Internet Space Misadventures

The blog of one Ali Aras, an FC, Provident, and member of CSM8.
0 | 11.5.2013 | 1 year ago

If you give a nerd a dataset…

…she’ll spend all damn week playing with it it.

Here is the product of about four lunch breaks and one long evening: a look at whatever I can glean about the CSM election from the actual ballots that were cast.

Some things are obvious from just looking at the results. TEST got screwed. The HBC in general did not vote in the kinds of numbers that the goons did. My estimates to Alikchi regarding Providence turnout were laughably high (c’mon, people). The infamous Bloc Ballot did not go as far as the propoganda/doomsayers said it did.

Single Voters

The first thing I looked at was the single voters, those who voted for only one person in the election. CCP put this in the stats devblog, but here’s a table version with the votes and the votes as a % of that candidate’s first place votes.


The percentage of total votes is, of course, influenced by how many total votes that candidate *had*, so sorting by that won’t match sorting by total number of solo votes. Solo voters are most likely those who either didn’t know anything about being able to vote for multiple people or those who were pushed to vote because they knew/supported a candidate, but didn’t care much about the election as a whole. Having a higher percentage of solo voters didn’t necessarily rule a candidate out; looking at the placement of the winners it doesn’t seem to have been a problem.

If those solo voters had instead voted for the candidate’s slate (or at least, the candidate’s preferred second choice)…nothing about the election would have changed. Their votes enhanced people’s vote totals, maybe keeping them in longer, but if they’d voted more intelligently (or in lockstep with their candidate’s preferences), the election would not have been different.

Common Pairs

So, okay, who *did* people vote for? Gevlon Goblin has a nifty spreadsheet visualizing this over here, and I have my own table versions. First, the most common pairs. All of the bloc votes other than wormholes are visible here. The Ripard Teg Ballot pulls in at third, followed by N3, an HBC variant, Russian voters, and Provibloc.


The number of potential combinations of votes influences what’s in the table. Wormholes aren’t visible here because there are 20 possible wormholer combinations, enough to put their pairs below the threshold for inclusion in this table. [Empty] makes such a good showing because there’s only one possible way to vote for just Greene Lee.

To take order into account, I looked at the most common sets of top-threes. Because these are sets, it doesn’t matter if someone voted mynnna/Kesper/Kaleb or Kesper/mynnna/Kaleb, they count as the same set. Since order isn’t important, wormholes make a strong showing here— Nathan Jameson, Cipreh, and James Arget were the most commonly-combined top-three. Not counting the Ali/Core/nobody else voters, Provi’s third choices were diverse; the remaining 600 or so Provi votes aren’t visible on the table. Again, [Empty] has an unfair advantage due to the few number of ways to vote for just one or two candidates.


After the top three or four ballots, the number of votes per set of three is really low. I plotted the number of votes for the top 100, and you can see how quickly the number of votes drops off.


If you want to see who was a likely second choice for each candidate, check out this spreadsheet. Here’s a sample of the results, but there were too many rows to include in this post.


The Big Blue Ballot

Throughout the election, one of the big things trumpeted from both the Bloc and “Independent” camps was the incredible power of the nullsec bloc ballot (by which, of course, it was meant the CFC/HBC ballot). So how well did it do in the actual election?

Well, they each got their top four in. The top HBC votes elected Sort Dragon, and the CFC votes elected Mynnna, Kesper North, and eventually Sala Cameron. The bloc voting was pretty consistent too. I looked at how many ballots had any of the HBC/CFC candidates in their top six. Here’s how many ballots had some overlap with the HBC/CFC ballot, graphed according to how strong that overlap was (1 candidate versus 6 candidates). It makes a neat little u-shape. 27,613 voters had no overlap at all, plenty of ballots had 1-3 candidate overlap (conscience voters), and then there’s a drop until you get to 6 candidate overlap.


When you look at who was removed in those 4 and 5 candidate overlap ballots, they’re mostly swaps of one bloc candidate for other candidates lower down on the ballot. These changes in organization didn’t really affect the outcome of the election.

If you look at the voting log, though, you can see that candidates who were not on the top four and were viewed as/ran as primarily bloc candidates lost out, big time. The great tactical moral of the story from this election is that it pays to build up a base outside of whatever support you may get from your buddies. If you don’t, you’ll stand a good chance of being eliminated early on, never to cash in on those trickle-down votes from slate voters.

blog comments powered by Disqus