Greedy Goblin

Thursday, April 17, 2014

Morlocks and Eloi: the dark secret of EVE Online

A few weeks ago I wrote a post named Morlocks and Eloi: the dark secret of RvB. I've found that Red versus Blue could be distributed into three groups that did equal PvP damage. The first group has 0.7% of the pilots in it, the second group has 3.3% and the remaining 96.0% also done 1/3 damage.

I attributed that to RvB being a scam that is designed this way, to feed "elite PvP-ers" with kills while they do their duty. I received the kill and loss data of Brave Collective for the last 6 months (2013 Oct - 2014 March) to show how a really newbie friendly alliance performs. Well, they don't do much better. The top group has 1.1% of the pilots, the second group has 3.6% and the remaining 95.3% also done 1/3 damage.

Oops. Almost identical numbers. So I turned to the last data I had, GSF 2013 kills and losses. The results are practically identical: The top group is 1.0% of the pilots, the second group has 4.2% and the remaining 94.8% also done 1/3 damage.

I've found back then that the top group has good ISK ratio, the middle group loses as much as kills and the bottom group is slaughtered. There isn't much difference here either if you look at the charts of RvB, BNI and GSF:

What does this mean? Two things: the first is that "elite PvP" is alive and well even in the alliances that claim to be the most newbie friendly and consider "elite PvP" a curseword. Secondly, it proves that PvP-ing is mostly irrelevant in the Sov scene. I mean it's obvious that the 87 pilots that form the "top 1/3" group of GSF, couldn't fight sov battles. Most sov battles have relatively low kills or no kills at all, because the enemy runs away. Sov warfare is rather PvE than PvP, fleet attendance and not PvP performance matters. Carebears are better for a Nullsec empire than PvP-ers.

What is the "dark secret of EVE Online"? That 95% of those who participate in PvP are slaughtered like sheep and even the most "newbie friendly" and "community based" alliances can't help them at all. Those who don't have the dedication and skill to get into at least the top 2-5% of the players are better off avoiding PvP completely and focus on making money, and flying "click broadcast, press F1" fleets. There is no such thing as "average PvP-er", just a 1% elite, a few % elite aspirants and 95% sheep.

Of course I'm not saying that PvP-ing is worthless. Hundreds of billions of ISK can be destroyed and even more stopped from being earned by hunting farmers and haulers. But this is a job available only to the top 1% elite.

Below you can see the elite PvP-ers of Brave Collective: Santo Trafficante, Seth Goa'Uld, Plask Arthie, Ripley Kaga, Nippery Slipple, Malphas Inanis, Leahcim Vinjen, Axandrias Darkstar, Janne Harkala, Black Bart Pirate, Andi Onthatop, Lumpymayo, Riper Dennom, Mattisen Arcanth, Mikey Ormand, Dardanos Herakles, Jack Heisenburg, Arik Alabel, Anna niedostepny, Sgtchronic, Toros Culzean, Le Petite More, Caprice Ohtar, Lefty Leftfield, chopper1k, Tyr Dolorem, Richard roarimlions, Larss Tivianne, Crazy HybridChick, Altrue, Nancy Crow, OriginalName IV, FricktheTurtle, Keesha Echerie, Vikkiman001, Verak Padecain, Braldye Sinma, pretty hair, Malik Fett, Calfis, Blue Ice, Dom Arilio, Drumin Cloud, Ryvarant, Thandion, watch yoself, Madison Madeveda, Skylif Beria, Bishop Halifax, Scrymjaw Dazai, Jobe Delaar, Nicholas Aideronne, Charles Wu-Wong, Revolt Sklor, Elena Cassidy, Forgione, dreamertt, D Montesquieu, Ayanamii Ikari, Alistair Chromewell, Andr Katelo, Nour Aerilon, Denae Soarus, Thirtyone Organism, Seb Solo, Nxtg3n, Socco, Laokoon Firetail, Clearly Knott, Panic Mangeiri, Motre, a DAMN PATRIOT, Chuck Fiinley, BrotherNick Tranquil, PillowFluffer, Xyvos, Janet Sukarala, Rae Kusoni, Moomincake, Colonel O'neil Tauri, Gullron, Tessa Gatria, Svana Shaishi, User Offline, ScentDeLaPwnage Loutte, Merling Erquilenne.

