >> Hi Craig, >> >> I'm forwarding your question to the STV-discuss list. You will >> hear from me or someone else soon. >> >> Jeff >> >> ---------- Forwarded message ---------- >> From: Craig Seeman < sec...@op...> >> Date: May 23, 2007 4:47 PM >> Subject: Bug with threshold? >> To: Jeffrey O'Neill <jc...@co...> >> >> Hi Jeff, >> >> If the threshold is 33 (using Hare Static threshold) then how is >> Bob elected in the final round with 25.1? Is the program is >> filling the third seat with the 3rd highest finisher even if not >> crossing the threshold maybe it shouldn't. >> >> >> Craig Seeman >> Sec. >> GPNYS >> >> >> Election: Hare >> Method: Meek STV >> Number of Ballots: 99 >> Threshold Name: Hare Static Fractional >> Threshold Value: 33.0 >> 8 candidates running for 3 seats. >> >> R|Dari|Roge| Ian| Bob|Tony|AnnL|AnnE|Davi|Exha|S urp >> --+----+----+----+----+----+-- --+----+----+----+---- >> 1|62.0|13.0|15.0| 9.0| 0.0| 0.0| 0.0| 0.0| 0.0|29.0 >> 2|33.0|18.6|20.1|22.1| 0.9| 0.0| 0.0| 0.0| 4.2| 0.0 >> 3|36.2| |33.0|25.1| | | | | 4.7| 3.2 >> >> Round 1: Count of first place rankings. >> Candidate Darin is elected. >> Round 2: Transferring surplus votes. >> Round 3: Eliminating candidate Roger. >> Eliminating candidate Tony. >> Eliminating candidate AnnL. >> Eliminating candidate AnnE. >> Eliminating candidate David. >> Candidate Ian is elected. >> Candidate Bob is elected. >> Winners are Darin, Ian, Bob. > ---------------------------------------------------------------------- > --- > This SF.net email is sponsored by DB2 Express > Download DB2 Express C - the FREE version of DB2 express and take > control of your XML. No limits. Just data. Click to get it now. > http://sourceforge.net/powerbar/db2/ > _______________________________________________ > STV-discuss mailing list > STV...@li... > https://lists.sourceforge.net/lists/listinfo/stv-discuss |
|
From: Craig S. <sec...@op...> - 2007-05-26 11:00:11
|
Sending this to the list again. Curious why Candidate Bob wins even though he doesn't make the threshold (not enough surplus votes). If the Hare system assumes the seat is filled despite not meeting the threshold one might want the option to disable this. I can imagine this situation with Droop Static where "bullet" voting results in a high number of exhausted ballots although I haven't attempted to simulate that yet. Craig Seeman Sec GPNYS At 4:57 PM -0400 5/23/07, Jeffrey O'Neill wrote: >Hi Craig, > >I'm forwarding your question to the STV-discuss list. You will hear >from me or someone else soon. > >Jeff > >---------- Forwarded message ---------- >From: Craig Seeman <<mailto:sec...@op...> >sec...@op...> >Date: May 23, 2007 4:47 PM >Subject: Bug with threshold? >To: Jeffrey O'Neill <<mailto:jc...@co...>jc...@co...> > >Hi Jeff, > >If the threshold is 33 (using Hare Static threshold) then how is Bob >elected in the final round with 25.1? Is the program is filling the >third seat with the 3rd highest finisher even if not crossing the >threshold maybe it shouldn't. > > >Craig Seeman >Sec. >GPNYS > > >Election: Hare >Method: Meek STV >Number of Ballots: 99 >Threshold Name: Hare Static Fractional >Threshold Value: 33.0 >8 candidates running for 3 seats. > > R|Dari|Roge| Ian| Bob|Tony|AnnL|AnnE|Davi|Exha|S urp >--+----+----+----+----+----+-- --+----+----+----+---- > 1|62.0|13.0|15.0| 9.0| 0.0| 0.0| 0.0| 0.0| 0.0|29.0 > 2|33.0|18.6|20.1|22.1| 0.9| 0.0| 0.0| 0.0| 4.2| 0.0 > 3|36.2| |33.0|25.1| | | | | 4.7| 3.2 > >Round 1: Count of first place rankings. > Candidate Darin is elected. >Round 2: Transferring surplus votes. >Round 3: Eliminating candidate Roger. > Eliminating candidate Tony. > Eliminating candidate AnnL. > Eliminating candidate AnnE. > Eliminating candidate David. > Candidate Ian is elected. > Candidate Bob is elected. > >Winners are Darin, Ian, Bob. |
