Thursday, May 9, 2013

The rise of the non-aligned voter (and fail of the null leaders)

Thursday is generally for business posts, and there was one for today. But it will be shifted to Friday to give space to this recent development.

The CSM election ballots are finally published and you can download it too. Many people did to calculate ordered lists, but I was interested in something else: the power of the blocks. Mynnna wrote: "Null bloc votes, as identified by Two Step: 9823+10957+5999 = 26799, 53.9% of the votes. Null bloc candidates: Five (Myself, Kesper, Sort, Sala and Progodlegend), which is about 35% of the council. I suppose you could count Malcanis as well, as the HBC ballot was instrumental in electing him and he is technically part of the bloc, but even that's only 42%. Basically, I'm saying that I think the people who were claiming that STV is all about rigging it for nullsec (ya'll know who you are) can put down their tinfoil hats. If anything we're under represented. "

Mynnna is wrong. He is wrong because Two Step was wrong identifying the voter blocks. Part of it was that he only looked for a few patterns, and looked for it everywhere in the ballot, while only the first spots had serious effect on the outcome. Below you can see a spreadsheet. Its fields show how many ballots have a certain candidate or pair of candidates in the top 7 places. I choose 7 because according to CCP that had 98.26% of the power of the ballots and because former CSMs had 7 Icelandic positions, so many people considered this an important number. The green background ones belong to one character, you can see that 12672 ballots had Mynnna on one of their top 7 positions. Next to it you can see "925", saying 925 people had both Mynnna and Ripard Teg on their top 7 spots.

Can you see the shocking result? Or just a bunch of numbers? Let me clear it up for you. At first I divided each line with the self-vote result, so every number of the first line was divided by 12672. From there the first line shows how big percentage of the Mynnna voters voted also for the other winner candidates. Then color-coded the fields, with the average value being white, below-average is red, above average is green. Finally I rearranged the candidates according to their cross-votes:

Now it's clearer, right? It shows that Mynnna, Kesper, Sort and Sala voters placed the other 3 to their ballots with high chance, though far from 100%. Only 75% of the Mynnna voters voted for Kesper, not really a good thing for CFC leadership. The cross-CFC-HBC votes were even smaller, though still high. What is more important here is that the rest of their lines and columns are red, meaning that those who voted for the CFC-HBC block ballot were unlikely to vote for other winning candidates (likely voted for non-winning block ones) and also that the voters of other winning candidates were unlikely to vote for them.

Following them comes the WH-pair. Their columns are red, so the non-WH voters were unlikely to vote for them, but their out-of-block lines are not red meaning that the WH voters were likely to support Mangala, Ripard, Trebor, instead of the WH 5 as instructed. Since the WH5 was a 5-long ballot, all of them fit into the top 7. According to the instructions, all WH pilots should have placed the 5 WH candidates to the top 5 positions in some order. Let's check it out:
Ouch! If everyone would follow the instructions, we should have seen 100% everywhere! The chance that a voter of a WH candidate also supported another WH candidates was around 50%! It seems the cooperation of the candidates is not enough to make their voters cooperate! Did it matter? To see that, you should know that STV runs in rounds, every round eliminating one candidiate. The ballots that had this candidate at the top position move to the next one on the list, so the rest of the candidate gain votes. Let's see how did it happen with the Wormhole 5:
The first eliminations gained 27 votes to the wormholers from voters who placed someone else first and wormholers behind. Then Cipreh got eliminated. At this move the WH top votes decreased by 155, which means that out of the 590 Cipreh voters 155 did not place another WH-er behind Cipreh. They gained 48 more voters with further eliminations and spillovers until Ayerson got eliminated, draining 310 out of 737, 42% of the Ayerson voters from the WH pool. Looking at the end of the run, it's likely that even that wouldn't get a third WH-er in, but still you can see the point. Same for the HBC-CFC block vote: when Kaleb Rysode was eliminated, 17% of his votes went to people who are not on the block list, half of it directly to the unaligned ones. When Banlish was eliminated, 41% of his votes leaked away from the null block.

After the two WH winners, there is a huge green field: the unaligned ones. See how these voters supported all candidates. They also got support from WH and Progodlegend voters, but gave no support to out-of-block candidates. It shows a huge number of votes supporting all of these candidates. Please note that the HBC member Malcanis belongs here. Despite what Mynnna wrote, the HBC votes were not getting him the seat. The Sort Dragon voters barely gave him more votes than they gave to Progodlegend and this was mutual. Those people supported Malcanis who supported the other unaligned ones. The null block was simply outnumbered by the unaligned ones. They were not simply voting randomly, they seem to like all these CSMs.

At the bottom corner sits Progodlegend all alone. His column is red, showing that no other winner had significant amount of voters who also supported him.

Hubris made nullsec leaders tell that they could place a basketball to their ballot and win. It seems their line members were not really motivated to support a basketball and many did not vote, or did not follow the official ballot! I was in TEST for months but never heard of Banlish, and probably other TEST voters neither. At least Sort and Sala often bridged me. For CSM9 the null leaders should take their own members seriously. Maybe they should run pre-eliminaries, finding candidates that have the real support of their members. Same goes to the Wormhole people. If those WH people who had no affiliation to any candidate had the same 50% "cross-support" as the affiliated ones, many WH votes were cast on non-WH candidates, mostly on the unaligned block. With 3 widely accepted candidates, they could get them all in.


Anonymous said...

There's nothing wrong with what Two Step did, but Mynnna interpreted the results incorrectly. Two Step's data shows what percentage of people put at least 3 from each bloc on their ballot, but since each person got 14 votes, lots of people are counted in more than one of his categories. Indeed if you add up all the totals it's well over 100%.