And the most ironic list: the elite PvP-ers of Goonswarm Federation: Warr Akini, Nina Hayashi, roaritron, Alex Gianturco, eC Cade, Rehtom Lamina, Ikolo, Benchab Mas, Aaron eVega, Mittani Alexander Gianturco, Proud Brother, rackarns, Unfortunate Uncle, Alexander Mittani Gianturco, JohnCrichton, shelly's Jihad, CowWarrior, Gad'ui, Silin Vespula, oodell, Furious Father, Vatek, ooeall, ooaell, Beryozka, Bubbles9001, Futility 101, LifesAbitch, ooeell, Mikalar, kaldarinka, kyrie1eison, Sipphakta en Gravonere, Durjaya Hemadri, Garnoo, 3Better, Cyber Yeti, Issabella, ooiell, Grizzly Baddums, Colosus blade, Bob Haldeman, Dungar's Sister, Gennel, Tamirr U'tath, Kasa Shadowman, Ministerial Duties, Aranorn, GubkaBob Mrazish, Blacklyche, M3AN FOOKER, sakhan Bravo, Alex Ribblestraib, FlamingLips, Gorn Arming, Sensei Chilinata, Gaara's sniper, sakhan Alpha, DaShmoo, AleksLuc, WildIRISH, PellePlutt, Forsaken Skipper, Lazei, Angel Logisticus, Dark Sn1p3r, ginmushi, Terminal Adtur, Agentile, Chikotilnik, Eliza d'Alikchi, Castle Sevas, Tiberizzle, Guuldan, Braygo Khallazar, Hentai-san Daivis, Monster Dude, Ymka Gotovcev, Gluttony the Voracious, Gilles de Rais, Lust the Lascivious, Wrath the Furious, Sloth the Indolent, Ebony Sniper, Seamus McLove, Jeanne D' Arc, Hassan- i Sabbah.

Update: Michael LeBlanc in a comment had a good idea, to display the pilots in an exponential scale. So I made the BNI chart. Please note that the X axis is not misaligned, there are 7600 pilots who got on killmails, but 3000 of them had no kills just losses, and ln(0) = -infinite:


Snow said...

Hey Gevlon, I'm one of those people in your BNI list.

Firstly, "Elite-Pvp" isn't a curse-word in BNI, in fact we have some corps dedicated to people who want to focus on small gang "Elite-PvP". Players are encouraged to take out cheap ships and die repeatedly to learn, when I started out I was terrible and died often, but after dying repeatedly to Agony members in Syndicate and learning from them I've been able to be the one who does the killing.

Secondly, I think it's incorrect to assume that the distribution of ISK efficiency in Brave/GSF are anything other than normal. If you look at other games (MMR in Arena, MMR in Dota, etc) you'll see a similar distribution where a small percentage of players are extremely good, a moderate portion are good, and the rest are terrible. What would be more interesting is to see how the ISK efficiency distributions between large alliances differ.

Eaten by a Grue said...

But a funny thing might happen if everyone takes your advice. If people stop PvPing on a massive scale, then losses from PvP will of course also drop dramatically. With fewer losses, this means far less demand for market goods, since, as far as I know, the EVE market is basically just stuff to replace your losses. So then the PvE won't be as fun either.

Anonymous said...

Congratulations! You've discovered a "power law" as it applies to Eve PvP. I'm sure that you'll find a similar law applies to World of Tanks, World of Warcraft, League of Legends. Not to mention just about any other online game...

Anonymous said...

oh boy thats a big can of worms. Just curious try it out on some other alliances. NC dot, -7-, YF, and for giggles noir or bl. I would be hugely interested in seeing what those numbers are.

Unknown said...

..Gevlon, the three entities that you have analysed are the big three that invite all who apply. Goons take all SA members who apply, and as we all know SA is a monster in terms of user base. So what your analysis shows is that extremely large entities who accept any and all applicants have a large membership who under preform. Shocking.

