CSM 8 Results: Curiouser and Curiouser

csmlogo-blackSo at last, after strange time-dilation effects delayed the publication of the CSM results for more than a week, we have the actual breakdown of how the voting went. While we knew who was elected, many of us in the CSM-watching space were curious as to the order of finish. Here’s the kicker – there are several different ways to count it. And those different ways of counting show very different outcomes.

So let’s go through three of those methods. First is Trebor’s re-run of the official numbers. It goes through in painstaking detail and shows who got eliminated and who got elected when in the mechanics, and I think should constitute the “real ranking” for statistical purposes. More on that below. Second is Two Step’s method of re-running the election for each round of vote reviews, which does not accurately show who did get in and in what order, but in my opinion does a more accurate job of showing who truly had the most support. Finally, Ali Aras did a “what if we used the old method” of First Past the Post (FPTP) that was used in all previous CSM elections [Side note: Ali herself disclaims that the FPTP analysis is scientific, and she’s correct – people would likely have voted differently in an actual FPTP – still, it is an interesting look at who voted for who in the #1 slots – more below.]

Here are the outcomes side by side. Below we’ll talk about the names in various colors.

# Trebor Script Two Step “Re-run” Process Ali FPTP
1 Mynnna Mynnna Mynnna
2 Ripard Teg Ripard Teg Ripard Teg
3 Sort Dragon Mangala Solaris Sort Dragon
4 Mangala Solaris Sort Dragon Mangala Solaris
5 Progodlegend Chitsa Jason Progodlegend
6 Ali Aras Trebor Daehdoow Ali Aras
7 Kesper North Ali Aras Korvin
8 Trebor Daehdoow Kesper North Trebor Daehdoow
9 Sala Cameron Progodlegend Mike Azariah
10 Malcanis James Arget Greene Lee
11 Korvin Korvin Chitsa Jason
12 Mike Azariah Sala Cameron James Arget
13 James Arget Malcanis Nathan Jameson
14 Chitsa Jason Mike Azariah Sala Cameron

So … what can we take away from all this, other than the remarkable consistency of who is on the list if we ignore order? Trebor’s script shows some very interesting information when you break it down. Here’s how things actually shook out.

The Clear Winners

Ripard & Mynnna were first and second right off the bat. Let’s be clear, they won right away in round 1 of 17, walking away from the rest of the pack. There’s no question at all that these two should be the permanent Iceland seats. [EDIT: Two Step rightly points out below that to get the Iceland seats, which are run as a separate STV-2 election, they needed a much broader base than just their personal #1 votes. Both had this very strongly in that election as well. My point (which could perhaps have been made better) is that no matter how you look at the votes, Mynnna & Ripard come out 1 and 2 on every single permutation I’ve seen. Calling it a mandate.]

Sort Dragon was very clearly 3rd, elected in Round 4. However Mangala Solaris would have come in higher when you looked at it through the “per round” lens that Two Step processed the list with. Officially, Mangala would be elected in the #4 slot, but far later, not until Round 13 of 17! UPDATE: June Ting did an interesting “Name Recognition” analysis – essentially the truest “popularity contest” measure in my mind – that showed Mangala’s broad-based strength. As measured by how many ballots his name appears on, Mangala was the hands-down champ. He was listed on 23,410 ballots, well ahead of Malcanis at #2 with 19,513. However both of them were not listed as high on those lists. In the Name Recognition list, Mynnna (14,476), Ripard (13,862) and Sort (14,999) were much further down.

So what the heck happened in between? Elimination after elimination after elimination – nine candidates booted between 3rd & 4th place.

The Wormhole Brawl … and Korvin and Mike

So in my prediction post I stated that “a wormholer will almost certainly take either the 5th or 6th place.” Using Two Step’s analysis, this appears to be true. In reality, the wormholers scrabbled over votes until the very end.

Cipreh and Ayeson were eliminated very early (eliminated 2nd and 6th respectively), but James Arget, Chitsa Jason and Nathan Jameson battled all the way to the end of the list, neck and neck for who would get on.