|
From: Jeffrey O'N. <jc...@co...> - 2007-05-23 15:30:53
|
---------- Forwarded message ---------- From: Craig Seeman <sec...@op...> Date: May 23, 2007 5:11 PM Subject: Re: OpenSTV - weights, eliminations? To: Jeffrey O'Neill <jc...@co...> At 1:10 PM -0400 5/19/07, Jeffrey O'Neill wrote: >Hi Craig, > > >>I thought that might be the case reading through the various >>descriptions of transfers and ties so I guess it might be good to add >>that on the readout. Maybe "Tie broken by look back" or "Tie broken >>by previous round total." > >Actually, that should happen. I'm not sure why it didn't here. Bug maybe? See email I just sent preceding this on another question/bug. > >>Also related to above is another form of voting I don't know the name >>of. Example, there are 6 ballots cast for multiple candidates >>(keeping things over simplified). Candidate A is named 1st on 3 >>ballots and not at all on 3 others. Candidate B is named 2nd on 6 >>ballots (has no 1st place votes and no ranking other than 2nd). >>Candidate B, with no 1st place votes, would win with wider support >>than Candidate A. > >This sounds similar to Condorcet or approval, both of which are >included in OpenSTV. Condorcet would be it for a single seat election. Slipped my mind. There are also Borda variations you might consider adding in which weights are not sequential. Baseball Writers Association uses 10, 7, 6, 5, . . . Some might want to use 5, 3, 1 or 4, 2, 1 We had used Borda to elect County officers in one county since people liked the idea of the lower rankings counting rather than being dependent on a transfer. We also used a modified form of Approval voting recently for seating people on various committees. People were able to case a Yes, blank (abstain) or No vote for any number of candidates regardless of seats. In the first round candidates with more No votes than Yes votes were eliminated. In the second round only Yes votes were ranked. It was done this so people could voice who they did not want to be seated on a committee. People had mixed feelings about this method though so that will change. Example Round 1 A 31-32 B 30-25 C 20-2 Round 2 A eliminated B 30 is seated C 20 is seated We've been doing a lot of experimentation with various voting system lately. BTW some of us have formed a working group whose goal is to use NYC's City Charter "quasi" initiative and referendum system to try to bring back some form of PR to the City Council. We've been in contact with Rob Richie at Fair Vote. Craig Seeman Secretary GPNYS > >best, >Jeff |
|
From: Jeffrey O'N. <jc...@co...> - 2007-05-23 13:57:17
|
Hi Craig,
I'm forwarding your question to the STV-discuss list. You will hear from me
or someone else soon.
Jeff
---------- Forwarded message ----------
From: Craig Seeman <sec...@op...>
Date: May 23, 2007 4:47 PM
Subject: Bug with threshold?
To: Jeffrey O'Neill <jc...@co...>
Hi Jeff,
If the threshold is 33 (using Hare Static threshold) then how is Bob elected
in the final round with 25.1? Is the program is filling the third seat with
the 3rd highest finisher even if not crossing the threshold maybe it
shouldn't.
Craig Seeman
Sec.
GPNYS
Election: Hare
Method: Meek STV
Number of Ballots: 99
Threshold Name: Hare Static Fractional
Threshold Value: 33.0
8 candidates running for 3 seats.
R|Dari|Roge| Ian| Bob|Tony|AnnL|AnnE|Davi|Exha|Surp
--+----+----+----+----+----+----+----+----+----+----
1|62.0|13.0|15.0| 9.0| 0.0| 0.0| 0.0| 0.0| 0.0|29.0
2|33.0|18.6|20.1|22.1| 0.9| 0.0| 0.0| 0.0| 4.2| 0.0
3|36.2| |33.0|25.1| | | | | 4.7| 3.2
Round 1: Count of first place rankings.
Candidate Darin is elected.
Round 2: Transferring surplus votes.
Round 3: Eliminating candidate Roger.