As to your conclusion that the bottom ~90% are just sheep sent out to be slaughtered. Can't you see that you get out what you put in?

If those ~90% get slaughtered, that was their own choice. Some improve to join the other two groups in your stats, and other stagnate at the bottom 1/3rd.

But all had fun, or else they would leave the corp and move on.

Gevlon said...

@Snow: The surprising thing is that the bad people can't be helped by their group to get results. I mean it was thought that if you are personally bad but fly under a good FC, you'll get kills. It seems you won't.

@Eaten by Grue: PvP in EVE isn't consensual. If you choose to avoid PvP, you can still be killed.

Unknown said...

<< But all had fun, or else they would leave the corp and move on. >>
That's not necessarily so.
Once a person is in a corp, he may be compelled to stay for reasons other than fun. For example, because even though he thinks current corp sucks, he isn't able to think of any ways anyone could do better, so he might as well stay.

People tend to keep playing MMOs long after they stop being actively fun, simply because they become a habit.


The power law is a very clear thing, but i don't think in other games the segregation is as extreme as in Eve. Though, to be fair, that depends on the metric - other games tend to not have anything directly comparable to "ISK destroyed". Even ladder rankings are not quite similar.

However, what i find weird is that such a large amount of people find themselves unable or unwilling to learn and advance. I personally blame the absolutely horrid user interface experience of Eve. You can have a hardcore world without turning interface handling in a game all of its own :/

Louis Robichaud said...

Aha, so RvB is *not* a scam then? Glad to hear it. I'm glad you have done the comparison, although you should have done that FROM THE START instead of throwing those accusations around...

... but better late than never.

Anyway, another aspect for you to consider. Loses per pilot. While the elite pilots have a much better win:loss ratio, the amount of loss *per pilot* is huge! Let's look at the RvB numbers again:

The "bad" group lost approximately 215 million per pilot per year. That's a very modest amount, less than a million per day. Meanwhile, the elite group lost 6.9 billion per pilot per year. That's over 30 times more! These elite pilots are getting a lot of kills, but they are paying a heavy price for it. They are risking blingy ships while the "bad" pilots are roaming around in t1 frigs.

Gevlon said...

@LR: RvB is still a scam for two reasons.
- It actively recruits new players with the offer of learning and participating in fun PvP (instead of the truth: be a sheep)
- It has no other gameplay. A Goon sheep can have success by ratting enough in Deklein to have a Dread he could whore on PL titans in B-R and he can also grind down regions, seeing the map shaped by his hand (this is true, not illusion). In RvB all he can do is undock and die.

Anonymous said...

There is no such thing as "average PvP-er", just a 1% elite, a few % elite aspirants and 95% sheep.

Real life capitalism works the same way.

Anonymous said...

The surprising thing is that the bad people can't be helped by their group to get results. I mean it was thought that if you are personally bad but fly under a good FC, you'll get kills. It seems you won't.

I'm curious if this is your famously "dewhored" data? because if it is, then you are modifying data to support your own presupposed conclusions.

Heres why - if you are a "bad pilot" flying with a "good FC" (and by bad pilot I'm going to say that means low skill points, or low ship class), and you fulfil a valuable role (hero tackle, ewar etc) then you will indeed appear on the kill logs of lots of kills and your total kills will be as high as everyone else who was involved. This is essentially the equalizing factor - if you participate and play an important role you are considered equal on the kill.

The "highest damage" metric isn't at all valid. If you look through the kill logs for any alliance for a month you will find that most of the time "highest damage" always goes to the same group of people - this is because they fly DPS, or caps. But on their own they cannot function. A battleship fleet lands on a bunch of cruisers and easily outmatches it? the cruisers will simply warp off if there i no tackle. Now the tackle might only get 0.1% of the kill value, but if he wasn't there NONE of the battleships would have gotten ANY kills. The tackle is a force multiplier and kill enabler.

Your data also completely avoids logistics. I fly in a wormhole alliance in armour gangs. We use a LOT of logistics. Lots of it. Logi pilots simply don't show up on kills at all - yet again, without them the fight never would have happened. And before you say logi should use drones, etc - they do. Logistics drones. Because the point of the fight is to win and hold the field, not to come out on top on the killboard and whore as much isk as possible.

