Pages

Friday, June 13, 2014

Explaining Chribba's Famous Graph

In a game in which players love graphs, perhaps the most famous is found on one of Chribba's many websites, Eve-Offline.net.  Many people, myself included, have used the weekly average concurrent users graph to interpret how players respond to events in EVE, or just to give a general sense of the health of the game.  But I'm not sure how many posts exist discussing the graph itself.

Weekly Average Concurrent Users For Tranquility From Eve-Offline.net
Each server cluster that CCP runs has its own API that Chribba uses to display the concurrent number of users.  These end points only supply the numbers for that cluster and no others.  Chribba not only displays the number of current users logged in, but the number of peak concurrent users over the past day, week, month, 3 months, and 6 months.  And, of course the average concurrent user graph.

One fallacy that has emerged, due to some shenanigans performed by CCP in regards to the user numbers displayed in DUST 514, is that the current user number displayed for Tranquility on Eve-Offline.net combines EVE and DUST 514 together.  I checked with Chribba and that is not the case.  Also, checking the data at the time DUST 514 went on-line confirmed that.  The average number of concurrent users did not magically jump by thousands one week.

The above graph shows the weekly average number of concurrent users on Tranquility since March 2006.  Each of the data points, as far as I can tell, represents the midpoint, not the start, of each week.  The last data point in the current graph is 5 June 2014, but the next update won't occur until Monday, when the 12 June 2014 data point is added.

Now, I know what some people reading this are thinking.  Do do they get their hands on this data?  The people who love a challenge will probably go out and try to get the data from the EVE API.  I'm not sure that really works, as I don't know how much historical data is available.  I'm pretty sure the weekly average is calculated and stored locally in one of his databases. So instead of reinventing the wheel, I always use the data in the graph.

Getting the raw data is easy.  Just right-click on the page, choose View Page Source (or whatever option your browser presents), and search on the string, "2006-03-09", the first week for which Chribba has data.  The page should look something like this...

An Example Using Firefox
Then just take the data and format as desired.  I copy the data from the browser into a text file and use a VBScript file I wrote to format the file so I can easily copy/paste the data into Excel. Or just use something better.  I use VBScript because it comes pre-installed with the Windows operating system (including Windows 8).  Hopefully I still have the script.

That's a brief overview of the famous graph and the data it displays.  I don't know if anyone is actually interested, but I figure I'd write a post explaining how I get my average concurrent user data.  And for those wishing to use the data, please credit Chribba, OMG Labs, or EVE-Offline.net.  I usually credit all three.

10 comments:

  1. Going to be interesting watching that thing spike in late July and early August, then melt away over the subsequent months.

    I read Danny Centauri's May 13th blog again that summarized the industry changes. I cannot fathom, under any set of circumstances, how CCP believes these changes will improve PCU and subs.

    ReplyDelete
    Replies
    1. All of the accounts I cancelled are still cancelled. I hope they got the message.

      Delete
    2. CCP has never said that the industry changes will improve PCU or subs. In fact, the seniors do expect high-sec industry-only account subs to drop, due to the usual player dissatisfaction whenever a large change is made to an old part of the game.

      However, CCP's recent house-cleaning of unnecessary staff has left the company in good shape, and any loss in subs is expected to be easily absorbed.

      Delete
  2. Thanks for the tip. I updated my today post, replacing the screenshot of the graph with an analyzed one: http://greedygoblin.blogspot.hu/2014/06/the-graph-that-didnt-shrink.html

    ReplyDelete
    Replies
    1. That new graph definitely looks better and supports your argument more effectively.

      Delete
  3. No idea on the concurrency numbers but there's no doubt DUST players are counted in the map statistics "docked and active" system user counts.

    That could lead one to assume they're counted in the concurrency numbers too.

    ReplyDelete
    Replies
    1. That's not the pcu count tho. Dust isn't in cludes in that as has been confirmed by ccp over and over

      Delete
  4. You do know that CCP began adding the DUST players to Tranqulity during the closed and open beta periods, which predate the availability of the DUST API, don't you? This is why there is no unusual spike. When DUST offcially launched, most of the players already had open beta accounts.

    And, CCP devs have confirmed that DUST players have "accounts" on Tranquility - this is how they share local with EVE players in certain systems (it is also why they are always stuck in their original NPC home system local forever on Tranquility, no matter what system they really are in on the DUST server).

    ReplyDelete
    Replies
    1. Really? The last I heard, the user numbers displayed in the DUST 514 client are higher than the ones displayed by the EVE client. The numbers displayed in the DUST 514 client are indeed Tranquility + DUST, but it does not work the other way around. This was confirmed by CCP.

      If you would like to convince me otherwise, please send me the proof.

      Delete
  5. Not sure how you could think any sound business manager makes the staffing cuts ccp just made and then says there ok to absorb a subs reduction for a game change that does not improve the game. I guess that's why the day the lights go out in eve is getting closer and closer....

    ReplyDelete