Monday, March 19, 2018

CSM 13 Election: Explaining The Single Transferable Vote

Starting with the election of the 8th Council of Stellar Management, CCP has used a modified version of the Wright single transferable vote system (STV) to conduct all elections. Before 2013, CSM elections were conducted using widely used First Past the Post (FPTP) system. Players would cast votes for a single candidate per account, with the 14 candidates receiving the most votes becoming either members of the CSM going to the summits in Reykjavik or serving as alternates. A simple to understand way of conducting an election.

The design of the single transferable vote system aims to elect candidates who are more representative of the players who vote in the election. For some background about STV, I'll turn to UK-Engage, a company that provides voting services to both public and private organizations:
"The Single Transferable Vote (STV) is a form of proportional representation voting system which uses preferential voting, usually in multi-member constituencies. Candidates don’t need a majority of votes to be elected; all they require is a known ‘quota’, or share of the votes, determined by dividing the number of valid votes cast by the number of positions to be filled, plus one..."

"Under STV, an elector has a single vote that is expressed by ranking the candidates in preference from ‘1’ until the elector cannot choose between the remaining candidates. As the count proceeds and candidates are either elected or eliminated, this vote can be transferred to other candidates according to the voter’s stated preferences. In a Single Transferable Voting system very few votes are wasted; unlike other voting systems, particularly First Past the Post, where the votes of all but the winning candidate are wasted."

"STV is arguably a much more representative and inclusive voting system as it gives voters more choice than any other system. This choice puts more power in the hands of voters, rather than the political parties: under other voting systems political parties can more easily determine who is elected. Under STV some would say that an elected representative is much more accountable to the electorate than to their party superiors."
UK-Engage also listed several advantages and disadvantages to the single transferable vote system. Distilling the lists down to those that apply to EVE, they are:

  • Under STV fewer votes are ‘wasted’ In other words, losing candidates receive fewer votes and run-away winners receive only the votes needed for victory. This means that most voters can identify a representative that they personally helped to elect
  • When voters can rank candidates, the most disliked candidate cannot win, as they are unlikely to pick up second, third and lower-preference votes.
  • By encouraging candidates to seek first, as well as lower-preference votes, STV removes the need for tactical voting. Tactical voting is when a voter supports another candidate more strongly than their sincere preference to prevent an undesirable outcome.

  • A system allowing voters to rank candidates is prone to behavior termed ‘donkey voting’, where voters vote for candidates in the order they appear on the ballot. CCP fights against donkey voting by randomizing the candidate list for each vote.
Explaining how the Wright STV system works is a little challenging. Instead of confusing the issue, I will post an explanation from Wikipedia, followed by an example based on a CSM election. First, the explanation:

"The system uses the optional Droop quota (the integer value of the total number of votes divided by one more than the number of vacant positions plus one) and the Gregory method of weighted surplus transfer value of the vote in calculating a candidate's surplus transfer value which is then multiplied by the value of each vote received by the candidates whose votes are to be redistributed, as is the case in the Western Australian upper-house elections.

"Unlike the Western Australian upper-house electoral system, the Wright System uses a reiterative counting process that differs from the Meek's method as an alternative to the method of segmentation and distribution of excluded candidates' votes.

"On every exclusion of a candidate from the count the counting of the ballot is reset and all valid votes are redistributed to candidates remaining in the count initially at full value.

"In each iteration of the count, votes are first distributed according to the voter's first available preference, with each vote assigned a value of one and the total number of votes tabulated for each candidate and the quota calculated on the value of the total number of valid votes using the Droop quota method.

"Any candidate that has a total value equal or greater than the quota is provisionally declared elected and their surplus value distributed according to the voter's nominated subsequent preference. If the number of vacancies are filled on the first distribution the results of the election are declared with all provisionally declared candidates being declared the winner of the election.

"If the number of candidates provisionally declared elected is less than the number of vacancies and all candidates' surplus votes have been distributed then the candidate with the lowest value of votes is excluded from the count. The ballot is reset and the process of redistribution restarted with ballot papers being redistributed again according to the voters next available preference allocated to any continuing candidate. This process repeats itself until all vacancies are filled in a single count without the need for any further exclusions.