38% of voters put a straight up null candidate at #1, and 35% of the council is null, so at that high level, things worked out pretty much as expected.

Gevlon, your analysis is very interesting, I might do some of my own and post some results here tomorrow. In particular I'd like to look more into the breakdown of the WH votes, as the 50% crossvoting seems shockingly low.

Anonymous said...

James Arget was part of both the WH and the Trebor bloc, so it makes sense that there were a number of people voting for him but not Chitsa Jason. The converse however is a little puzzling.

2841 accounts had Chitsa in the top 7 but not James. Of these, 784 (about 25%) had a WH candidate in their top spot, 714 of which were Chitsa himself. Of those, 225 (roughly 33%) voted Chitsa and noone else, while 440 (60%) voted for a non WH candidate in slot 2, about half for Trebor and about half for Null.

Of those who had an HBC candidate in the first spot (6026), 80% put a null candidate in the 2nd spot and 5% had only cast a single vote. For CFC (7080), 91% had null in the 2nd spot and 2% with single votes. N3 on the other hand only had 66% vote null in the 2nd spot.

Of course, it's possible that some people put N3 people at the top of their list without actually being in N3 (similarly for HBC/CFC but their numbers were healthier)

Anonymous said...

interesting post - seems the HBC was better at voting for CFC candidates than vice versa.

As for Banlish - it's hard to believe you never noticed him as he is a very active forum poster who loves composing massive walls of (well-written) text.
He also is the CEO of Di-tron Heavy Industries (which used to be Bobby Atlas' corp).

Foo said...

I disagree that the WH group had a hope of seat number 3.

Foo said...

(For those missing part of the conversation, Gevlon commented back at my blog.)

I am now unsure of what you meant by "With 3 widely accepted [WH] candidates, they could get them all in." By definition, widely accepted candidates get in.

I have a different take. Some voters will vote for a bloc, others for personally known candidates.

If you restrict yourselves to only accepting popular bloc candidates, you will still have leakage, and not even pick up whatever portion of votes from the lesser known candidates.

The only time you can afford unknown (or lacking in popularity) candidates is when you have 'safe seats'.

First preference votes are very important to a ticket. To this end, I have a different message for leaders.

Next time leaders should motivate people to vote.

Gevlon said...

I think the people voted, just not as their leaders told. I don't think many "highsec carebears" supported Malcanis, Mangala or Ripard. Their voters are probably in null blocks, just gave the finger to their own leaders. Similarly many WH people cast votes for only 1-2 WH5 members. I assume that many cast votes that did not have a single WH candidate.

What we see here is a rift between the political elite and the electorate. The elite considered it evident that any WH (nullblock) candidate is good as they all fight for things like POS rewrite. They only tried to convince voters to place them on the first place, they did not even think of the possibility that they won't be on the ballot anywhere.

They did not bother convincing their "own" voters, campaigning among them, so the voters gave them the finger and voted for candidates who are nice, despite will not represent them and did not even offered it.

I assume most of the Ripard-Trebor-Malcanis-Mike votes are protest-votes as these candidates didn't really have a platform, did not say "I will fight for this and that". Their campaign was mostly "I'm a knowledgable, nice guy". This assumption is also supported by the fact that the voting number decreased for the first time. Most people did not see a "this is my guy" candidate, simply picked the least bad.

Anonymous said...

So I took Two Step's blocs (as I don't know all the candidates). 3 of the elected candidates were from 2 blocs, so to simplify I just counted votes in two blocs as 2 votes, one for each bloc, and said that 17 people got elected. 8 candidates were not in blocs but received less than 2% of the vote at any stage, so I just ignored those votes. The simplifying assumptions will change the numbers a bit, but not enough to significantly skew the results.

The makeup of the elected council matches within error (ie either rounding up or rounding down) of the percentage of total votes cast no matter where you cut off the ballots, top vote only, top 2, top 3, all the way to top 14, with Trebor being the only exception, getting 1 extra candidate vs a round 4-7 cutoff.

This suggests to me that even with more voters sticking to their bloc, the outcome would not have been different.

I can say much more confidently though, as I pointed out earlier, this shows that Mynnna was wrong, everyone is represented pretty much exactly right (at the bloc level)

CFC, WH and HBC gained popularity mid ballot, HBC gaining 40% (of their top slot vote), WH gaining 25%, but both lost that support by the end of the ballots. Those votes came from Trebor, N3, the Russians and Provi whose strongest support was in the top spot.

Anonymous said...

Your assumption, that if all WH voters would have voted for all 5 candidates, the cross reference should be 100%, is false.

It fails on the premise, that there only WH voters who would pick a WH candidate.

I am high-sec dweller. But since there is no way to fill the ballot that way, my exploring interest added 3 WH candidates (but e.g. I can't stand Chitsa). And no way I was going to add 0-sec bloc candidates (while a CSM without them isn't a good representation, they don't need my help to get there), so that added independends to fill the ballot.

So in the end, my ballot was high-sec -> WH -> independends I like. Of course I can't guess how many WH votes were created this way. But at least with me, it won't be 100% that way.

The same reasoning can be applied to 0-sec bloc candidate: Surely there are 0-sec non-bloc dwellers that think the one or other bloc candidate is a worth the vote.

Anonymous said...

I was in TEST for months but never heard of Banlish, and probably other TEST voters neither. At least Sort and Sala often bridged me.

BWAAAAHAHAHAAA!!!! Sometimes leadership get stuck in the echo chamber and believe their own propaganda!


Aeek said...

Who else is a high sec carebear going to support other than Ripard, and Mynxee back in the day. Interesting and intelligent blogs.
Would never have known about Mike without Ripard's plug.

Subscribe to the goblinish wisdom