Welcome to Jester's Trek.
I'm your host, Jester. I've been an EVE Online player for about six years. One of my four mains is Ripard Teg, pictured at left. Sadly, I've succumbed to "bittervet" disease, but I'm wandering the New Eden landscape (and from time to time, the MMO landscape) in search of a cure.
You can follow along, if you want...

Wednesday, August 31, 2011

Some curves aren't

A little over a month ago, I wrote a blog post called "Not all curves are pretty".  If you take away the player guides that I write for EVE and for Global Agenda (both of which are far and away the most-read things on this blog), it's the most-viewed blog post I've ever written.  It swept through the EVE blogs, news sites, various alliance forums, and even hit a site or two in the external media.

The post itself covered some mathematical analysis of the average number of players logged into EVE Online over a five-year period starting in March 2006.  The source of the data was the API output of eve-offline.net, a website that's intended to help players determine if EVE is itself off-line or if it's just their own network connection.  Still, the site does periodic downloads of the active logged-in player count from the EVE API.  That data was available for download... right up until not long after my post hit.  These days, that part of the site says the API data is "temporarily unavailable."

That was a bit of a problem because at the end of "Curves", I stated that the trend-line for the average users logged into EVE at that time was on a sharp downward trend and that it'd be interesting to look at it in another month to see if that trend continued or stabilized.  Since then, I've gotten numerous requests for an update, an update that's been difficult due to lack of data.

Well, in the last few days, I've received the data.  This time, I'm going to keep my sources to myself, but I am convinced of their accuracy.  Both sources provided me with data, and their data not only tracks with the previous data that I used to build the previous chart, but with each other.  I have no reason to suspect at this time that they got their data from the same source, and the data looks consistent, so I'm going to go ahead and run with it.

I dropped the new data into the previous spreadsheet that I built for "Curves".  I then decided to build a new spreadsheet and normalize the data.  This involves taking out the data drop-outs and discrepancies caused by EVE patch days.  Instead, I'm going to stick to full 24-hour EVE days where full data is available (with a couple of exceptions here and there).  That didn't adjust past math too much, but did make some differences to the 2011 numbers.  My updated spreadsheet with the raw data reflecting this is also available if you want to do your own analysis.

Here's the updated "all-time" chart (click to embiggen all charts):

But that's a bit tough to see, so I also generated a chart for 2011 only:

Don't get too excited over the dramatic-looking curve.  Note that the bottom line of the X-axis is 26,000 players logged in, not zero.  Significant events that have taken place in 2011 are surprisingly easy to see on this graph, I was pleased to see.  You can see the typical post-patch dip for both Incursion Phase 3 and the "Sanctum nerf patch" released in the first week of April.  Alliance Tournament IX is easy to spot, as are the Jita riots.  It was a bit tougher to understand what happened in EVE on March 2 that caused the big spike there, until I started doing incursions myself.  After that, it was easy.

For those wondering what the percentage in EVE's growth was for +60 days after Incarna's release?  Before I normalized it, it was up a bit.  After normalization, the true number is -0.2%.  The normalized +30 day number changes from the -1.8% I published on July 20 to +0.1%.  In short, either there's been no growth from Incarna, or any growth has been neutralized by unsubs caused by this summer's dramas.

As a matter of fact, what struck me the most about the new, normalized chart, was the 30-day rolling average for the last 30 months.
  • That number on February 22, 2009: 29,113.
  • The same number on August 22, 2011, 30 months later: 29,841.

The short spike of new players after Incursion was released aside, that's a difference of 2.4%.  In 30 months.

That's not a curve.  That's a flat-line.  "Stagnant" doesn't even begin to cover this.  Bet you money, Hilmar has charts that looks very much like these two on his office wall somewhere, right now.  It's got to be one of the biggest things on his mind, if not the biggest.

Assuming the big spike in logged-in players after Incursion dropped was old EVE players coming back to check out all the changes and to spend their Learning SPs (something that's implied by a lot of forum posts from the devs, CSM Summit minutes, and other sources), those players seem to have turned right around and left again.  Incursion did not draw them back to stay.  Incarna is doing nothing to bring them back a second time.