Final result:

James Arget has 2651.6 votes
Chitsa Jason has 2612.74 votes
Nathan Jameson has 2430.01 votes

In the end, less than 183 votes separated #14 Chitsa from #15 Nathan, and James and Chitsa ended in a virtual tie, less than 39 votes apart. UPDATE: James may have been significantly helped simply by sheer volume of votes based on June’s list. On that list James is the #3 person in terms of number of ballots he appears on, 17,148 ballots. James was on the CFC list, the wormhole list, and likely made the list of Ripard, Trebor, and E-UNI voters.

But here’s the kicker: Fully half the CSM – everyone after Ali – was not elected until the very last round, #17 of 17. The quota needed to get elected was 3106 votes. This means that had there been more candidates, the vote would have gone onward. Instead, eliminations are what got James Arget and Chitsa Jason in. It’s also what got Mike Azariah and Korvin in – there were simply no challengers left.

Trickle-Down Politics

One other thing is interesting to me in the data – who were the biggest beneficiaries of STV? In the end, these folks aren’t terribly surprising, but there are some surprising elements in the numbers.

  1. Kesper North – 3514 Trickle-Down votes. He got enough from Mynnna and others to elect an entire candidate all in and of themselves. This makes sense because he was #2 on the CFC list. Kesper’s progression through the STV is the most interesting of any candidate. It looks at first like he will come in 4th, in Round 6 … but in the end his massive trickle vote pile gets pushed around and back all the way to the very end, when he is elected in 7th place way down in Round 17.
  2. Malcanis – 1966 Trickle-Down votes. Malcanis was a unique candidate. He had a strong corp and Alliance base. He had strong endorsements from across the community. He had a strong forum reputation. But he was halfway down the list of both of the blocs, putting him at risk of not getting on without community support. A third of his trickles came early on, from Ripard Teg voters (584 votes). Another third appeared suddenly in the very last round after Sala was elected – from HBC voters (679 votes). A third more were from candidates across the board – trickling in from across the spectrum of New Eden to push him over the top. Interestingly, Malcanis didn’t need all of the Ripard Teg votes … which cascaded primarily to Nathan Jameson … but not enough.
  3. Sala Cameron – 1750 Trickle-Down votes. Much like Kesper benefitted from being the CFC #2, Sala was the HBC #2, and thus picked up everything not needed by Sort Dragon.

UPDATE: In addition, June’s Name Recognition list highlights these folks in a further. On that list Malcanis is #2, Sala is #5 – they simply appeared on a very large number of ballots. Kesper is #11 on that list, reflecting that the bulk of his roll-down was direct from Mynnna.

Too Late Past the Post

So if they were the beneficiaries, who lost out? Well, as Ali Aras points out in her FPTP post, it is likely that people would have voted quite differently in a FPTP election. It is likely that Kesper still would have gotten in due to his bloc voting differently, but Malcanis might have missed out. Still, clearly this isn’t scientific. Still, I believe that most people (other than Banlish voters) would have put their #1 choice first. Thus while not scientific I think Ali’s list is still a good predictor of who else would have had a chance. So who didn’t get on that might have been able to?

  • Greene Lee. As a current CSM member, Greene still had a shot. He came in 17th. Together, Greene and Korvin had just over 3700 native votes. Greene got almost no trickle vote whatsoever (only 241 votes … compare to Kesper’s 3514). Combined one of them could have placed higher, but there were not enough votes to get both Russians elected.
  • Nathan Jameson. Nathan finished 15th of 14. It is remarkable how close we got to having 3 wormhole candidates on the CSM, and I congratulate him on a very close run.
  • Banlish. Seriously, TEST voters should just be ashamed. The new CSM has barely taken office, the HBC has broken up, and TEST is unrepresented. He can’t even say he was 15th … he finished 16th, behind the #3 wormhole candidate. Epic fail voting. Epic. [EDIT: As Two Step notes below, if it had been a 15-person CSM Banlish would actually have gotten in ahead of Nathan due to further trickle-down. But this doesn’t diminish my point … Banlish should have been in the top 10 had TEST served itself well.]

