Wednesday, July 16, 2014

The Game Of Graphs: What's Happening In Low Sec?

EVE players are known for their love of graphs and charts.  My post from last Monday that used a graph of the average concurrent users matched up with in-game events proved pretty popular and created some discussions.  But one of the talking points I've read in a couple of places leaves me a bit puzzled.
I'll use Niden's article, "EVE Is Dying",  on Crossing Zebras as the example, as that article is the most prominent I've read so far.  His argument is that whatever malaise has fallen upon EVE has missed low sec.  I won't argue against that premise as low sec certainly feels busier. A PvPer has chased me out of my normal ice mining system and my chosen backup system has too many Procurer and Skiff kills posted on the killboards for comfort.  But that is anecdotal evidence and I want hard numbers to back up my gut feeling.  But I don't think that Niden and others have provided those numbers.

Looking at Niden's numbers, he came up with a projected number of ships killed in low sec for 2014 that is based on the second half of 2014 performing exactly as the first.  He also did a comparison of the number of ships killed between low sec and null sec by year from 2011 through 2014, using the projection to argue that the number of ship kills keeps going up.

The 2014 projected totals really bothered me and so I started running numbers on my own a few days before Niden's article was published on CZ.  But instead of using the yearly numbers, I used the monthly totals and created the below graph.

I combined the monthly ship kill data from Dotlan with the average concurrent user (ACU) data from Chribba's  I converted the monthly totals from Dotlan to daily totals in order to do a proper comparison between months with varying numbers of days.  I included four graph lines for ship kills: one each for high sec, low sec, null sec, and the total number of ships killed in known space.  Wormhole totals are not included due to the EVE Online API no longer giving that information starting in June 2014.  But as many of the comparisons I read were between low sec and null sec, I don't see the omission causing an issue.

I need to point out two items that could cause discrepancies in the results. First, I had to massage the data from EVE Offline to make the weekly data fit into calendar months in order to match up with the monthly data from Dotlan.  The second is that the numbers from Dotlan are not 100% accurate due to unexpected downtimes of CCP's API service and maintenance down times of Dotlan.  But after examining the data, I believe any discrepancy is minor and will not affect the overall analysis.  With those caveats stated, let's begin.

First, looking at the data, I see that, except for two exceptions, if the average concurrent user mark increases in a month, so does the number of ships killed in low sec.  If the ACU total goes down, so does the average number of ships killed in low sec each day.  The exceptions are August 2013, when the ACU fell and the number of ships killed increased, and May 2014, when the ACU rose but the number of ships killed decreased.  So in my judgment, any argument that states that the number of PvP kills in low sec is divorced from the ACU for Tranquility fails.

Second, PvP activity, as measured by ship kills, is declining in low sec.  Looking just at 2014, the number of ship kills declined from 12,701 per day in January down to 10,131 in June, the month that Kronos launched.  But making such a comparison is possibly unfair due to seasonal factors1, so I then compared June 2013 and June 2014.  In June 2013, low sec saw an average of 12,151 ships killed every day.  That indicates a year-to-year drop of 8.4% in the number of ships killed every day.

I should note that June's average of 10,131 ships killed per day is the second lowest total since December 2012, the month that Retribution launched.  The only month with a lower average was September 2013, when only 10,046 ships were killed on the average day.

So if PvP activity, at least as measured by kill mails, is not responsible for the increased activity in low sec, what is?  Surprisingly enough, the answer is low sec's PvE content.2

The increase in NPCs dying in low sec began with the arrival of TEST after they had lost their sovereignty to the CFC in September 2013.  From September 2013 to February 2014, the average number of NPCs dying each day in low sec to capsuleers rose 37.4%, from 190,652 to 261,889.  The drop in the number of NPCs dying in March and April corresponds with TEST joining the H.E.R.O. Coalition at the beginning of March and H.E.R.O.'s subsequent deployment to null sec.

In fairness to TEST, they did not just carebear during their stay in low.  The average amount of ships killed each day from September 2013 to February 2014 also jumped from 10,046 to 15,497, an increase of 54.3%.  In the four months since the formation of H.E.R.O. and the beginning of its null sec adventure, the number of ships killed each day in low sec has fallen back down to 10,131.

PvE activity increased slightly again in May, but the Kronos release tore down the floodgates.  With the exception of one station in Pure Blind, the new Mordu's Legion ships are only available as drops in low sec belts.  The rush to find BPCs for the ships led to an increase of 25.4% in the number of NPCs killed between the months of May and June.  That activity, in my opinion, is the leading cause of low sec feeling a lot livelier than the other parts of New Eden.

Undoubtedly, the number of ships killed per day falling by 9.1% between May and June is directly related to pilots spending more time ratting and less time fighting each other.  But that is another reason why I believe any API server failures or Dotlan maintenance sessions would not affect the analysis.  Those types of interruptions in data collection would affect both sets of data equally.  Or the proportions of the data discrepancies would not vary widely enough to make a practical difference in a judgement that the new low sec PvE content, and not PvP activity, is responsible for making low sec feel as vibrant as it does today.