I believe even more strongly that the winter expansion is "sink or swim" time for CCP.  It must succeed.

EDIT (31/Aug/2010): A couple of comments have already come in saying "2.4% isn't flat!"  I wouldn't normally respond to that level of coherent thought, but this kind of comment really deserves special attention.  So just for the people who don't think 2.4% over 30 months isn't flat, here's a graph of 2.4% growth over 30 months:

Decide for yourself.

EDIT (31/Aug/2010): Another version, for those who want to accuse me of manipulating the first graph.  Same data, going back to the beginning, 30-day rolling average, six month slices.  You tell me where it's flat and where it isn't.

Again, decide for yourself.  But if you want to argue with this one, you might want to join me in a place I call "reality."


  1. you got numbers,
    you got graphs,
    still you manage to make an interpretation that is hardly making sense.

    first of all, in the overall graph you can see that during the summer periods there has allways been a low, and that it allways goes up after wards.
    also if you compare last year to this year, you see that it is still more.

    while your try to tie this to patch days and make a black prognosis to ccp based on it has no grounds.

    also, for ccp the relevant number is the number of subscribed accounts not the number of average concurrent users logged in.

    looking at the overall tone of you towards ccp, in this and other posts it occurs to me that your missinterpration is deliberate, unfortunately this kind of writing (as seen on en24, your blog, and a few others in the blogscene) seems to what creates users for the blogs.

  2. I think with the amount of scandal caused by incursion and the amount of people saying they would cancel to still be 2.4% up is a decent achievement.

    You can anaylse and report on stats in different ways, I think there are thousands of companies who would welcome a small increase during a global recession.....

  3. I'm sorry, but the factual data just don't support your position. I keep hearing about this "summer slump." OK, let's look at 2009-2010.

    30-day rolling average on 10/1/2010: 30124
    30-day rolling average on 07/1/2010: 32770
    30-day rolling average on 04/1/2010: 30895
    30-day rolling average on 01/1/2010: 31522
    30-day rolling average on 10/1/2009: 28972
    30-day rolling average on 07/1/2009: 30341
    30-day rolling average on 04/1/2009: 30973
    30-day rolling average on 01/1/2009: 26220

    Where's your summer slump again? Because it seems to be taking place in the winter.

    And as I've said already, the data clearly show that from summer 2009 to summer 2010 to summer 2011, there aren't "more" players. The numbers are completely flat.

    Can you provide actual numbers in support of your position? I've given you access to my raw data.

  4. 2.4% is more. How is that flat ?

  5. And just to be clear, the slump in summer 2011 wasn't a slump, it was the downhill slope of the spike caused by Incursion. The first chart in this blog entry clearly shows this.

  6. For those who say "2.4% isn't flat!" I've added a third graph to this blog entry... just for you.

  7. 2.4% is growth, however you want to try to picture it.

    If it was a 2.4% reduction Im sure you would be saying it was a drop....

  8. There is a global recession at present. Most businesses are struggling over the past two years just to stay afloat.

    To increase the amount of customers by 2.4% under such conditions, in an extremely competitive market is not a decline, which is how some people seem to portray Eve.

    Its growing, maybe not as quickly as some would like, but its still growing.

  9. That third graph is using a scale of 0-100,000.

    No wonder a 2.4% increase is completely flat. On a scale of 100,000 an increase of 2.4% of 50,000 is halved on the graph.

    Whilst it still wont look drastic increase if you used a scale of 40,000 - 55,000 it wouldnt be so flat.

    Stats and graphs can be presented in different ways to tell different stories.

  10. Strictly speaking, we do see a decrease in players online during the summer and to a lesser extent in the fall as compared with the winter and spring, but this might be due to (1) CCP's expansion release schedule and (2) the exact definition of "summer" and "spring" used in the analysis. Whether there was a true "slump" in players online in the summer of 2009 is also questionable, since on 22 June 2009 CCP banned approximately 6000 botting accounts; the number of players online dropped by approximately 3000 between 22 June and 23 June, which would account for most of the decline in players online that summer. Consequently, we could well conclude that the first true summer "slump" occurred in the summer of 2010, following Tyrranis.