Your "de-whore" numbers effectively show you who gets top damage. You then list those who get top damage yet avoid the fact that top damage is entirely expected for DPS pilots. You are taking a very simplistic view of eve battlefield dynamics.

Gevlon said...

@Anonymous: check the bottom chart of

In this I made an alternative de-whoring: everyone got equal share of the kill who got on the kill. So the tackling frig with 0.01% damage got the same value as the titan who doomsdayed it. No big difference.

Anonymous said...

It made no difference to the RvB dataset. Surely even you can see that statistically speaking the rest of hte alliances kill record is more significant and worthy of a decent look?

And you still haven't solve for logi, non-combat alts and the like

Anonymous said...


Undocking and dying is an inherent part of pvp.

For RvB to be a scam, from the reasons you set, you would have to show that

a) New pilots think they are doing anything other than exploding gloriously
But, most importantly:
b)That they are enticed to sign up for RvB for reasons other than "constant pvp opportunities".
If you can show RvB offering industry, sov, PvE, Mining etc, and not delivering, then it is a scam, but, afaik, RvB has never advertised as anything but "Undock and get shot, and shoot back".

You are saying it is a scam if I work at a bakers and am expecting to be chopping meat as well.

Unknown said...

Its not the group's fault that they cant help people not suck at pvp. One of the biggest reasons for this majority who get no kills relative to losses, is that they rarely log in. When they do log in, they will fly some expensive ratting/mining ship/freighter, and get ganked. Then they do a little pvp, and log in again in a few weeks. Thus, their ratio sucks. How would the group help that guy, and why the hell is it their duty to do so?

Lucas Kell said...

"The "highest damage" metric isn't at all valid. If you look through the kill logs for any alliance for a month you will find that most of the time "highest damage" always goes to the same group of people - this is because they fly DPS, or caps. But on their own they cannot function. A battleship fleet lands on a bunch of cruisers and easily outmatches it? the cruisers will simply warp off if there i no tackle. Now the tackle might only get 0.1% of the kill value, but if he wasn't there NONE of the battleships would have gotten ANY kills. The tackle is a force multiplier and kill enabler."
This is a pretty important note, and it extends to logi considerably. Generally in larger scale combat, logi will be focussing on keeping up the people that will most contribute to the destruction of the enemy, which amplifies anyone that is already ahead of the curve, while not contributing to the damage because they are logi. Then when logi go down, they still contribute a loss, further extending the gap between the two groups. Then look at other aspects of combat. Look at the CFCs FYF. Those are generally the newer players in ewar and tackle fit T1s, they are unlikely to land much damage, but they plow into an enemy, neuting them and disabling their logistics, which is a massive part of the battle.

I don't think that it's a difficult concept to understand that a small group of people will excel in combat while the vast majority don't, but to suggest they are just sheep to be slaughtered to the extent claimed here is far from truth. It's just another display of why killboard stats don;t mean everything.

Anonymous said...

You're actually surprised that an extreme version of the Pareto Principle applies here? This fact, that 95% of PvP kills are concentrated in a small player base, has been known in EVE for years.

In the real world, 80% of the work is done by 20% of the personnel. But in the real world, for most first-world jobs, you can't come and go as you please. You have bills, and you can't just show up to a job and start working. As a result, the bottom-8 tends to be more productive through sheer force of organizational momentum.

In EVE, back before CCP stopped publishing such data, the average subscription length was 3-4 months. There's little evidence that has changed. You can come and go as you please, so there's no organizational momentum to force you to produce output.

Why? Because people can and do play for fun. Not everyone is interested in being an "elite PvPer." They'd rather play for their 3-4 months (at a cost comparable to any AAA title), be an average PvPer, have fun, and move on to the next game. The time, effort, and money to become a 60M SP elite PvP pilot doesn't interest a huge segment of the gaming population.

Anonymous said...