Lessons Learned

It was a new system, and now that we have one election under our collective belts, it will get tougher for independents next time. Presuming the method remains the same…

  • I expect first off that the object lesson of Banlish will ensure more Alliance voting and less Coalition voting. The null blocs likely will readjust their strategies to have several groups of voters, voting with specific patterns of #1-#5 votes. They’ll vote 14-candidate slates to be safe, but count on their top 2-5. [EDIT: Mynnna tried this below and it actually made the CFC results worse at some level, better at others (be sure to read all of his comments!). It would be more efficient to do it in multiple blocs, but they would need more voters to pull it off. However it happens, I have every confidence that given a year to work on it, Mynnna & company will find a good set of tricks to keep their representation high. It’s worth noting that Mynnna himself is a bit of a “transcendent” candidate – while he was heavily Goon-driven, he had appeal and respect outside of the CFC. Another good example would be Malcanis. This may be the way of the future candidate – a solid reliable core base but a broad appeal. That can only be good for the game.]
  • The wormholers got lucky this time. The amount of candidate equivalency could have backfired as easily as succeeded. In my opinion, we need to narrow the field earlier next year, and field no more than three candidates; ideally two. [EDIT: Foo wrote a great response to this point in the comments below. I encourage you to read it. In essence it says wormholers were basically guaranteed one seat, and two was a likelihood even with the photo finish. He suggests that the wormholers would potentially be better off with an agreed rank order, but beyond that no worries. I buy that.] We voted with remarkable discipline considering that we have no official ties to one another, and across the five candidates we picked up an additional 1488 non-wormholer votes. Makes me proud to be a wormholer myself.
  • There are some new blocs and old half-dismissed blocs to be reckoned with.
    • The Frustrated Disenfranchised. Ripard Teg #2, beating out the leader of the HBC. Trebor #8. Mike Azariah #12. Need I say more?
    • The Warriors of Highsec. Red vs. Blue and associated groups, plus likely some wardeccers and EVE University students put Mangala Solaris in the #4 slot, handily above most of the rest of the field.
    • N3. They ran a terrible slate this year in my opinion, but they proved they could get someone in, and at a high level. Progodlegend took #5.
    • Provi-Bloc. Never heard of them before a month ago. That’s a bloc? Ali Aras, #6, including 118 votes from Ripard Teg voters. ’nuff said.

OK, I’m done with CSM for a while. This post was assembled hastily, and I’m by no means a STV scholar, so please feel free to call out any issues in the comments and I’ll tweak as needed.

Congratulations again to the winners. Do us proud.

Advertisements
This entry was posted in CSM, CSM 8 and tagged , , . Bookmark the permalink.

12 Responses to CSM 8 Results: Curiouser and Curiouser

  1. mynnna says:

    For what it’s worth, the very first thing I did with the ballot was take all the variations of the CFC ballot (there are many), reassign them to the “proper” CFC ballot, and then split them three ways equally, with a third placing Kesper first, a third placing Kaleb first, and a third placing myself first.

    It didn’t make a damn lick of difference. Kaleb getting eliminated early was an artifact of us not having enough raw votes, period.

    • mynnna says:

      I should revise my previous statement. It doesn’t make a lick of difference to the end results, though the way the mechanics play out are a little different. Ripard gets elected first, then Sort. I don’t actually get elected until the fifth round, and Kaleb actually winds up going out on the very last round. The election order looks to be ripard -> sort -> me -> mangala -> pgl -> ali -> trebor -> james -> korvin -> sala, with the other four getting in only after Kaleb is eliminated, by default – a skin of the teeth scenario, basically, though I don’t have Trebor’s script on hand so I can’t say how close it is.