Eliminating candidate Tony.
Eliminating candidate AnnL.
Eliminating candidate AnnE.
Eliminating candidate David.
Candidate Ian is elected.
Candidate Bob is elected.
Winners are Darin, Ian, Bob.
Same ballot ran with Droop by comparison
Election: Droop
Method: Meek STV
Number of Ballots: 99
Threshold Name: Droop Static Fractional
Threshold Value: 24.8
8 candidates running for 3 seats.
R|Dari|Roge| Ian| Bob|Tony|AnnL|AnnE|Davi|Exha|Surp
--+----+----+----+----+----+----+----+----+----+----
1|62.0|13.0|15.0| 9.0| 0.0| 0.0| 0.0| 0.0| 0.0|37.2
2|24.8|20.2|21.6|25.8| 1.2| 0.0| 0.0| 0.0| 5.4| 1.1
3|24.9|21.1|22.2|24.8| | | | | 6.0| 0.1
4|27.2| |37.3|27.7| | | | | 6.9|17.9
Round 1: Count of first place rankings.
Candidate Darin is elected.
Round 2: Transferring surplus votes.
Candidate Bob is elected.
Round 3: Transferring surplus votes.
Eliminating candidate Tony.
Eliminating candidate AnnL.
Eliminating candidate AnnE.
Eliminating candidate David.
Round 4: Transferring surplus votes.
Eliminating candidate Roger.
Candidate Ian is elected.
Winners are Darin, Ian, Bob.
|
|
From: Jeffrey O'N. <jc...@co...> - 2007-03-19 19:30:20
|
Announcing OpenSTV version 1.1 -- software for computing the single transferable vote, instant runoff voting, and many other voting systems. More details and download links are at http://stv.sourceforge.net. This release includes three new methods: (1) Scottish STV -- Rules that will be used in Scotland for the first time in 2007. (2) Green Party of California STV (3) Approval Voting We would greatly appreciate assistance in further developing this software. We are looking for bug reports, feedback on usability, Python programmers (or people willing to learn), and web developers (something like Drupal or Xoops). Please contact me if you are willing to help. best regards, Jeff O'Neill |
|
From: Jeffrey O'N. <jc...@co...> - 2006-10-31 19:27:07
|
Hello, I just realized that I never announced to this list that OpenSTV 1.0 was released on October 10th. If you haven't heard already, then please go to stv.sf.net to download the latest version. The sourceforge site keeps track of project statistics and October has been by far the best month since the start of the project: Downloads: 170 (previous high was 135 in April 2005) Total Web Hits: 1396 (previous high was 991 in October 2005) Rank: 1411 (previous best was 2101 in October 2005) best, Jeff O'Neill |
|
From: Jeffrey O'N. <jc...@co...> - 2006-10-10 16:06:43
|
Announcing OpenSTV version 1.0 -- software for computing the single transferable vote, instant runoff voting, and many other voting systems. Previous releases of this software were called pSTV, but the new name better reflects the purpose of the software, which is entirely open source. More details and download links are at http://stv.sourceforge.net. In addition to the new name, we have a new programmer. Jonathan Lundell has joined the project and among other contributions has created an installer for Mac OSX. Any assistance would be greatly appreciated. We are looking for (1) bug reports and feedback on usability, (2) Python programmers or people who would be willing to learn, and (3) web developers (something like Drupal or Xoops). Please contact me if you are willing to help. Jeff O'Neill |
|
From: <Dga...@ao...> - 2006-08-28 05:22:26
|
In a message dated 28/08/2006 01:04:50 GMT Standard Time, Matthew Clarke writes: Hi Jeff, I'm thinking of using your program for counting an STV-IRV election. On some test data, I've noticed that often multiple candidates are eliminated in a single round. My understanding of the IRV system (hand counting) leads me to believe that only 1 candidate is eliminated per round. Of course I am happy with the results of the elections, it is merely this procedural step that I'm curious about. Scrutineers would obviously ask me this question upon the seeing the result output, so I would like to have an answer! Hello Matthew Just to provide a simple example. Suppose in an IRV election for a single seat the following ballots are cast: 49: A 27: B 13: C>A 7: D>C>B 4: E>D>C>B The number of votes with a first preference for C,D and E is 24 less than the total number of votes with a first preference for B (27). Therefore C,D and E cannot win and can safely be eliminated as a group at the second stage of the count. Multiple eliminations of candidates who cannot win is a method of speeding up the hand counting of an IRV/STV election. David Gamble |
|
From: Jonathan L. <jlu...@po...> - 2006-08-26 19:12:44
|