    Regardless of whether such a thing exits, and what its causes are, historically the number of players online begins to increase approximately one to two months prior to a major expansion, and crests approximately one to two months after the expansion is released, regardless of season. This provides a good explanation for the 2010 summer "slump," where the summer expansion was released in May 2010, and the decline in players did not occur until July.

    Incarna is the first expansion in three years where this pattern has not occurred. There has been absolutely no increase in players online prior to expansion, and no growth in players online following the expansion.

  11. Apologies for the comment spam, but would like to also note that the number of players online has been a good proxy for the number of active accounts in the past. As a matter of common sense, a person will usually not keep paying a subscription for a service he does not use for long; this is especially true for non-recurring subscriptions, such as accounts funded by PLEX. Furthermore, the number of players online is a good indicator for the health of the game as a whole. If fewer players are logging on, then the game is no longer fun for those players, and they may well find something else to do with their time and money. The danger, which EVE is not yet facing, is that eventually the decrease in player activity will become self-sustaining. To quote Kirith Kodachi:

    "[T]he bitter old vet disease took control of the corp leadership and took the wind out of the sails, going into a spiral of not enough people logging in because no fleets were up because no one was leading them because no one was logging in."

    By the looks of it, the players online count has stabilized over the past month. Hopefully CCP will produce an expansion which will be able to attract and retain new players.

    Also, 2.4% growth over two years is essentially meaningless both in absolute and in relative terms. Past expansions have shown much higher permanent growth in players online in just a few months.

    The recession doesn't have much to do with it in my experience anyways. Some of the folk in the corp I'm in have been unemployed for months due to the recession -- they pay for their one or two EVE accounts via PLEX. Nor is a stable subscriber base particularly good news if CCP has been expanding its operations in reliance on growing subscriber numbers.

  12. "Also, 2.4% growth over two years is essentially meaningless both in absolute and in relative terms. Past expansions have shown much higher permanent growth in players online in just a few months."

    I'm not sure in which world you're living in, Katsuko, but if there were "much higher permanent growth" there would be more than 2.4% increase in player numbers over this period of time.

    I expect that once "Establishments" come out and people can interact in-station, subscriber numbers will pick up significantly. I'll be surprised if it's less than 5% higher EVE Online subscribers on 30 day rolling averages six months after the multiplayer aspects of Incarna come out. The shorter term will no doubt see the usual post-expansion bulge as people come back to check out the new content, then gradually unsubscribe as they realise that computer games can't substitute for meaningful sex.

    Sure, it's just "sims in space" or "space barbies" at this point in time, but I'm expecting that a significant number subscribers be logged in simply because they want to use EVE as an IMVU substitute, with flying-in-space being a novelty aspect.

  13. All I'm seeing here is a man who created a blog to run for a spot on an elected council for a video game.


    Rage quit.

    Couldn't stay away and now badmouths the game at every turn.

    That about sum it up?

  14. @Mara Rinn: Not entirely sure what you're trying to say here, and apologies since my comments are not always entirely coherent. To clarify: I'm arguing that 2.4% growth in players online over an entire two years' worth of fairly expensive expansions in terms of development resources isn't something to be proud of given CCP's business model, and might very well be simply noise in the data. The comment was aimed at the anonymous folk who've been saying that 2.4% is good or respectable. It's not.

    What we're seeing between 2009 and 2011 are sharp spikes in player counts around expansion releases, but the long-term player activity has been basically stagnant for two years aside from what might be a slight upward trend. For comparison, between 2006 and 2008 the number of players online increased by approximately 50% (it's a bit too late here for me to do anything sensible with Excel/SPSS at the moment, so apologies for the lack of precise numbers), ignoring expansion-produced spikes. Between 2008 and 2009 the number of players online increased by approximately 20%, ignoring the post-Apocrypha spike. Arguably that number would be higher if CCP had not banned around 6,000 accounts that summer. Since then, however, the number of players online has barely increased in real terms. The post-Incursion increase in players online topped out at approximately 35,000 as measured by the 30-day moving average; it is now back to essentially where it was prior to Incursions. So there has been no meaningful growth in player activity, and by extension in long-term players, for almost two years.

