1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
  2. Greetings Guest!!

    In order to combat SPAM on the forums, all users are required to have a minimum of 2 posts before they can submit links in any post or thread.

    Dismiss Notice

[Developer Blog] CSM8 Election Statistics

Discussion in 'EVE News' started by EVE News, May 7, 2013.

  1. EVE News

    EVE News RSS Feed
    RSS Feed

    Joined:
    Apr 12, 2011
    Messages:
    768
    Likes Received:
    0
    The elections for the 8th Council of Stellar Management (CSM8) are now all wrapped up and CSM8 is already in action reviewing providing feedback on many of the features coming in Odyssey.
    [​IMG]
    With the changes in this years election system to using the Wright Single Transferable Vote (Wright-STV), we at CCP thought it might be interesting to take a look at some of the important statistics regarding voter behavior and how ballots were counted. In the new voting system, voters were able to fill out up to 14 slots of their ballot, or "vector", in order of preference. The ballot if then processed by the Wright-STV code to ensure that it is optimally used to elect candidates in the voters order of preference without wasting the vote on those who can't use it or who do not need it, until there are no more candidates on their ballot (this is reffered to as exhaustion of a ballot). The percentage of the ballot exhausted to elect candidates is equivalent to the ballots "strength" in the election in comparison to other ballots.
    First up, we looked at the length of the ballot and examined how much value got exhausted per length
    Ballot Length
    Exhausted Value
    NumBallots
    Avg Exhaustion
    1
    2776.542908
    5349
    51.91%
    2
    726.688604
    4187
    17.36%
    3
    1258.299119
    10014
    12.57%
    4
    258.901186
    2900
    8.93%
    5
    135.865479
    2737
    4.96%
    6
    133.058269
    2760
    4.82%
    7
    34.767622
    1077
    3.23%
    8
    16.867895
    925
    1.82%
    9
    23.326805
    1145
    2.04%
    10
    4.779356
    613
    0.78%
    11
    4.92942
    389
    1.27%
    12
    3.273959
    423
    0.77%
    13
    4.049469
    414
    0.98%
    14
    97.557479
    16769
    0.58%
    Graphs of the last two columns:
    [​IMG]
    [​IMG]
    We also found it interesting to look at which slots on the ballots were used to elect candidates: This shows on average, how powerful each preference slot from 1-14 on each ballot was at electing a candidate to the council.
    Slot
    Sum of strength
    Percentage
    1
    26662.60939
    60.29%
    2
    10874.87429
    24.59%
    3
    3233.184747
    7.31%
    4
    954.072052
    2.16%
    5
    1252.032231
    2.83%
    6
    298.957331
    0.68%
    7
    178.279415
    0.40%
    8
    109.412348
    0.25%
    9
    492.259013
    1.11%
    10
    42.164441
    0.10%
    11
    25.816052
    0.06%
    12
    18.608337
    0.04%
    13
    14.566362
    0.03%
    14
    66.256416
    0.15%
    [​IMG]
    Here are some interesting high-popularity ballot orders and how they got applied. These are what many would refer to as "bloc votes". We bolded the bits that we found interesting.
    3964 votes followed this vector:
    ['mynnna', 'Kesper North', 'Kaleb Rysode', 'Sort Dragon', 'Sala Cameron', 'Banlish', 'Artctura', 'Malcanis', 'Unforgiven Storm', 'Awol Aurix', 'DaeHan Minhyok', 'Mangala Solaris', 'Psychotic Monk', 'James Arget']
    43.65% of your ballot expended to elect mynnna
    39.48% of your ballot expended to elect Kesper North
    Kaleb Rysode was eliminated in round 3 with 610 votes
    Sort Dragon was elected before your vote got here
    15.87% of your ballot expended to elect Sala Cameron
    Banlish was eliminated in round 16 with 2173 votes
    Artctura was eliminated in round 5 with 736 votes
    Malcanis was elected before your vote got here
    Unforgiven Storm was eliminated in round 11 with 1383 votes
    Awol Aurix was eliminated in round 4 with 687 votes
    DaeHan Minhyok was eliminated in round 1 with 246 votes
    Mangala Solaris was elected before your vote got here
    Psychotic Monk was eliminated in round 14 with 1844 votes
    0.99% of your ballot expended to elect James Arget
    1639 votes followed this vector:
    ['Sort Dragon', 'Sala Cameron', 'Banlish', 'mynnna', 'Kesper North', 'Kaleb Rysode', 'Awol Aurix', 'Artctura', 'Malcanis', 'Unforgiven Storm', 'DaeHan Minhyok', 'Mangala Solaris', 'Travis Musgrat', 'progodlegend']
    74.47% of your ballot expended to elect Sort Dragon
    Sala Cameron was elected before your vote got here
    Banlish was eliminated in round 16 with 2173 votes
    mynnna was elected before your vote got here
    Kesper North was elected before your vote got here