At 7:41 PM -0500 8/26/06, Jeffrey O'Neill wrote: >Dear Matthew, > >I´m on vacation so I´m forwarding your email to the OpenSTV discussion >list. I´m sure you will get a response soon. > >best, >Jeff > >---------- Forwarded message ---------- >From: Matthew Clarke -- GPC - PVC <mc...@gr...> >Date: Aug 23, 2006 7:37 PM >Subject: IRV Question >To: jc...@co... > > >Hi Jeff, > >I'm thinking of using your program for counting an STV-IRV election. > >On some test data, I've noticed that often multiple candidates are >eliminated in a single round. My understanding of the IRV system >(hand counting) leads me to believe that only 1 candidate is >eliminated per round. > >Of course I am happy with the results of the elections, it is merely >this procedural step that I'm curious about. > >Scrutineers would obviously ask me this question upon the seeing the >result output, so I would like to have an answer! Hi, Matthew. and greetings from the Green Party of California. In brief, the behavior you're seeing is an optimization. The general idea is that we can eliminate the bottom n candidates if the total of their votes is less than the votes for the remaining candidate with the smallest vote count. The reason is that we know a priori that if any of the n candidates to be eliminated got *all* the votes from the group of n, she'd still be eliminated in the next round because she'd still have the lowest number of votes. The limiting case of this optimization is that at any point if the top candidate has an absolute majority, we're done. I'm not sure that it's a significant optimization in the single-winner (IRV) case, but it's presumably worthwhile in multiple-seat STV elections that are computationally expensive to count. >-- >Matthew Clarke > >National Office Manager/Directeur de Bureau >Green Party of Canada/Parti Vert du Canada >PO/CP 997 Station B Ottawa ON K1P 5R1 >(direct) 613-562-3558 (toll-free/sans-frais) 1-866-868-3447 (fax) >613-482-4632 >mc...@gr... - mc...@pa... >www.partivert.ca - www.greenparty.ca -- /Jonathan Lundell. |
|
From: Jeffrey O'N. <jc...@co...> - 2006-08-26 17:41:20
|
Dear Matthew, I´m on vacation so I´m forwarding your email to the OpenSTV discussion list. I´m sure you will get a response soon. best, Jeff ---------- Forwarded message ---------- From: Matthew Clarke -- GPC - PVC <mc...@gr...> Date: Aug 23, 2006 7:37 PM Subject: IRV Question To: jc...@co... Hi Jeff, I'm thinking of using your program for counting an STV-IRV election. On some test data, I've noticed that often multiple candidates are eliminated in a single round. My understanding of the IRV system (hand counting) leads me to believe that only 1 candidate is eliminated per round. Of course I am happy with the results of the elections, it is merely this procedural step that I'm curious about. Scrutineers would obviously ask me this question upon the seeing the result output, so I would like to have an answer! -- Matthew Clarke National Office Manager/Directeur de Bureau Green Party of Canada/Parti Vert du Canada PO/CP 997 Station B Ottawa ON K1P 5R1 (direct) 613-562-3558 (toll-free/sans-frais) 1-866-868-3447 (fax) 613-482-4632 mc...@gr... - mc...@pa... www.partivert.ca - www.greenparty.ca |
|
From: Jonathan L. <jlu...@po...> - 2006-08-19 11:33:06
|
At 2:24 PM -0400 8/18/06, Jay Batson wrote: >BTW -- I'd already made our counting method choice before I received >your email, and ended up choosing BC (vs. Meek), mostly because I >didn't know any better. Too late now to change it. But at least >there's *something* specified. :-) Thanks. That's the important thing. I do have one gripe about BC, though, especially for small elections, and that's that it rounds up the quota. There's no real excuse for that in a system with fractional surplus transfers. -- /Jonathan Lundell. |
|
From: Jay B. <bat...@si...> - 2006-08-18 11:24:44
|
Thanks for the reply. I may end up handling this slightly differently than originally expected, so I may not need to make this change. (I'm going to add some PHP-style stuff to *collect* votes on our website, try to collect them *all* there (even if we, the admins, enter hand-supplied ballots using a back end interface), then have the php code output a file that contains the resulting votes. I can write this code to put the information out in a format OpenSTV can already consume. If this ends up being the direction I take, I'll not need to push anything back. But if we do end up doing the work, I'll be happy to contribute it. Thanks for your replies. Cheers -jb BTW -- I'd already made our counting method choice before I received your email, and ended up choosing BC (vs. Meek), mostly because I didn't know any better. Too late now to change it. But at least there's *something* specified. :-) Thanks. On