"The Wright System takes into account optional preferential voting in that any votes that do not express a valid preference for a continuing candidate are set aside without-value and the quota is recalculated on each iteration of the count following the distribution of the first available preference. Votes that exhaust as a result of a candidate's surplus transfer are set aside with the value associated with the transfer in which they exhausted."
Confused? I was when CCP first announced the use of STV in 2013, the dev blog instructed players to visit Wikipedia for an explanation. I needed an example before I truly understood how the algorithm worked. Having covered five CSM elections using the STV, I think I can do a little better than the generic mechanical explanation. In the below example, I will not only go through the mechanics, but introduce some of the situations that occur in a CSM election.

Example Election

In our example election, 8 candidates are vying for 4 seats. We can divide the candidates into 3 security zones, each with their own political dynamics.

Null security space: In our mythical election, we have three null sec voting blocs. They are:

Goonswarm Federation (GSF) - Goonswarm historically is the electoral powerhouse in CSM elections. In our example, GSF is running two candidates, Aryth and Innominate, winners in both the CSM 11 and CSM 12 elections. The Goon campaign team usually supports other null sec entities, so will finish off their voting slate by supporting candidates from fellow Imperium member The Bastion as well as Brave Collective.

The Bastion - Smaller alliances within the Imperium often attempt to win a seat on the CSM. In our example, The Bastion's Sullen Decimus, a member of CSM 11, will try to win a third seat for the Imperium. The Bastion made a deal with Brave Collective to put his name second on their ballot if The Bastion would do the same for Brave's candidate.

Brave Collective - Brave always seems to win whenever they officially back a candidate in a CSM election. In our example, Brave will run CSM 12 member Yukiko Kami. Brave's voting slate will mirror The Bastion's, except for putting Yukiko in the top slot and moving Sullen down to the second spot.

Wormhole space: The dynamic in wormhole space is a little different than in null sec. Instead of working within established major blocs, wormholers got together and two candidates, Noobman and ExookiZ, rose to the top. The backers of each candidate promised to put the other faction's candidate in the number two slot to ensure at least one wormhole candidate would win. Wormholers didn't care about candidates from k-space, but didn't want to waste the other two slots, so filled them in with Sullen Decimus and Yukiko Kami.

High security space: High sec is a very splintered, unorganized area of EVE. In our mythical election, two candidates representing two very different playstyles will run. They are:

High sec gankers: The high sec ganking community is very loud, very passionate, but very small. To maximize turnout, the famous ganker James315 is running in our election. Since Goonswarm's Ministry of Love often operates with the gankers in high sec, James315 fills the remaining three slots on the ballot with members of the Imperium.

High sec carebears: To say that high sec miners, mission runners, and haulers do not like high sec gankers is an understatement. But the residents of high sec, with the exception of the gankers, are a pretty apathetic lot. The turnout from those who enjoy PvE in high sec is usually very low. Additionally, in our example, they could only come up with two other candidates to vote for: Sullen because he did some mining stuff on CSM 11 and ExookiZ due to giving some presentations at Fanfest.

The Vote

Initial Vote
In our election, 1800 players voted. To calculate the number of votes required to guarantee election, known as the Droop quota, we use the following formula:

(Number of votes cast / number of positions + 1) + 1

The initial quota is 361. The number of votes for each candidate at the beginning of the round is below.

From the audit file

Innominate starts the round in last, but that won't last long. Aryth has 650 votes, which exceeds the quota. Aryth wins the first seat and the extra 289 votes he didn't need fall to the second candidate on the Goonswarm ballot, Innominate. At the end of the round, the tally is:

From the audit file

Since the voting does not have 4 winners, the algorithm removes the candidate with the lowest vote total, Sullen Decimus, and distributes his votes to the number two candidate on The Bastion ballot, Yukiko Kami.

Round 2

Round 2
The number of ballots is still 1800, which means the quota is still 361. The tally for each candidate at the beginning of round 2 is:

Beginning of round 2, from the audit log

Not much happens in the round. Once again, Innominate receives the overflow vote from Aryth to stay in contention for a seat.

End of round 2, from the audit log
Since the voting does not have 4 winners, the candidate with the lowest vote total, ExookiZ, is eliminated and his votes distributed to the number two candidate on the Wormhole #2 ballot, Noobman.