    As far as Establishments attracting new players, I sincerely hope you're right. Sadly, given the current state of CQ and CCP's history of abandoning features, I am not terribly optimistic. I'm also not convinced that players will pay $15 per month for a glorified and extremely resource-intensive chat room that can't even be customized. The pretty graphics might get prospective players interested in the game, but it will require either meaningful gameplay or strong social connections to retain those players. I am not hostile to Incarna, but realistically the odds are against CCP pulling off something spectacular this coming winter.

  15. Katsuko, when talking about growth of the user base the recession is very relevant as new players DO have to pay subs.

  16. @katsuko, the banning is an interesting point. If 6000 accounts were banned thats a massive reduction in users. They have also been banning people regularly recently.

    Therefore a 2.4% increase has to replace those banned and is therefore a big increase on new players who would replace the plex paying botters.

    2.4% overall may not be a massive number, but with the botters banned the actual increase of paying players could be 10%

    Again, its easy to look at stats and make a blanket statement, but its not as straightforward as 2.4% growth isnt enough.

  17. Remember, you can see the Unholy Rage dip in active logged-in players on these charts. It takes place in summer 2009.

    Ironically, it's easiest to see in the sarcastic "Where is it flat?" chart in the second edit. Unholy Rage causes the dip that takes place right before the "definitely not flat" part. But it's also pretty easy to see in the first chart, where the pink line (representing the 30-day rolling average) bottoms out a few months after the release of Apocrypha.

  18. Well done M8t .... Keep up the good work! I think its time CCP got a realty check... and hope it does not bounce. I did not renew my 3 accounts today and they on 6 monthly sub cycles.

    Well doen HILLMAAAAR !

  19. Anonymous @ 1152

    While banning 6k botters would result in an immediate hit to the concurrent numbers... those botters would simply return, with new account. Now it's worth noting that while training a new Hulk pilot, or ratting Tengu you'd have less reason to log in on a day-to-day basis until you had basic skills trained... but please don't imagine any serious RMT bot controllers would simply throw in the towel after being banned once every few years.

  20. I'm having trouble seeing the interpretations that some are making from these graphs. These graphs show player activity and I don't think you can make any kind of firm link between this activity and subscription levels or CCP income. I think that the number of players online is an indicator of the level of interest, but not of subscription rates. It's as possible that the peaks of activity are due as much to existing players taking a greater interest in the game as it is new subscriptions. Is the activity increase due to idle players getting active? Players reactivating unsubbed accounts? Or new subscriptions? Or active players just spending more time in-game?

    Consider the peak at the Outartoh Sansha LE - there's a big jump in the 7-day average there. I very much doubt that many new subs happened at that time just for that event. It's much more likely that existing players just got into it and spent more time in-game than they did before or after.

    People spent a whole lot more time playing after Quantum Rise, but are those new subs? You can't tell. I'd be surprised if there weren't a bunch of new subs in there, but it could be low level players finding reasons to play more? Or a combination. And why did activity drop shortly after Apocrypha? Did Apocrypha do it, or was it just the effect of Quantum Rise finally wearing off and players returning to a more sedate activity levels?

    Note the jump in activity during the Jita Riots, but we know, or think we know, that a significant number of people were unsubbing in protest at the same time. Activity up but subs down at the same time.

    Hilmar may well have these figures on his desk, but he'll be paying much more attention to the subscription numbers.

  21. You make an excellent point, but the counter-point to it is the drop in activity after the Incursion spike.

    People who log in to look at your game, play for a month, then log out and never log back in again? That's bad news for you whether they cancel their subscriptions or not.

    Right now, we don't have any way of knowing what the subscription numbers are because CCP is keeping those to themselves, except for brief flashes.

  22. Well Doen Jester .. for saying the hard stuff. Respect to you for take the flack in a good spirit.

    But your right i have noticed some key players leaving or taking a break. I dont think EvE is going to fall over but in the last 8 months something has definitly gone wrong for CCP. That figures/Graphs are not good even at 2.4% growth, considering cost of labour increasing and globle inflation. We taking 30 months(2.5 years)... thats not good.

    My 3 subs (of which 2 are 6 monthly) will be due for renewal over the next 3 weeks and i really dont intend to renew them.