Aug 16, 2006, at 10:49 AM, Jonathan Lundell wrote: > At 8:54 AM -0400 8/16/06, Jeffrey O'Neill wrote: >> I bet excel has an option to export csv files with other delimeters. >> I only have OpenOffice and I think it does this (at least it reads >> files with other delimiters). > > Excel will export with tab or space delimiters (in the latter case, > variable-length strings of spaces for formatting purposes). > > At 1:17 AM -0400 8/16/06, Jay Batson wrote: >> There's a second problem I'd like to see solved, too: We'll only >> have around 50 or so votes (for 5 seats), and historically we've >> recorded the name of the voting member alongside their votes. If >> we put this info into the spreadsheet, it would introduce info >> into the input file that OpenSTV would barf on. > > It'd be nice for OpenSTV to trim comments (by some definition of > comment) from input lines, I agree. My Perl counter preprocesses a > ballot line with something like ^\s*(.*)(#.*)? (in principle, with > suitable greed adjustments, saving \1, and then silently discarding > the line if the result is empty. > > However, it's easy to deal with this in Excel; I can think of two > methods. > > 1. Select the column of names, clear it, save as tab-separated > text, and then undo the clear. > > 2. Select the rankings without the names, paste into another sheet, > and then save the second sheet as tab-separated text. Text-format > saves only a single sheet. > >> The most useful scenario for us would be to use a voting process >> we'll collect ballots, ballot-counters will sit and enter the >> ballot choices into an Excel spreadsheet, go through them a second >> time to confirm entries, and then feed the resulting file into >> OpenSTV. > > I've been through processes like this. It works fairly well, but > make sure that a different person is looking at the paper ballots > for the check than for the count. If you have the volunteers, > consider entering the election twice, with independent counters, > and if the two results don't agree, compare them (to make that > easier, number the ballots before counting to help ensure that > they're entered in the same order). > > At 1:17 AM -0400 8/16/06, Jay Batson wrote: >> I'm hoping to be using OpenSTV in an election in September for the >> Directors of a non-profit. > > Free (as in beer) advice: if you haven't done this already, make > sure that the group has agreed on a specific counting method before > the election. I helped count a Green Party election once in which > the BC and Meek results were different, and an argument over the > results lasted more than a year. > > It's less important which method you choose than that you choose > one in advance. In my view (and I think Jeff's), Meek is always > preferable, unless you want to count by hand. BC is hard to count > by hand when you have more than two seats to fill, and I'd strongly > resist any of the other hand-count-friendly methods (especially > with only 50 ballots). > > > -- > /Jonathan Lundell. ---------- Jay Batson Acting Managing Director, Chairman bat...@si... +1-978-824-0111 |
|
From: Jonathan L. <jlu...@po...> - 2006-08-16 07:49:36
|
At 8:54 AM -0400 8/16/06, Jeffrey O'Neill wrote: >I bet excel has an option to export csv files with other delimeters. >I only have OpenOffice and I think it does this (at least it reads >files with other delimiters). Excel will export with tab or space delimiters (in the latter case, variable-length strings of spaces for formatting purposes). At 1:17 AM -0400 8/16/06, Jay Batson wrote: >There's a second problem I'd like to see solved, too: We'll only >have around 50 or so votes (for 5 seats), and historically we've >recorded the name of the voting member alongside their votes. If we >put this info into the spreadsheet, it would introduce info into the >input file that OpenSTV would barf on. It'd be nice for OpenSTV to trim comments (by some definition of comment) from input lines, I agree. My Perl counter preprocesses a ballot line with something like ^\s*(.*)(#.*)? (in principle, with suitable greed adjustments, saving \1, and then silently discarding the line if the result is empty. However, it's easy to deal with this in Excel; I can think of two methods. 1. Select the column of names, clear it, save as tab-separated text, and then undo the clear. 