    Kaleb Rysode was eliminated in round 3 with 610 votes
    Awol Aurix was eliminated in round 4 with 687 votes
    Artctura was eliminated in round 5 with 736 votes
    22.90% of your ballot expended to elect Malcanis ß This explains the bump in the Strength/slot graph at #9
    Unforgiven Storm was eliminated in round 11 with 1383 votes
    DaeHan Minhyok was eliminated in round 1 with 246 votes
    Mangala Solaris was elected before your vote got here
    Travis Musgrat was eliminated in round 10 with 1103 votes
    progodlegend was elected before your vote got here
    2.63% of your vote was exhausted
    595 votes followed this vector:
    ['Ripard Teg', 'Malcanis', 'Trebor Daehdoow', 'Ali Aras', 'Mangala Solaris', 'Nathan Jameson', 'Korvin', 'Corebloodbrothers', 'riverini', 'Roc Wieler', 'Psychotic Monk', 'James Arget', 'Unforgiven Storm', 'progodlegend']
    63.40% of your ballot expended to elect Ripard Teg
    32.82% of your ballot expended to elect Malcanis
    Trebor Daehdoow was elected before your vote got here
    Ali Aras was elected before your vote got here
    Mangala Solaris was elected before your vote got here
    Nathan Jameson was eliminated in round 17 with 2430 votes
    3.77% of your ballot expended to elect Korvin
    Corebloodbrothers was eliminated in round 12 with 1575 votes
    riverini was eliminated in round 9 with 992 votes
    Roc Wieler was eliminated in round 8 with 895 votes
    Psychotic Monk was eliminated in round 14 with 1844 votes
    James Arget was elected before your vote got here
    Unforgiven Storm was eliminated in round 11 with 1383 votes
    progodlegend was elected before your vote got here
    The 1-vote vector peeps broke down like so:
    577 "Greene Lee"
    469 "Mangala Solaris"
    454 "Corebloodbrothers"
    410 "Korvin"
    280 "progodlegend"
    258 "Ripard Teg"
    247 "Trebor Daehdoow"
    239 "Psychotic Monk"
    234 "Nathan Jameson"
    225 "Chitsa Jason"
    207 "Malcanis"
    180 "Steve Ronuken"
    176 "Mike Azariah"
    175 "Sala Cameron"
    169 "PsychoBitch"
    156 "Ali Aras"
    107 "riverini"
    102 "Travis Musgrat"
    82 "Ayeson"
    82 "James Arget"
    82 "Unforgiven Storm"
    75 "Banlish"
    67 "Roc Wieler"
    62 "Sort Dragon"
    55 "Kesper North"
    48 "mynnna"
    47 "Cipreh"
    28 "Artctura"
    25 "Awol Aurix"
    21 "DaeHan Minhyok"
    10 "Kaleb Rysode"
    When we ran the STV-2 election to find the two most widely supported CSM members to be the new 2 permanent summit attendees in the 2+5 system we found that Mynnna and Ripard Teg were victorious, but that the difference between the 2nd and 3rd place candidates for the 2-seat guarantee was 299.11 votes between Ripard and Mangala (with Ripard winning).
    Finally, let’s take a look at how many votes actually affected the outcome of this election in comparison to the CSM7 elections.
    For CSM8 this number is easily found. The equation looks a little something like “candidates straight up elected + remainder votes for each person elected by default”, or in number form “10 * 3106 + 3091 (Korvin) + 2885 (Mike) + 2651 (James) + 2612 (Chitsa) = 42299 Votes”. This means that of the 49702 votes cast in the election 85.1% of them had a direct effect on the makeup of the council.
    Finding this number for CSM7 is a little more difficult. 14,813 votes were cast for candidates that did not make the top 14. These are "under-votes" and had no effect on the final council. The more difficult area is determining “over-votes” (votes cast for a victorious candidate that did not affect their standing and could have been used to elect a different candidate). If you consider being in the top 14 to be “unaffected standing” then there were 26,348 over-votes cast in the CSM7 elections. If you feel that being one of the top 7 delegates is "unaffected standing" (as many do) then there were 12,495 votes cast for those in the top 7 that did not affect the election outcome. This means that of the 59,109 ballots cast in the election, 31,801 ballots affected the outcome of the top 7 and 17,948 ballots affected the outcome of the top 14; That comes out to be 53% and 30.3% enfranchised votes respectively.
    This drastic improvement from 53% enfranchisement to over 85% is exactly what we were looking for when we made the move to Wright-STV. Now that a voters opinion is more likely to actually count towards the results of the election, our next big push will be towards visibility and voter turnout for the CSM9 elections.
    [​IMG]
    If you would like to do your own analysis of the votes you can find the full list of ballots cast in .blt format here and the ballots we used in the STV-2 election here. Also, you can find the previously published Wright-WTV code used to process the ballot here.

    Continue reading...