Round 3

Round 3
The number of ballots is still 1800, which means the quota stays at 361. The tally for each candidate at the beginning of round 3 is:

Beginning of round 3, from the audit file
The cooperation between the wormhole candidates worked, with Noobman exceeding the quota in round 3 of the voting simulation. Innominate, as was the case in the first two rounds, receives 289 extra votes from Aryth. With Noobman meeting quota and winning a seat, his 39 excess votes go to Yukiko, allowing him to surpass Lorelei's total. The tally at the end of the round is:

End of round 3, from the audit file
Round 4

Round 4
With the elimination of Lorelei Ierendi in round 3, the high sec carebear voting slate expires, reducing the number of ballots from 1800 down to 1570 votes. The result is a decrease in the quota from 361 down to 315.

Beginning of round 4, from the audit file
The reduction of the quota increases the trickle-down from the candidates who already exceeded the quota. So instead of Innominate receiving 289 of Aryth's excess votes, he receives 335 votes. Yukiko receives more votes from Noobman, increasing his total from 39 to 85 votes. Due to the decreased quota, Innominate exceeds the quota by 20, winning the third seat and passing those votes to Yukiko as well.

The final round, from the audit file
I'm not exactly sure why the algorithm indicates Yukiko had only 314 votes while the actions during the round show Yukiko should have 315 votes. Either way, Yukiko surpasses James315's vote total of 310 and wins the final seat.


Perhaps a game as famous for its complexity as EVE Online should use a complex voting system like the modified Wright single transferable vote for determining the winners of the Council of Stellar Management election. Others may argue that using a single transferable voting system adds needless complexity to an election that historically has low voter turnout. I just wish I didn't need to write a 2100 word essay in order to explain how the voting system works.

Friday, March 16, 2018

The CSM 13 Election Season Begins Monday

What I deem the EVE silly season, known to players who know about the event as the Council of Stellar Management elections, officially kicks off this year on Monday, 19 March. Once again, I plan to cover the event both here on The Nosy Gamer as well as on my election campaign coverage site CSM Wire. But before getting into my planned coverage, I probably need to cover what the heck a CSM is first.

The Council of Stellar Management is a body of 10 players chosen by the EVE player base to “represent” the players to the game’s developers. I used scare quotes because CSM members do not have the same type of power as a legislative body in a Western-style democracy. Think Beijing, not Westminster.

In other words, the CSM does not get to dictate the developmental roadmap of EVE Online. In some respects, the CSM acts as a player focus group, with the development staff bouncing ideas off the members. The most visible sign to the player base are the two summits between CCP and the CSM held every year in Reykjavik, Iceland. The CSM and CCP also hold weekly meetings where they also discuss issues, with the meetings recorded for later viewing by those unable to attend due to real life concerns such as work.

Unlike a normal focus group, the CSM maintains in contact with the developers year-round (or at least when the office isn’t vacant due to the staff going on holiday). In addition to text chat, the CSM has access to the developers’ Confluence, CCP’s internal wiki.

Where the term representative comes closer to the ideal of a representative body is when the CSM interacts with the players. The members of the CSM often act as a communications conduit between CCP and the players, explaining actions in ways that CCP’s messaging sometimes sugarcoats. Members also reach out to the player base looking for feedback on certain areas that CCP may explore in the future.

The members of the CSM are elected to a one-year term in an election based on a modified Wright single transferable vote system. Players with a subscribed account (aka Omega accounts) may vote for up to 10 candidates, with an algorithm determining the winners. Players can cast as many votes as they have subscribed accounts. Theoretically, a player running a 100 account skill point farm could vote 100 times.

This year, CCP changed the timeline of the election. Instead of holding the election in February or March and announcing the results at Fanfest, the official kick-off of the campaign begins at Fanfest with the announcement of the candidates. The application period only runs for five days, from 19-23 March. Firm dates for the election itself are not available, but the announced time frame is sometime in late April or May. Afterwards, the CSM will onboard during the traditionally quiet summer season when the CCP staff are off on holiday.

My planned coverage on the blog is to try to explain some of the details around the election. For example, the single transferable vote is pretty complex and a turnoff to a lot of people. As for CSM Wire, I will only include those candidates who post in the campaign thread on the official forums until the official candidate list is announced at Fanfest. Since the instructions to candidates suggests making a post, I hope candidates start posting as soon as they apply and don't wait until the Fanfest announcement. I currently don't plan on updating the site while I am in Reykjavik with new candidate pages. Deleting a page is easy. Creating pages takes some time, and I'd rather spend that time running around Iceland and talking to people I only get to see once a year.

