Friday, July 18, 2014

The Butterfly Effect In EVE: Bloggers And Data

I realize that Friday's traditionally are reserved for flaming other bloggers, not praising them.  But Gevlon did a nice piece of quantitative analysis today using data from the past 18 months from Dotlan and came up with some interesting conclusions.  I would say that even Goons wouldn't argue too much with the post.  Of course, Goons are Goons and the post was written by Gevlon, so insults would fly.  That's just the nature of EVE at this point.

That got me to thinking about how Gevlon came to write that post.  He credits my Wednesday post for giving him the idea, but I started gathering data in response to a post by Sven over at EVE Lost and Found, "The State of EVE – Some Player Activity Statistics".  I thought he may have uncovered some data (I didn't realize Dotlan stored data back to 2010) that would validate a theory I have.  Turns out he didn't, but I learned something.  And a hat tip to Sugar Kyle for pointing out his blog to me.  I've added it to my blogroll.

He, in turn, was inspired to write his post due to the reaction to my post using the average concurrent users mark to note historical events in New Eden from the launch of Retribution up to the launch of Kronos.  I think the doom and gloom was getting to him.  Interestingly enough, though, he credits my post explaining how to pull the average concurrent user information from EVE Offline as the true start.  What really makes things interesting is that I wrote that piece partly in response to another post of Gevlon's, although I had considered posting the instructions for some time.  Gevlon's post just spurred me to action.  Gevlon then took the instructions and updated his post with some more cool analytic work in a way I hadn't considered before.

Just to allay fears that Gevlon and I sit in a little echo chamber discussing the merits of quantitative vs qualitative analysis, I should bring Niden into this post.  Niden's post took ideas from Sven's post (i.e. the ships killed projection for low sec in 2014 is pretty obvious) and pruned them down to concentrating on PvP.  Niden tends to focus on PvP, while I try to look at all aspects of low sec.  Then again, I'm a carebear living in low sec, so that's where my biases lay.  But I think that both Niden and I agree that low sec is a more lively place than before.  I think we just analyze the data differently.  Also, I'm a bit leery of using projections as no one really knows when CCP will choose to shake up the ant farm.

I probably should conclude this post with a chart I came up with when Blog Banter 52 was still a thing back in January.  Just because the ACU is declining doesn't mean I think subscriptions are also.

I have other reasons to believe that.  Just relying on the ACU?  Kind of weak.


  1. You should be careful when reading Gevlon's posts. More often than not, his logic is flawed in a number of ways and the data he presents is either incomplete (he often takes in only the things that support his preconceived thesis), manipulated (notice how he discards facts that prove him wrong as "irrelevant") or plain simply made up (famous wormhole wealth analysis fiasco or pretty much everything related to having fun in games). Maybe this one is indeed worthy of something, but his recent obsessions drastically skew everything he posts, so it's hard to call him trustworthy.

  2. You dont have to trust ne one if you know his sources thats the nice thing about math's and data

    1. If you know all sources are valid AND you know the interpretation is valid, you can trust that person. Gevlon is known for fabricating/manipulating data (as in "let's ignore some of the data because it proves me wrong") and for making a lot of baseless assumptions and only then trying to beat his own data to fit those assumptions.

  3. "Just to allay fears that Gevlon and I sit in a little echo chamber discussing the merits of quantitative vs qualitative analysis"
    Nobody has that fear. Gevlon has proven a number of times (including once recently) that he actually doesn't know what his data actually means. He learned how to use a couple of excel functions to make numbers appear and knows in theory what those positive and negative numbers should mean, but he doesn't actually understand what he's analysing. This causes a bit of a problem as he tend's to assume that if the data he feeds into his excel spreadsheet goes bright red or bright green that there is a definite causal relationship between those two data sets.

    For example, in the UK, traffic accidents have declined over the past few years, as has the usage of cheques. If you feed the figures into excell and build a correlation table, you will see that there's definite signs of correlation. Gevlon would take that to mean that the use of cheques causes car accidents.

    In this case, he may have stumbled onto some fact, but also may not, since the data provided alone is not enough to decide categorically if the correlation actually means the data sets are linked or if it's merely coincidental. His final conclusion seems a little far-fetched considering the only method of generating isk highsec has that is better than null is trading, which wouldn't show up on any of the figures he used.

  4. Lol in all fairness I hated the guy since he was all like I maxed out wow gold so now im gonna buy a titan and win eve. But really you check the sources and do your own math it either adds up or not. 2 of my kids have maxed wow gold now. They both suck at eve. but they are still in middle school lol
    All that said pcu does not equal subs. And lots of stuff is going on this summer game wise not to mention the stupid nice weather in the Midwest. Between that and getting into the landmark beta my play time is down as well, but I still make enough every month to keep my 4 accounts plexed, For now ne way once they hit a billion I might have to rethink my playstyle. Personally im just hoping eq next doesn't get fucked up by Sony I could use a new sandbox

  5. If, at this point, there is still anyone doesn't believe that subscriptions are declining, I don't know what to say to them.
    If there was *any* way to make it look like subs were going up, CCP would have employed it and announced an increase at fanfest. Or at any time since.

    What I would like to know is at what rate new PLEX are being injected into EVE. If it's stable, or even not much declined, then something weird is going on. If, as I rather suspect, that's falling as well, then there is Trouble At The Mill, because something like 25% of CCP's income was from PLEX the last time they mentioned any figures. As long as the PLEX are flowing and ISK-affordable, people who aren't playing will keep their subs going "for free" if only to keep that skillqueue ticking over.
    But with PLEX prices increasing inexorably, a hell of a lot less people are doing that. Anecdotally, virtually everyone I talk to is trimming down on "that one account they don't use much".

    1. I've been looking at PLEX sales on the market. Contracts are not included because I couldn't figure out how to get that information. Here's a comparison between June 2013 & June 2014.

      June 2013: 154,573
      June 2014: 123,599

      PLEX sales in the in-game markets are down 20% year-over-year.

  6. They are making more off plex cuz with the price so high more people buy them so tbh I doubt ccp cares about the multi boxers cutting back. They are still getting paid. Oh I unsigned one account cuz plex is too expensive that will show them......except some whale just bought 100plex......tbh id say 25% coming from plex sales means ccp doesn't need ur sub ne way

  7. Doesn't the analysis of EVE and NCSoft not really care about subscriptions. I.e. the number that is important is revenue, which subs are a proxy for. CCP's gaming revenue is essentially EVE and NCSoft breaks revenue out by game.