In the paragraph starting "Of course I'm not saying that PvP is wothless" are you meaning that ganking freighters and industrials, undock-podding and looking for site runners are activities only the elite can do? My personal experience is to the contrary - I'd say these are among the least challenging destruction-based activities in the game

Von Keigai said...

Gevlon, as one of the nonies says, you've discovered a power law. In specific, check out the wiki on Pareto principle, which is most applicable in this case:

The Pareto principle (also known as the 80–20 rule, the law of the vital few, and the principle of factor sparsity) states that, for many events, roughly 80% of the effects come from 20% of the causes.

The 80/20 split shows up in a lot of human affairs. (The split you are seeing is the top 20% iterated three times.)

Unknown said...

I think you are taking the wrong lesson from these numbers - the numbers are good, but there are multiple ways they can be interpreted - you cannot merely assume a why, which is what you do when you attribute the term Morlock and Eloi to the the numbners.

Let us look at real world examples: in the military you have a large number of standard troops - many of which will die if they engage in combat. However, you have have much smaller and better equiped elite groups that tend to dish out the dying more than receive of it.

There will always be elite groups simply by benefit of individual differences. Everyone can cook, not everyone can be a chef. Everyone can pick up a gun, not eveyone can be a crack marksman.

Equal opportunity never gaurantees equal outcomes. All it does is provide everyone the same opportunity despite background to try for it. Whether it is RvB, BNI, or Goonswarm, you have groups offering players of all levels opportunities. But just because everyone has the opportunity, not everyone is equal in skill.

Just like you: you are an elite Trader. It is where you excel while many others fail to come close. In a sense, using your standards, you are a Morlock where trading is concerned among a sea of Eloi.

Unknown said...

Any chance you can post your worksheet so I don't have to re-parse your data files? Basically I'm just looking for a list of pilots and their total (unwhored) kills and total losses, which I imagine you've already done.


To add to what other posters are saying, you already saw from the RvB activity graph that activity has roughly a negative exponential distribution.

If the kill distribution is also negative exponential, then, depending on the mean, you will generally have the top third be a small percentage, and the middle third be about 3x as big, which is roughly what your numbers show (the middle third is bigger than it ought to be, so it's not exactly an exponential distribution).

Moreover, if losses are roughly independent, then you will end up with a very high K:L ratio for the top third, mediocre for the middle third, and bad for the bottom third. Which is exactly what we see. Except diving into the actual numbers, a completely independent distribution would be significantly more skewed than what you show, suggesting a pretty high correlation between kills and losses (which isn't surprising, activity drives both, and we already know this about RvB as you posted losses/pilot in each group)

The 95% bottom third may have a terrible K:L ratio overall, but this does not mean that all 95% of them do. In all likelyhood, close to half of them actually have a positive K:L ratio, you just skew the results by removing the people with the best and leaving the people with the worst.

Snow said...

@Gevlon: I don't necessarily agree that it's shown that bad players can't be helped by this analysis. An interesting study would be to compute monthly improvement in un-whored isk efficiency for new players or new members of RvB/GSF/BNI/EUNI and see which group has the largest average increase in efficiency for new players.

Tegiminis said...

This is natural for any/all competitive pasttimes. Very few football players join the NFL, very few Dota 2 players go to The International, and so on.

Broadcasting it as some big secret just makes you look out of touch. This is how it is and always has been for all games. The vast majority of players are average/below average, a small group of players are good, and a tiny group of players are the best.

Also, you fail to take into account some things. For example, if I'm primarily a logistics operator (not heal logi, but fuel logi), chances are I'll have virtually no ISK killed but plenty of losses.

EVE is a team game, especially on the strategic level, and some of that means having "bad stats" to enable your team to do better. It's like yelling at the tank in a dungeon because he's not topping the DPS charts; of course not, he's a tank.

Arrendis said...

oodell, ooeall, ooaell, ooeell, ooiell <--- One guy, multiboxing that high skill level, btw. oodell's a baller as fuck pilot.

I mean it was thought that if you are personally bad but fly under a good FC, you'll get kills. It seems you won't.