Currently, 7 candidates have posted campaign threads, with several current CSM members yet to publicly declare. I expect to see a lot more people running this year, as many candidates are working behind the scenes trying to gather support. To everyone running, good luck!

Friday, March 9, 2018

The Guardian's Gala 2018: An After Event Report

With the end of the Guardian's Gala event I thought I would try something a little different. Normally most people, including myself, give previews of the events. Hardly anyone ever looks back at the event. After getting one character up to the 700 point prize and another up to the 500 point prize, I think I can provide a valid perspective on the event.

First, an overall opinion. The Guardian's Gala event was not designed for newer players (or at least, newer players who had not purchased skill injectors). With the changes to the Alpha skill set and the ships they can now fly, sufficiently skilled Alpha characters could run the sites. I think their best choices were shield-based ships like the Drake and Cyclone, with the Gila a viable option for returning players with the skills. I’m not sure how many new Alpha characters have the skills to fly the Gila well.

I used a Claymore, a Minmatar command ship, to run the sites. I chose the Claymore because, in addition to the ship’s inherent tech 2 resist profile, the ship gives bonuses to both missile rate-of-fire and active tanking. I also chose to fit two command burst modules to enhance the ship’s defenses. While the event NPCs never broke through my shields into armor, I did go through a fair amount of command burst charges and Cap Booster 400s powering my Xtra-Large Ancillary Shield booster. I also did not lose a drone during the event, with my drone repair costs totaling under 50,000 ISK. To end the breakdown of consumables, I used approximately 50,000 Nova heavy missiles. I chose to go with the tech 1 missiles because at the beginning of the event I made the determination that the event was not worth using tech 2 or faction missiles to complete. I might have used tech 2 missiles if I ran any sites in low security space, if only to clear the sites faster to avoid interlopers attempting to kill me.

Now for a couple of details I did not learn from my testing on Singularity. First, warping into the sites at 50km and then aligning at speed to a celestial produced an interesting response from the NPCs. The frigates would warp to a perch and then warp down on top of my ship. The cruisers, on the other hand, would start to slow boat toward my ship.  If I could fire fast enough, I could kill the initial cruisers in the site before the frigates could reach me to apply webs. Next, I found out that when the loot ships appeared on grid within 100km, the best course of action was to turn around and close the range, as the ships would not leave grid to bounce closer. Turning around also gave me a chance to kill more frigates, which upped the number of event NPCs I could kill in each site. The last change involved other players warping into the site. The player closest to the structure would gain the most aggro to the NPCs. At first, I would enter the sites at 100km to make the NPCs have to travel farther on the theory that would mitigate the amount of damage I took. That plan worked but left me vulnerable to having a site swiped. The move to warping in at 50km was an effort to counter the stealing gameplay.

CCP also instituted a couple of balance changes to the event sites. The most noticeable occurred on the third day of the event. Once the boss ship with the loot was defeated, the NPCs would pack up and leave the site. Before the change, the NPCs would continue to web the ship down and try to kill you. The change to the NPCs giving up and warping off was very welcome and led to quicker completion of the sites.

The next area to look at is the event prizes. The event gave players prizes for reaching 100, 300, 500, and 700 event points. The minimum number of sites one could run to get to the 700 points was 42, or three per day if one logged in every day of the event. I thought the prizes for the first three plateaus were appropriate for the level of work. At 100 points was a new type of firework, which was also a drop. I wound up with 5900 of the fireworks. At 300 points, CCP rewarded players with an advanced skill accelerator with granted players a +12 bonus to all learning attributes for 24 hours (48 hours if the character trained Biology V). These advanced accelerators, along with the standard accelerators that only provide a +10 bonus to all learning attributes, also dropped during the event. The premier reward was a new type of implant that gave a +5% boost to web range. These implants dropped in the VIP rooms, which only spawned outside of high sec. For high sec players, the only chance to get one of these implants was to reach 500 event points.

The 700 point reward, a 72-hour accelerator that granted a +10 bonus to all learning attributes (144 hours with Biology V), I felt was optional. In high sec, the drop rate was about 2%, so if one dropped, the final reward was a bonus for either an alt or selling on the market. If CCP wanted a prize that people wouldn't feel required to grind towards and burn themselves out, I believe they succeeded.

One of the most popular questions about events like the Guardian's Gala is whether the rewards make the event worth running. Over the course of the event, I completed 92 event sites in approximately 15 hours. I consumed a little over 10 million ISK worth of command charges, cap charges, and heavy missiles during the two weeks of the event. Those are the costs. Now for the rewards.