2. Select the rankings without the names, paste into another sheet, and then save the second sheet as tab-separated text. Text-format saves only a single sheet. >The most useful scenario for us would be to use a voting process >we'll collect ballots, ballot-counters will sit and enter the ballot >choices into an Excel spreadsheet, go through them a second time to >confirm entries, and then feed the resulting file into OpenSTV. I've been through processes like this. It works fairly well, but make sure that a different person is looking at the paper ballots for the check than for the count. If you have the volunteers, consider entering the election twice, with independent counters, and if the two results don't agree, compare them (to make that easier, number the ballots before counting to help ensure that they're entered in the same order). At 1:17 AM -0400 8/16/06, Jay Batson wrote: >I'm hoping to be using OpenSTV in an election in September for the >Directors of a non-profit. Free (as in beer) advice: if you haven't done this already, make sure that the group has agreed on a specific counting method before the election. I helped count a Green Party election once in which the BC and Meek results were different, and an argument over the results lasted more than a year. It's less important which method you choose than that you choose one in advance. In my view (and I think Jeff's), Meek is always preferable, unless you want to count by hand. BC is hard to count by hand when you have more than two seats to fill, and I'd strongly resist any of the other hand-count-friendly methods (especially with only 50 ballots). -- /Jonathan Lundell. |
|
From: Jeffrey O'N. <jc...@co...> - 2006-08-16 05:54:26
|
Dear Jay, No this is not an oft-asked question, and I'm glad to help as much as I can. Before I respond in detail to what you wrote, have you considered entering the ballots through the pSTV (which will be named OpenSTV in the next release) program? I think this would be significantly easier than entering the ballots in Excel (try the auto rank feature). It would be straightforward to have a second person check the ballots with pSTV too. If the current ballot entry dialog does not meet your needs, perhaps your programmers could instead adapt pSTV's ballot entry code to suit your needs rather than convert an excel file? I am willing to provide assistance, but I am mostly unavailable until October. > Of course, you can see the obvious first problem: Excel could output > a "Raw" file, but lines would be comma-separated, not space- > separated. Now, clever use of Unix 'tr' could solve the problem, but > this requires we have a Unix dweeb around. I bet excel has an option to export csv files with other delimeters. I only have OpenOffice and I think it does this (at least it reads files with other delimiters). > There's a second problem I'd like to see solved, too: We'll only > have around 50 or so votes (for 5 seats), and historically we've > recorded the name of the voting member alongside their votes. If we > put this info into the spreadsheet, it would introduce info into the > input file that OpenSTV would barf on. Yes, it would barf, and I agree this would be useful information to have in the ballot file. > 1) Do you foresee any significant difficulties with modifying the > ballot-file-reading code to read a source ballot file that use comma- > separated entries on lines (vs. spaces)? This would be the code for reading raw files. I don't see any problems as long as the candidate names are alphanumeric. > 2) Do you foresee any significant difficulties with modifying the > ballot-file-reading code to ignore everything on a line after the > standard Python comment indicator is encountered (#) (so we could, > e.g. put a #member-name into the last column on a line in Excel)? I see no problems. > 3) Assuming I have these changes implemented, is this code you are > interested in enough that you'd like us to feed the changes back into > the open source code base? Possibly. Generally, I like to limit the number of available ballot formats to avoid confusion. The next ballot format I would like to add would be an xml one, but I have yet to see a good one. If your code is very useful, easy to use, and would appeal to a wide audience, then I would want to add it to OpenSTV. I hope this helps, Jeff O'Neill |
|
From: Jay B. <bat...@si...> - 2006-08-15 22:17:56
|
