Friday, July 11, 2014

The State of EVE – Some Player Activity Statistics

By some accounts, EVE is not in good shape. Some people are worried about the recent drop in player subscriptions and a drop in concurrent online users, specifically, and others are worried about perceived "stagnation" in various parts of the game, such as null sec. The Nosy Gamer kicked off most of the recent discussions about the state of EVE with his post, first, covering the average number of players online, and then in his more detailed looks at those numbers, the latter of which got picked up by The

Episode 45 of the Crossing Zebras Podcast is a good example of the doom-and-gloom response these numbers have generated for some (surprising since, by their own admission, the guests on the episode haven't been playing much EVE lately--but not for lack of content, they explain, as their null alliances have had frequent OPs lately, which they have simply not wanted to participate in because they are not big or newsworthy. Rather, my perception is that these guests are the sort of players who at least currently wait for content to generate, not the type to create it for themselves or others).

But player subscription numbers, online users, and personal assessments of the state of the game (or null sec, etc.) are not the fully story. What about player activity? What about spaceship content? 60,000 players logged in at one time--a number surpassed a few times in 2013--is mostly irrelevant if those players are in station and not engaged in any actual gameplay and if they deem no activities worth pursuing besides skill training. Is there any way to gauge player activity? 

In what follows, I cobble together some stats taken from Dotlan--stats on ship kills, NPC kills, and jumps, by type of space. I compare 2012 with 2013, and then I compare 2013 with the first half of 2014 projected over the rest of the year (I'll explain this more below). 2013 saw massive improvements in every category on 2012, as many players would expect, but 2014 does not appear to be such a bad year after all--at least thus far. In fact, low sec and, to an extent, null sec, seem better than 2013 in terms of player activity. Obviously the numbers are limited, and ignore important fine-grained details such as the fact that the first 3 months of 2014 were likely better in most categories than the next 3 months, consistent with the numbers on average concurrent users. Nonetheless, they paint a fairly positive picture of the state of EVE.

2012-2013: Ship kills, NPC kills, and jumps by region

To recap the past few years, 2012 saw the release of Inferno and Retribution. EVE was on an upward trend after CCP had learned valuable lessons from Incarna etc. etc., with development focusing on fundamental gameplay mechanics, particularly just getting people out into space. 2013 continued this trend with Odyssey and Rubicon, while also started to lay the groundwork for future big features. 2013 is often considered one of, if not the best year for the game, with player subscriptions reaching all time highs among other news-worthy statistics, such as largest battles (by player count) and most players logged in at one time, not to mention the complete revitalization of low sec, etc.

With the first half of 2014 came Kronos, which I predicted would be a fantastic expansion for thehealth of low security space. Kronos marked the end of the traditional expansion cycle (for now at least), preparing the way for the last half of 2014 to bring four separate releases—Crius in July, Hyperion (Sep), Oceanus (Nov), and Phoebe (Dec). However, the first half of 2014 also experienced a decline in subscription numbers. Let's start by looking at how 2012 compared to 2013:

Alas, no graphs, but what's important is pretty clearly communicated in the changes from 2012 to 2013. Start with ship kills. 2012 was a good year compared to the years prior, yet 2013 blew it out of the water. High sec and null sec both saw a small increase in ships killed, but low security space (literally) exploded. In 2012 and, really, ever year prior, if you wanted fights your best bet was to avoid or pass right through low sec and head into null, which even in 2012 saw quite a bit more action. From 2013 on, though, that fundamentally changed, with low sec not only completely overtaking null in ships killed, but edging closer to high sec than null ever did. Wormhole space also saw a decent increase in kills, hurray! These numbers, though, aren't surprising. Retribution and Odyssey both did more for the state of low sec than any expansion before them--and probably more than every expansion together. Rubicon helped, too, but also helped null with the warp speed changes and interceptors specifically.

I also included total ship kills in the top ten systems to see if the increases in ship kills were due to popular systems getting (much) more popular, or if the fighting was more spread out. It turns out that the top systems in every type of space improved, becoming much more active specifically in low sec, but not enough to account for the total changes in ship kills. I also checked the top low sec systems relative to the total ship kills in low sec. I excepted the rapid growth of Brave Newbies would concentrate PVP in low sec more than it was in 2012, but it turned out that the top systems in low sec still only accounted for a little over 5% of the total low sec PVP--a fantastic sign of the health of low sec I think, as it suggests that the activity remained spread out.

Now let's look at NPC kills (Dotlan classifies this as faction kills, which I assume are NPC kills). Interestingly, even though 2013 experienced growth in terms of player subscription as well as massive growth in terms of ship kills, NPC kills on the whole declined in every type of space. I don't know enough about PVE in EVE to explain this, but a few causes standout to me: First and most obviously, players PVP'ed more and PVE'ed less. So these declines are actually a good sign in some ways. Second, some changes would account for less NPCs being killed--specifically, Odyssey removed NPCs in (the old) relic/data sites. I am not sure if this is enough to entirely account for the change.

Finally, what about jumps? Jumps indicate players are traveling in space, regardless of what content they pursue. The number of jumps in high sec stayed almost exactly the same, null declined, while low sec of course grew nicely. The obvious message is that high sec stayed balanced with the new players coming in replacing some older players leaving the game or leaving for low sec. In terms of what areas of the game actually felt that growth from 2012 to 2013, then, it was low sec.