There's two parts of this assumption that you're not accounting for. First, you've got organized groups that train their FCs and promote based on merit and competence, you wind up with the bad FCs not staying FCs. And a 'good' FC won't take a fight that'll get his people slaughtered - so when the forces of two of these organized blocs fight one another, if it's a very lopsided fight, the outgunned FC will tend to extract as best he can, minimizing losses. The result: no matter how good the 'winning' FC is, a fight that doesn't happen won't get his fleet kills.

Second, a 'good' FC can only influence a fight so much. Part of it is what Clausewitz called 'friction', the unavoidable complications that arise unexpectedly due to the sheer nature of life. Part of it, though, is if you have a 'good' FC with a fleet full of 'personally bad' pilots (the vast majority, as you've demonstrated), then their poor individual performance will degrade fleet performance.

I'd also posit that in any significant fleet action, between 10-20% of the fleet is logistics, who skew the numbers downward by sheer merit of CCP not crediting them for any of the fleet's kills, despite the logi making those kills possible (by keeping the fleet / FC alive long enough to get those kills). So every one of those logistics pilots, if they fly on 100 fleets, and lose 3 ships, during that month, will show up with a Kill/Loss ratio of 0 / 600,000,000.

This gets skewed in the other direction for the 'high end' pilots - you morlocks - by capital ships showing up to subcapital fights. The odds of a BL ishtar gang inflicting significant losses on a CFC or N3 carrier fleet, for example, are pretty low. Now, this is clearly evidence of high-skilled characters - after all, they can sit in a carrier - but at the same time, the amount of player skill needed isn't much above the level of 'align, lock target, press F1' used by a line battleship pilot. Yes, there's the cap chain and remote reps to manage, but in practice, cap chains are simple to run if you're paying any attention at all, and remote reps don't work much different from anything else: watch for broadcast, ctrl-click the broadcast, activate high-slot on the target.

Gevlon said...

@Von Keigai: in BNI 8% did 80% of the damage.

@Michael LeBlanc: see update in post

@Snow: the main problem with that is most pilots have only a few kills during a year, so improvement cannot be charted

@Tegiminis: I seriously doubt that the underperforming 90% are all fueling towers and flying remote reps

Arrendis said...

@Tegiminis: I seriously doubt that the underperforming 90% are all fueling towers and flying remote reps

No, but they can also be ratting/mining alts. I mean, I've got multiple corpmates who each run 10 mining alts and 10 cyno alts - all of whom don't get kills, but definitely take losses. And jeez, Suas has to have like 50* ratting carriers out at any given time.

* - ok, that might be a slight exaggeration.

Unknown said...

So I parsed some of the data myself. The below is just the first data I've looked at, which is December Brave data.

I get roughly the same result as you do overall, 1.3% (of people who appeared on at least one KM) do 33% of damage. 5.9% do 66% (so the middle group is 4.6%). The first group has an overall 2.67 K:L Ratio, the middle group 1.49, and the bottom 94% have 0.41. Not terribly different than your 6 month data, so I'll stick with this.

Here's the thing though, 55% of Brave pilots in December has a K:L ratio better than 1, and 56% were better than the overall average of 0.85. Far from 95% being 'slaughtered'

Moreover, if you do your same analysis, except sorting by losses, you find that 33% of the losses are taken by 2.4% of the players, the next 33% are taken by 6.9%, and the final 33% by the remaining 90.7%. K:L Ratios aren't as extreme as measuring the other way, but still the 91% has a K:D of 1.14 while the the 2% has a K:D of 0.57.

So splitting the data by losses suggests the exact opposite of your conclusion - that most of the player base is feeding off a very small minority of bad players.

My favorite bit though, is that 42% of the 'Morlocks' when sorted by most damage done, are 'Eloi' when sorted by most damage taken!

Guldur said...

That was the first step. The next will be when you understand that it is how the entire world works for ever and forever. And then seeing how pointless is our lives, you go suicide yourself. This is called in many SCI-FI stuff the Ascension...

Unknown said...

Why is there no list of the top 1/3rd of RvB? i'm curious to see if people i fly with are part of that list!

Gevlon said...

@Quinten Sarn: clicky link on top of post.