1.  In comparison, the number of ships killed in the first half of 2013 rose from 8,689 in January 2013 to 12,151 in June, so an argument that the drop is due to the season probably isn't valid.

2.  For this post, I am not including other PvE activity like exploration or mining.  I only have statistics for the number of NPCs killed.


  1. Excellent analysis.

    1. Agree with Anon. You've been doing some really great work and solid, reality-based analysis of the data. Keep it up, and thank you!

  2. I dont know if its just the summer or if its just local to where i live in eve but the low sec US TZ feels dead. Guess its a good thing I got the Landmark beta to distract me.

    1. Fire Bush, head to Huola.

      Looks like the Caladri- Amarr are hammering the Minmatar. I got an email from a guy who lives in Huola calling for as much help as he can. Apparently the main antagonists are a caldari alliance made of ex-Nuli Secunda.

      I blew through there last night in US TZ primetime and there were 175 in local, hammering away at each other.



  3. What I find interesting is the null-sec line. For all we hear about null-sec dying, the number of kills in null has remained pretty constant, and possibly trending upwards a tiny bit. For all that we talk about null-sec killing Eve, this doesn't appear to be the problem.

    Perhaps we can even draw a conclusion from this data. Look at the high-sec kills line. I wonder if perhaps many of those unsubbing are high-sec residents...

    1. The lines sort of makes a mockery of the risk for reward theory.

    2. Only if you put it into relation to # of chars there.

      The nullsecgraph only shows that wars do not matter overall killwise and the people living in null are more involved in the game so they are not that correlating to the ACU.

      For lowsec it might be nice to add the addition of clonesoldierspawns.

  4. That year to year drop in kills may have something to do with Brave moving out of losec. That happened right before June this year. If you did some analysis of the increase of kills in Catch over the same period, you might be able to make the correlation.

  5. Crossing Zebras is tabloid writing at best. Keep putting those fucks in their place.

  6. Sorry it took some time for me to reply, being EiC of CZ does take it’s toll ;)

    Let me preface this by saying you are the master of stats and I’m noob by comparison, I fully expect I could be wrong. With that in mind, here’s the issues I have.

    Undoubtedly the number of kills in LS is not divorced from the ACU. However, compare it to the kills from NS and HS in your graph. NS remains the least affected, still rising steadily. HS kills are the most volatile and follow the ACU more closely. The line for LS kills looks like the median between those two but tilted slightly upward, showing the most positive development out of the three. The interesting part to note here comes around Feb 2014 to now.

    (Note for the below: I’m using the Gal/Cal war zone as the basis here because it is the most active, has the most effect on LS stats, and the one I have any real knowledge in.)

    Allow me to explain. After farming the crap out of FW, TEST left, leaving an exposed Caldari militia while the Gallente set about the campaign of taking the war zone back. It culminated in the taking of Inna April 1st. Since then GalMil have basically been mopping up what’s left of CalMil, a militia that hasn’t really found its feet until very recently. All this is a considerable factor in the stats you see for the period. Less WT’s, less fights, less kills - along with the effect from the ACU curve. At the end of your graph the pendulum is still swung far in the favor of the Gallente, this means a low amount of kills compounded by the summer slump, AT and the world cup.

    History tells us that the pendulum swings however and we’re beginning to see the first signs of that in the Gal/Cal war zone. It is not unreasonable to think that the latter part of the year may at least somewhat closely resemble the first, augmented by the ACU. Assuming the ACU remains static or even follows its current path and assuming LS kills are as affected by it as they have been in the past still puts projected LS kills for 2014 at the level of 2013, if not better. Templis CALSF and others are taking the lead on re-inventing the Caldari Militia and we are seeing other entities join the fight on the side of the Caldari. Naturally, the Gal/Cal war zone isn’t the same as LS, but it does play a dominant role in the stats, historically speaking.

    So why do I think LS will be less affected by the ACU? Because I believe that the main leak in accounts is represented by HS alts, owned by older NS players, responding negatively to Crius because it requires them to change how they operate. The wound has been cleaned out and it hurts, but as they say; it will get worse before it gets better. As a new breed of industrialists that find ways to seize the opportunities offered by Crius emerge, the alts will return, but fewer in number because slots no longer need to be held. Secondly, Crius is proving to be a boon to LS industry, but LS has not yet developed the culture to make the most of it.

    Another thing to consider is the release of Kronos. The introduction of Mordu’s rats and sites are what’s behind the spike in NPC kills, as you point out. However, this will probably normalize and be much lower in the latter part of the year. When novelty tapers off more people will return to PvP, thus moving from one statistic to another. Yet another thing to remember is the plex-farming nerf that has killed off a slew of alts just recently, adding them to the ACU statistic without affecting PvP kills in LS in any significant fashion, lessening the effect the ACU has on the projection of kills.

    Ok, those are my garbled thoughts. Don’t really have time to go on or sort them out further atm, but feel free to crush my arguments all the same :)


    1. Not going to get into the why's and wherefore's of your opinions of low sec and null sec participation in the game, but I will comment on your opinions of high sec manufacturing alts.