Hi -- I'm just now getting started with this, so forgive me if this is an oft-asked question. (I did a quick archive search, unsuccessfully, before posting this.) I'm hoping to be using OpenSTV in an election in September for the Directors of a non-profit. I'd like to create a ballot file (either type) that OpenSTV can use. The most useful scenario for us would be to use a voting process we'll collect ballots, ballot-counters will sit and enter the ballot choices into an Excel spreadsheet, go through them a second time to confirm entries, and then feed the resulting file into OpenSTV. Of course, you can see the obvious first problem: Excel could output a "Raw" file, but lines would be comma-separated, not space- separated. Now, clever use of Unix 'tr' could solve the problem, but this requires we have a Unix dweeb around. There's a second problem I'd like to see solved, too: We'll only have around 50 or so votes (for 5 seats), and historically we've recorded the name of the voting member alongside their votes. If we put this info into the spreadsheet, it would introduce info into the input file that OpenSTV would barf on. Finally, I'm probably going to hire some consulting programmers I know in order to make desired changes in OpenSTV to fit our purposes. Before I set our programmers off on their task, I thought it would be best if I asked the following questions on this list. (Assume, for purposes of these answers, we have capable Python programmers at our disposal.) 1) Do you foresee any significant difficulties with modifying the ballot-file-reading code to read a source ballot file that use comma- separated entries on lines (vs. spaces)? 2) Do you foresee any significant difficulties with modifying the ballot-file-reading code to ignore everything on a line after the standard Python comment indicator is encountered (#) (so we could, e.g. put a #member-name into the last column on a line in Excel)? 3) Assuming I have these changes implemented, is this code you are interested in enough that you'd like us to feed the changes back into the open source code base? Thanks for your answers. Cheers -jb ---------- Jay Batson Acting Managing Director, Chairman bat...@si... +1-978-824-0111 |
|
From: Jeffrey O'N. <jc...@co...> - 2006-06-27 11:09:57
|
Dear list, I am planning some significant changes to pSTV, and I would appreciate your feedback and help. I am planning the following: (1) Changing the name from pSTV to OpenSTV. The next release of the code will be OpenSTV 1.0. I have already registered OpenSTV.org which points to the sourceforge page. (2) A slicker web page. I have set up a space to test the Xoops platform but I am open to other development tools. (3) A prettier logo and icons. I'd like to keep them generally the same but make them look nicer. (4) Clean up the user interface. It is currently a bit klunky so any feedback for making it more usable would be greatly appreciated. I would also like to add an option for html output and maybe also graphs. I'd greatly appreciate any help with any of the above. Also, beta testers would be helpful. I will post to this list when beta versions are available. best regards, Jeff O'Neill |
|
From: Jeff O'N. <jef...@gm...> - 2005-10-30 17:37:38
|
Two announcements: (1) pSTV version 0.9 is now available. (2) A much improved version of the STV Database is now online. ======================================================= (1) pSTV 0.9 New methods include N. Ireland STV and the Supplemental Vote. Condorcet, Meek, and Warren methods are much faster than before. Methods available: - SNTV - IRV - Supplemental Vote - British Columbia STV - ERS97 STV - Cambridge, MA STV - N. Ireland STV - Meek STV - Warren STV - customizable random transfer STV - customizable fractional transfer STV - Condorcet (completion with SSD, IRV, and Borda) - Bucklin - Borda - Coombs More details and download links are at http://stv.sourceforge.net. ======================================================= (2) STV Database 1.2 The STV Database includes two tables. The first table contains statistics of ballot data for 931 elections. The ballot data includes government elections (e.g., Cambridge, MA, and Ireland), university elections, and artificial elections. The second table conatains election results for the ballot data with 17 election methods (Borda, Bucklin, CambridgeSTV, CofE, Condorcet, Coombs, ERS76, ERS97STV, IRV, MeekSTV, NIrelandSTV, QPQe, QPQr, SNTV, SuppVote, WarrenSTV, XXvote). Users can browse the database online with SQL queries or may download the tables for use on their own computers. For more inforamtion, go to http://stv.sourceforge.net. New users should read the users guide at http://stv.sourceforge.net/guide.html. |
|
From: Jeffrey O'N. <jef...@gm...> - 2005-09-08 19:19:02
|
pSTV version 0.8 is now available. The visible differences are (1) and improved user interface, and (2) the ability to select candidates to eliminate before the votes are counted. Behind-the-scenes improvements include (1) that the ERS97 rules are now implemented in fixed point and not floating point which makes them more robust, and (2) that some methods are faster than before. Methods available: - SNTV - IRV - British Columbia STV - ERS97 STV - Cambridge, MA STV - Meek STV - Warren STV - customizable random transfer STV - customizable fractional transfer STV - Condorcet (completion with SSD, IRV, and Borda) - Bucklin - Borda - Coombs More details and download links are at http://stv.sourceforge.net Jeff O'Neill |