2013-2014: Ship kills, NPC kills, and jumps by region

This much is all likely expected. But what about the transition from 2013 into 2014? We are only 6 months into 2014, and massive changes are on the way both for the game as well as for the way content is introduced to the game, via the 10-release model that replaces the expansion cycle. So, I can only speculate on what the next 6 months will bring. So, my comparisons will assume that the statistics of player activity during the second half of 2014 will be roughly like that of the first. Given the negative assessment some players have about the first half of 2014, this would, presumably, be quite bad indeed. To calculate these projected numbers I simply took the Dotlan stats for the first 6 months of EVE and projected them over the next 6 months (they include the full month of June but not the first week and a half of July), the first column being the previous statistics from 2013, the second the projected 2014 statistics:

The tl;dr version is that the first half of 2014 (and thus the projected state of EVE for the full year) is no where near as bad as some people are describing, at least in terms of player activities in game. Let's look at the comparisons in more detail, starting with ship kills. High sec space and wormhole space both experienced small declines, with high sec ship kills returning closer to its state in 2012. Wormhole space gained so much momentum in 2013, though, that the (projected) -6.66% decline in ship kills only negates a small protion of the gains in 2013. And that's the bad news.

Now for the good news: Low sec, amazingly, is even better in 2014 than in 2013! In fact, low sec comes within  few hundred thousand kills of high sec. Given the amount of ship kills that, historically, occur in high sec, that's pretty crazy and a great sign of how healthy and active low sec is. Now for null: apparently, null does even better in terms of ship kills in 2014 than in 2013. That's a surprise, given that the EVE news sites and blogs rarely go a week without posting about the "stagnation" of null. There is clearly a stagnation of news about content in null (rather a self-fulfilling prophecy for news sites, given that they control this fact) yet not a stagnation of fights. Keep in mind that in March of 2014 Braves Newbies and the HERO. coalition moved largely to null sec, out of low. This helps explain the continued positive growth of ship kills in null sec, and it makes it even more remarkable that ship kills increased in low sec when one of the biggest groups largely left. Total ship kills stay roughly the same, which means people are fighting less in high sec and more and more in low sec and, to an important but lesser extent, in null. So much for the percieved stagnation.

What about the other two categories? They mostly continue the trends of 2013--except for one important case: NPC kills in low sec. While no where near the number of NPC kills in high (which continued to decline), NPC kills in low rose by nearly a quarter in 2014. That means right around 25% more NPCs being killed, which means more people in low sec, in space, killing NPCs of various sorts--that is a pretty massive change. NPC kills also slightly rose in null, likely due to the "cold war" status of the region and the increased use of afk-tars. Interestingly as well, jumps in low sec continued to rise, while jumps in null also reversed the 2012 to 2013 trend and increased modestly.

Some Final Thought: Why the decline in the ACU? 

In sum, 2014 is not a downward spiral of stagnation and despair. In some areas, 2014 saw a decline from 2013 (e.g., NPC kills) but many of these trends were present in the 2012-2013 period and many of them are likely positive signs of the game's health. In fact, however, in terms of player activity, 2014 is in most cases about as good as 2013, and in some cases even better.

This makes the drop in the ACU and, presumably, in player subscriptions all the more surprising. I don't have a fully explanation obviously, but I do have three important pieces to the puzzle:

1) There might be roughly the same number of flesh and blood players in EVE in 2014 as there were in 2013, but less newer players (as evidenced by the drop in high sec ship kills and NPC kills as well as from public numbers) and, notably, less alt accounts, which translates to less alts logged in at the same time or one after another, if only to train skills. The decline in alt accounts is likely due to two factors: The continued increase of PLEX prices, and the new ability to train multiple characters on one account. This has led many people, it seems, to close an alt account or two, but, more importantly, it also signals that EVE has not actually lost very many individual players.

2) Players seem to be doing more when logged in. Player subscriptions are down, and concurrent number of players online are down, but some forms of activity, such as low and null sec ship kills, are up. This suggests that players are spending less time logged while pursuing no content--such as sitting afk in stations. Instead, players seem to be doing more while logged in for less time. This might indicate a positive improvement, that players are able to better maximize the enjoyable parts of the game compared with their play time in 2012 or 2013. 

3) Regression toward the mean. 2013 was, by most accounts, the best year of the game's history. When athletes have a banner year and then experience a decline the following year, thousands of new pages and TV time are spent explaining it in rigorous detail when, in the vast majority of cases, their overall abilities have not declined but their averages have regressed to a more accurate status. That, perhaps, is what is going on with EVE in the decline from 2013-2014 in stats like concurrent users online and new player subscriptions, etc. 2013 was an amazing year, and it is probably too much to expect CCP (or the players who generate content) to surpass or even match that. Instead, it would be much more realistic to hope for a continued, small but lasting growth in important statistics.

And that is roughly what we see thus far in 2014. Low sec is better than ever. So here's hoping CCP can do for the other areas of space what they did for low, because it has been a massive success.


  1. part of the drop in subs i think is the active players are using fewer alts. i stopped using five alts and shrunk the number to two. part of the motivation was high plex prices... but also ships were redesigned esp. mining ships and multiple ships were not necessary anymore, esp. orcas when i was mining. a decline in missioning is visible. there is no longer any need for standings to anchor a pos in hi sec. just some ideas anyway

  2. It is difficult to take your 2013-2014 comparisons seriously when your 2014 projection has no valid basis (e.g., you cannot find any year using any known metric where months 1-6 resemble months 7-12). Maybe you could follow up with a comparison of first half of 2012 to 2013 to 2014?

  3. @mordis: Even that wouldn't help. Every analysis I've seen over the last few days of player activity data has been based on a flawed methodology, including this one. All of them, including this one, ignore that February 2014 was one of the most active months in the history of EVE, something that is simply not going to be repeated in the rest of 2014. And yet every analysis just assumes through their methodologies that it will.

    You could get closer to a full-year analysis by taking the February data out completely, finding a average for the remaining five months, multiplying *that* number by 11, and then adding the February data back in. Even that wouldn't be accurate (because activity this year is reducing month by month on a linear scale) but it'd at least be closer.