    Once again Respect ...!

  23. I am not hostile at all to your points and I thank you for taking the time to put the data out. I'm afraid however it still doesn't tell us enough of the post Incarna story

    Incarna's CQ is incredibly resource intensive, in a way that those that did have CQ loaded would have quite a bit more reason to:

    _ close their client when not actively usin it, perhaps while playing another game or doing rl work in another application
    - stay logged in to fewer clients

    I do not have cq active on any of my 5 accounts... but for good reason or not, I feel like the program still uses more resources while docked even with CQ disabled.

    Because i know little of such things and have heard of cpu usage spikes, the idea of leavling mutiple clients logged in just to periodically look into market prices doesn't seem like a good idea anymore (besides, I'm mostly taking a break,,,even forgetting to log in to keep my skill queues running unstopped, which is a sign of true distancing from the game... I've got isk to pay for 5 acounts for 5 years and make more isk a mont logging in for 10 minutes per acount a week than it takes to pay for them ... that is a problem in itself for ccp in some shape or form in some ways, but it also makes me a big plex sink for them with a tiny amount of cost for them per account I keep active -- no forgone revenue and tiny variable data costs)

  24. You have to look at the data a little more granularly as well, as long as you're also looking at the big picture.

    The daily average is dropping, but the daily high isn't so much.

    What is happening is that US prime time has MUCH less players, EVE is losing the interest of many US gamers during their prime gaming time.

    While Euro primetime is easily at 45k+, US is at around 26k+, US timezones are dropping. Can you look at a cross section of US average hours (0100 UTC - 0500 UTC) from 12, 6, and 3 months ago to today?

  25. This data really worries me. We can only hope that CCP is honest and frank with itself and what it want to do about the current reality. This is no time for group think. I continue to hold out hope that this period will be the input-us for a resurgent CCP and Eve online.

  26. @Anon1007: no, you're not wrong. I have CQ turned off on all four of my accounts, but The Door by itself still consumes a ton of resources.

    That said, you can still reduce the resource usage to near-zero by minimizing the client. I doubt very many people are logging out of EVE over it.

  27. Anonymous said...
    you got numbers,
    you got graphs,
    still you manage to make an interpretation that is hardly making sense.

    first of all, in the overall graph you can see that during the summer periods there has allways been a low, and that it allways goes up after wards.
    also if you compare last year to this year, you see that it is still more.

    while your try to tie this to patch days and make a black prognosis to ccp based on it has no grounds.

    also, for ccp the relevant number is the number of subscribed accounts not the number of average concurrent users logged in.

    looking at the overall tone of you towards ccp, in this and other posts it occurs to me that your missinterpration is deliberate, unfortunately this kind of writing (as seen on en24, your blog, and a few others in the blogscene) seems to what creates users for the blogs.

    come on hilmar could you make it a little less obvious when you try to shout down people who back up their statements with facts. i get that your worried you'll be out of a job soon but seriously dude if you want to make everyone not believe the well thought out, spell checked fact backed statement in this blog you have to do better than point your finger at it and wish it away.

  28. Looking at your graphs, my only quibble is that the area labeled as the Sanctum nerf should really be labeled the mining bot ban, since the big wave of bans took place on April 9-10, which was just ahead of the Sanctum nerf. And for those that point out the 6000 bans, only about 300 of those accounts were eventually perma banned. That's why there was a big increase in PCU about one month later.

    And for those arguing that the daily PCU hasn't taken a hit, that's incorrect. A rolling 7 day average of daily peak concurrent users taken from Eve Offline showed a 10% decrease from July 31st to August 31st. For me, the scary part is that from the release of the dev blog about the roadmap for null sec on August 15th to the end of August saw the numbers drop over 7%. I'm really hoping that is a seasonal number and not a reaction to the dev blog.

  29. Proof that accounts were banned in April 2011, please. The only large-scale banning that I know about took place in summer 2009.

  30. Jester, growth of companies or assets is understood as being exponential. Therefore you should really plot the graph not on a linear scale (as it is), but on - maybe additionally - on a logarithmic scale.

    As comparison, it is said that a healthy company should grow 8% per year. In the five years that EVE Offline has recorded, growth has been 10.8% p.a. - but that is not evenly distributed as your new graph will show.