      Yes, null sec players are moving their high sec mfg alts to null, or even deleting them permanently from the game because of some redundancy with their existing mull sec chars.

      But there are a vast quantity of high sec mfg players and their alts that are also leaving the game because of Crius, and they won't be returning, in any shape or form, because high sec mfg is dead, at least any profitable mfg.

      As for low sec manufacturing getting a buff with this disaster, not a chance. No low sec mfg group can ever bid on a team, because they immediately post a huge bullseye on them for every bored griefer group in the game with a supercap fleet.

      Eve's PCU is at the lowest it has been for years, and I see no reason why it will grow in the short or medium term.

      The null sec cartels, courtesy of mynnna's control of the CSM and CCP's slavish desire to keep the CSM happy, has the desolation of high sec almost complete. Missions and Incursions still have to be killed off as profitable ventures, but that is moving along, and high sec account numbers dropping continues accordingly.

      Null sec stagnation is not being addressed until at least summer 2015, so no reason to expect a surge in accounts there.

      And as for low sec, FW is the only activity level that is growing, and that can only reach a certain activity level before it flattens.

      Eve, and by extension CCP, is in a real tailspin, and it has no one but itself to blame, because of their desire to keep the cartels happy. CCP could reinvigorate the game in one day if they had the balls: At downtime totally remove all sov mechanics and remove every supercap from the game. Null then becomes a free-for-all where you actually have to live in and protect every piece of turf you want, and there is zero chance any cartel has the manpower to protect every system they control today. Smaller groups could get a toehold again in null. But that is not going to happen.



    2. While I'm not as pessimistic as Dinsdale, I don't see low sec outpacing high sec industry. I think the teams will keep that from happening. Besides the fact I'd never be able to win a bid for a team, do I really want to draw that type of attention to myself? A PvP alliance may do that for its industry folks in order to attract fights, but that's about it. Unless FW is going to be bidding for them, but I don't get the feeling that will happen.

    3. "Besides the fact I'd never be able to win a bid for a team..."

      No, you won't. The best teams will go to the null sec alliances, who can outbid any of the remaining industrialists.

      And forget about trying to freeload. They already plan to move the teams between systems, including in high-sec, where they are setting up new POSes. The reason for moving the teams around is to avoid the cost increases which would occur if they kept the teams in the same system forever, as other players converge to the systems where the null sec industrialists have put the best teams.

      And, don't be stuipid enough to believe that they are going to be madly shipping stuff around to keep up with moving teams - they are simply expanding and setting up parallel manufacturing in multiple systems. When one system is finished with a team, they temporarily shutdown production in that system and start up production in the next system with a new team. Wash and repeat, forever.

  7. Low-sec isn't dying for the simple fact that CCP doesn't know what to do with low-sec, and has, therefore, pretty much left it alone. Thank God.

    The drop in overall player numbers is due primarily to reduced industry player activity in high-sec. Since the industry changes were first announced, the PCU graphs have been dropping steadily, as the casual high-sec industrialists have been letting their subs run out, and other industrialists have been consolidating to fewer accounts and logging on less frequently (pretty much shutting down their high-sec operations). Sure, the big null-sec alliance industrialists have been scaling up to take more control of high-sec industry, but they only make up a small percentage of the industry players in the game (although with disproportionately greater influence over CCP's direction - ref: mynnna). So, you should see more active high-sec POSes after Crius, but they will mostly belong to null-sec alliances.

    The downward trend in high-sec industry players will continue after Crius, maybe even accelerate a bit, since Greyfail's most recent devblog basically was a big F-U to these players. It will probably level off in six months or so, and we'll then be able to judge the impact of these not-so-smart changes.

    The real question is whether or not the drop in player accounts, and corresponding drop in revenue, will force another round of layoffs at CCP. All hype and speculation aside, whether CCP is hiring more people or laying more of them off is much more indicative of the health of CCP and the future of its flagship product.

  8. Looks like Gevlon has taken your analysis and somehow managed to change it into "ACU shows that null sec players are playing in highsec!". Quite amazing.

    1. Actually, having read his post, he didn't take my analysis and twist it. He took the data and went far beyond what I did. Basically, he confirmed things that null sec players have said for years. Think about it for a minute. Do you really think Gevlon wants to validate The Mittani's arguments?

  9. I doubt he does, but his analysis is still flawed by his preconceived conclusions. Clearly nobody has told him that correlation does not imply causation. It's just as viable that the increase in hs activity and the decrease in null ratting both occur due to an independent cause. He leads off all of his "analysis" with a conclusion, then uses excel functions wildly until he has something that can support them.

    1. I don't see where he tortured the math in his post. He did a quantitative analysis. Also, in the case you are complaining about, his conclusions were at the end of each section.

      Basically, he was showing his work to show how he arrived at this conclusions. And yes, he did draw conclusions. That's why you do that type of analysis. But at least we know why he thinks the way he does. If you don't like his analysis, tell him why it's bad. Don't just say, "Gevlon wrote something, it must be wrong!"