Something I failed to notice during testing on Singularity is that the "boss" NPCs granted bounties. The Guardian Angels battlecruiser paid 250,000 ISK while the battleship paid 400,000 ISK. Surprisingly, the battleships appeared 57.6% of the time. All told, I received 31,100,000 ISK in bounties during the event. Obviously, the bounties are present just to recoup ammunition costs.

The drops are where the real ISK is made. The boss NPCs dropped four types of items in high sec: 30-minute stat boosters, fireworks, SKINs, and learning accelerators. I didn't keep track of the booster drops, as I don't use them and didn't think about doing an after-event post until the Guardian's Gala completed. I'm going to estimate a 40% drop rate, as I seemed to receive either a booster or an accelerator with each event site completed. I received 5900 of the new Angels Arisen firework for a 64% drop rate. Fireworks, while cool, don't provide too much ISK.

Ship SKINs (or Super Kerr-Induced Nanocoatings) do bring in some ISK when not just activated for personal use. I received 55 unique SKINs and 88 total SKINs. At this point, I wonder if I just misplaced 4 SKINs as what I have currently listed is a 95.6% drop rate.

I'm not sure why some of the icons are different. Perhaps they were added for this year's event. But when adding the value listed above to the bounties for completing sites, the ISK/hour was only 14 million ISK/hour.

Before moving on to accelerators, I want to mention the implants that dropped in the VIP rooms. Players running the event only in high sec had access to the 5% web boost implant available as the 500 point reward. Yesterday, the implant sold for an average of 53 million ISK. That pushed my ISK/hour up to 21 million ISK/hour.

Skill points are the most valuable commodity in EVE, so the fact that the learning accelerators bring in the most ISK should not come as a surprise. I received 55 standard (+10 bonus) and advanced (+12 bonus) accelerators for a 59.8% drop rate. I also received 2 of the 72-hour Serpentis accelerators for a 2.1% drop rate. Add in the 2 advanced accelerators and 1 Serpentis accelerator I received as event rewards, and I had a lot of accelerators.

I am currently consuming my remaining advanced accelerators and plan to finish my advanced training period by consuming the Serpentis accelerators at the end. I wound up selling 15 of the standard accelerators to buy orders in Rens for 11.5 million ISK each. The additional 172 million ISK pushed my ISK/hour up to 32.5 million ISK/hour.

The additional boost to my training is the real reason I run these events. Keeping characters training with skill boosters throughout the period nets over 750,000 skill points. I am doing that for two characters, with a third joining about 10 days into the event. Since all my characters currently training are over 80 million skill points, I would have to purchase 5 large skill injectors in order to get 750,000 skill points. I figure that over all three characters I will receive approximately 13 large skill injectors worth of points. Buying 13 large skill injectors in Jita would cost 10.5 billion ISK using yesterday's average price of 808 million ISK. I don't know of too many activities in EVE where I can get that type of cash for only 15 hours of work.

For myself, I find events like the Guardian's Gala worth the time reward wise. I will say I was sick of running the sites by the end. After the fascination with the challenge of the new NPC behavior wore off, a PvE grind is a PvE grind, no matter what game I play. For the next event, though, I need to limit my ambitions. Getting all the skill points possible is hard to resist.

Friday, March 2, 2018

CCP's War On Bots and Illicit RMT: Fast And Furious Dev Blogging

"The damage the bots do is not to the game’s economy, it is their visibility and the perception that CCP isn’t doing anything about them that causes hurt. The developers know there will always be some bots to combat - 'No game has ever completely gotten rid of this problem,' CCP Guard says. 'We think of it more like a war of attrition.' Even so, CCP need to convey that they /are/ fighting them and that players who report bots are helping to clear the game of RMTs."

One of the major challenges to covering CCP's War On Bots and Illicit RMT is detecting ban waves. CCP didn't make a big deal out of them. In January, I had the surreal experience of watching a major anti-RMT/botting operation unfolding while at the same time the EVE Online sub-reddit, official forums, and even the gaming media exploded about CCP's inaction against botting. Apparently, CCP has changed its strategy, offering up the security team for interview with outlets like Kotaku and PCGamersN. CCP Guard, CCP Peligro, and CCP Grimmi even showed up on the Talking in Stations podcast to talk about their efforts.