      Anyway the point is that turnout rules all. I’m not sure how many additional votes in either scenario we’d need to change the outcome and get Kaleb in, but I suspect it’s a similar number in either case. Perhaps that’ll be an experiment for tomorrow. Games like shuffling the vote order as you suggested could pay off in corner cases, but could just as easily screw you over in other corner cases.

  2. Rhavas says:

    Good point – thanks for running the numbers (seriously not my strong point). From that point of view, it looks like you made the right choice because it got you one of the confirmed Iceland seats. Still, I have no doubt you will come up with something clever next year. 🙂

  3. Foo says:

    I think your analysis, especially on what wormhole candidates should have done, stands for first past the post elections. I disagree with it for preferential elections. I started to write a comment, realised I was waffling for a wall of text, so posted it at my blog http://foo-eve.blogspot.com.au/2013/05/breakdown-of-wormhole-csm-vote.html

    • Rhavas says:

      If you’ve read this far, go read Foo’s post. It’s excellent relative to the “lessons learned” of the wormhole vote and from someone (unlike myself) who is very familiar with STV. I’ll be revising my comments based on this.

  4. Two step says:

    A couple of things:

    Firstly, Banlish was actually #15, not Nathan, despite Banlish getting eliminated earlier. If you run the election for a 15 person council, more votes trickle down from HBC/CFC ballots, and those push Banlish past Nathan.

    Secondly, Ripard and mynnna were not the top 2 because they got elected right away. In order to get a top 2 slot, both had to have over 10,000 votes, and mynnna started out with 5782, while Ripard had “only” 4181. If both of them were not on a bunch more ballots they would not have gotten those top 2 slots.

    • Rhavas says:

      To the first point, I think that would make a third column in my table above. This was based purely on Trebor’s processed results, not on a 15-, 16- or 17-member council to determine “true chances” of Banlish, Nathan and Greene.

      To the second point, this might well be a confusion about STV on my part. What I saw in the results was a quota (3100-something IIRC) which both easily met with 5782 and 4181. But what I read into that is that, and what the processing shows, is that those overvotes all flowed down to other candidates in the 14-person check. So … I’m presuming you’re talking about the “STV-2” vote that was actually used to calculate the Iceland attendees. Certainly in that one you are very much correct, and I’m sure that’s why they use them. Perhaps I could have worded it better around the Iceland attendee “second processing” portion.

  5. mynnna says:

    So it turns out I’m wrong (shocking, right?) With the shuffled vote scenario it would take somewhere between 594 and 669 extra votes spread across the three theoretical ballots to get Kaleb in, assuming no one else’s voting changed. For the brute force approach – one list being voted to strictly – it would take, well…I stopped messing with it somewhere around 1100.

    Something for us to remember next year, I guess. 🙂

  6. Ali Aras says:

    Regarding provibloc, Provi turnout was lowwww, not even enough for them to get one seat on the council. There were only something like 1200 provi voters (defined as people who voted Ali Aras/corebloodbrothers in either order). I’m not going to deny that the ~600 Provi first place votes helped a lot (as did the trickle-up from the other set), but 3/4 of the initial votes came from “outside provi” and ~61% of the votes that actually elected me were outside provi votes. I’m still playing with numbers/graphs to find out exactly who else they were.

    I’m hoping that having an actual CSM member makes them more motivated next year, though. Who knows! Either way, it’s like herding cats to get Provi organized.

  7. Lei Merdeau says:

    (just found this via your NDA post) I’ve been voting proportional for years. This was different in that there was no Party Party type joke candidate to vote first (100% transfer, my vote is worth MOAR!) Also I’m a blog voter rather than a bloc voter. Mynxee back in the day. This time I put Ali 1st on Ripard’s plug, never heard of her but good platform and figured she needed it more.

    As for major alliances gaming the system, that’s likely to be counter-productive, CCP will rightly be dubious of the CSM – much better for them to maximise someones vote for the status.
    Also, any CSM member is a partial loss to their corp, so do you want to lose extra heavy hitters?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s