|
From: Jeffrey O'N. <jef...@gm...> - 2005-08-25 06:45:51
|
Hi Mike, This information is in the Ballots class in file ballots.py. The variable is Ballots.nSeats. You can access it from the Election class using Election.b.nSeats. I figured that when people are casting their votes, they will know how many people will be elected so I put the number of seats there rather than in the Election class. Please let me know if you need more information. Jeff > I am using the pSTV python module and am having trouble figuring out how to > tell the module how many people I want elected. In the Windows application > this is very simple. > > Thanks so much for your time! > > Mike |
|
From: Michael D. C. <Mi...@ID...> - 2005-08-24 19:46:50
|
Hello, I am using the pSTV python module and am having trouble figuring out how to tell the module how many people I want elected. In the Windows application this is very simple. Thanks so much for your time! Mike |
|
From: Christopher J. <jer...@ma...> - 2005-06-19 16:22:42
|
I just wanted to share a thought on surplus transfers with STV: In the help menu for pSTV, it says: "With fractional transfers of votes, secondary surpluses must be allowed. |
|
From: Jack V. <ja...@12...> - 2005-04-22 17:11:40
|
First, to Jeff, I wasn't specifically requesting such a feature, but thanks for entertaining the notion anyway, at least long enough to dismiss it. :-) To Jesse, Say a ballot in the 2004 US Presidential Election looked like this: 1. Nader 2. Kerry 3. Bush A typical STV ballot would be something like: 2 1 3 or 3 2 1 or 1 2 etc. With duplicate rankings permitted a voter may divide their vote for Nader and Kerry, giving each of them a first choice rank: 1=2 3 That's how ChoicePlus Pro notates such a condition. The voter has indicated both Nader and Kerry as a first choice. The ballot is then somehow divided into two separate ballots, each having one-half a vote in value. It can, obviously, (and does) get more complicated than this. Because elections for a national non-profit with membership distributed geographically all over the U.S. have high mailing and printing costs, aside from other election-related expenses, a bungled election could well constitute a financial calamity. I am also concerned that, if at some point ChoicePlus Pro becomes outdated software, Pacifica will have nowhere convenient to turn for help. It would be much easier for Pacifica to drop permission of duplicate rankings. Jack On Apr 22, 2005, at 4:56 PM, Jesse Mortenson wrote: > I don't understand this notion of duplicate rankings, but I'm > interested to know how not having that feature would have been costly > for Pacifica. Jack, could you give me a synopsis on or off list > sometime? > > Thanks, > Jesse Mortenson > St. Paul, MN > > Jeffrey O'Neill wrote: > >> I have never heard of any other organization or government body >> allowing duplicate rankings. Personally, I think it is a very bad >> idea to do so. With pSTV, I am focussing on implementing STV as used >> by various government bodies. Since I have a long todo list already, >> I won't be able to implement duplicate rankings. >> best regards, >> Jeff O'Neill > > > > > ------------------------------------------------------- > SF email is sponsored by - The IT Product Guide > Read honest & candid reviews on hundreds of IT Products from real > users. > Discover which products truly live up to the hype. Start reading now. > http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click > _______________________________________________ > STV-discuss mailing list > STV...@li... > https://lists.sourceforge.net/lists/listinfo/stv-discuss |
|
From: Jesse M. <te...@gm...> - 2005-04-22 16:37:51
|
I don't understand this notion of duplicate rankings, but I'm interested to know how not having that feature would have been costly for Pacifica. Jack, could you give me a synopsis on or off list sometime? Thanks, Jesse Mortenson St. Paul, MN Jeffrey O'Neill wrote: > I have never heard of any other organization or government body allowing > duplicate rankings. Personally, I think it is a very bad idea to do > so. With pSTV, I am focussing on implementing STV as used by various > government bodies. Since I have a long todo list already, I won't be > able to implement duplicate rankings. > > best regards, > Jeff O'Neill > |