  31. @Tsu: the "flat?" chart plots from 29,113 to 29,841, the average number of EVE players, 30 months apart.

  32. Jester, frankly speaking I do not understand your response to my comment. ;-)

    I too believe, that the number of active players is not as high as it should be. And, that Incarna etc. didn't help to raise it.

    In the meanwhile I have used your work and data to create a, in my opinion, nicer graph with Google Chart. You can find it here:

  33. Wondering how many defending the numbers as good are ccp trolls.

    Well no I'm not. Statistical analysis is a funny thing, spin doctors love to ride the numbers in their direction, argueing over it will never stop.

    That being said there is an appearent greater amount of discontent for every generation of eve player ( generally thinking of generations in terms of year you started )... as we can only percieve what we have experienced and therefore those who have been around longer and stayed in the "know" may realize this is serious, and those who have not may think this is monumental and unheard of ( because they havn't heard of it being so bad before...they were not there. )

    Eve will live or die, arguments will continue, the fact is we as a player base have opinions and Eve has given us a lot of control over our environment but it is still CCP's game and they can do what they want, how they want and can very well die off like any other gaming company that existed...what will be sad is because this game depends on them being around to run servers their legacy will be lost when they do die unlike the games of old that you could keep/copy and run on your own.

    I think they have legs still and so far while I'm not liking some of the things they are changing, it is their game to change...and if it no longer becomes something I like...there are many more fish in the sea.


  34. Please use decent axes on your graphs. To take the example of your last plot the y axis should go from 15,000 to 30,000 give or take a bit. All that whitespace smacks of manipulation and if anything better axes would actually help prove your point. Its also good practice to make the data points you are plotting more obvious so the trend lines look less arbitrary.

  35. Wow. Just wow. I can't believe anybody would believe that 2.4% over nearly 3-years represents "growth". Do I really share space with such retards?

    Any number above 0% is growth - congratulations - go get yourself a merit badge. However, 2.4% over 3-years represents stagnant growth. For you slow kids in the back of the class, it simply means you're going nowhere slowly.

    It's actually worse than a drop.

    It fools you 'tards (and CCP investors) into thinking everything is just spiffy, when in fact, it stinks to high heaven.

    Wake up and get educated. 2.4% is a joke.

  36. Very interesting, and I applaud you for the legwork behind this.

    If you are trying to make the point that the numbers are influenced by the economic environment, might I suggest normalizing the numbers to some appropriate economic indicator? I'm not even sure what this would be, others will know better than I. Maybe normalizing to some entertainment sector indicator would reinforce your argument.

    While I agree with some of the rationale behind making the y axis logarithmic, it may make the features harder to see. Usually when I'm begging for log scale is when ratios of highs to lows get very high, or there's a clear exponential dependence.

    Thanks again and good luck.

  37. In response to Jester's comment that "you can reduce CPU usage by minimising the client" - you can't fly a ship with the client minimised. EVE Online still consumes memory which isn't swapped to disk when it is minimised.

    For my purposes, I log in, the client is active for a while, and I quit it as soon as I can in order to reclaim resources for other processes on my computer. I only have 3G of RAM, and with Incarna I can now only run one client on this machine. I'm lucky that I can run 1 client, but thanks to Transgaming my X1600 card is now unofficially supported.

    Turning off CQ only goes so far. If I zoom out while flying in space (zoomed out to the point that the entire grid is one hundred pixels, thus no objects are being rendered) I can run two clients, but the moment one of the docks, the other can't run.

    I doubt I'm the only one running a minimum spec computer.

    Aside from the apparent drop in subscribed players that might be attributed to people logging out to keep their computers cool, I wonder if EVE has simply attracted as many people as a persistent space combat MMO with the highest death penalty in the MMO market is ever going to attract?

    I'll continue waving the flag for space barbies until a few weeks after establishments are launched.

    I wonder if the losses caused by Empyrean Age and Tyrannis can be reversed by properly implementing Faction Warfare and Planetary Interaction? I wonder how many more players will log in and play when a suitable model for null sec sovereignty is implemented?


Note: Only a member of this blog may post a comment.