EDIT:
Can’t seem to mark the answer as answer to let anyone curious skip all the sidesteps / trolling.
The CCP video the numbers come from is linked below by Kezrai
It is just a small quote from 3:05 in the video, It is a general statement about a year of Eve, not about 2025. So since Eve player count seems steady (online count steady, player count kept secret by CCP), this does mean about 40% of the players quit and is replaced by 40% new per cycle, but the cycle time is 12 months, not just 4 months as it would have been if these numbers were form 2025. Giving about 3% players replaced per month.
He does say players, whether that is actually players or accounts is usually rather vague in Eve, but in this context it could be the actual player (as best CCP can make out) and not accounts.
40%-45% of the current players were not playing the game at the beginning of the year. Meaning we have a lot of new players playing the game, and potentially returning players.
As Aisha said, typically when CCP discusses these things, they are referring to “new accounts” and not “unique players”. They’re also usually referring to “active accounts” which varies, but typically means something like “has logged in during the past X months” or “has played at least X hours during the past N months”.
So think of this as “out of all the recently active accounts, almost half of them are new accounts created in the past 5+ months”. This is in line with some other data CCP has published showing that there’s a high turnover of new accounts, but not many of them stay for the long term.
I infer something a bit different with this number - as the churn rate is ~95% (or roughly 500 new players retained per week). Over 4 months this only works out to about +8,000 new players added to the game.
I think it’s more likely that this number reflects 40-45% of the total player base being inactive at the beginning of the year, or around 550-600k accounts.
I have no idea what metric would be used to determine this. Maybe no activity for a set period of say 90 days? It’s also entirely possible that I’m way off the mark here…
This shows more people online than one year ago. I know it is not showing us new people come and go. Also take note the line graph slumps in the summer months and pops back up in the fall and winter. Good weather take its toll on gaming.
Despite all the jaggies in the graph you see the same slump in the summer time and rebound in the winter. People do this on all MMO games. They kick the tires. I guess I should go comment on Srima’s thread now.
Somehow CCP claims to know the difference between accounts and players. For example, in a summary of the battle of FWST-8 which is the biggest battle in EVE where the server didn’t malfunction (and back then was the biggest battle period), the listen the number of toons which participated, the peak number of toons on grid at a time (which was apparently not the same as the previous one, I guess some people d/ced or joined in late), and the number of players. A surprising number of people in the battle never had more than 1 toon logged in during the time at the battle.
They do know this. The main thing that ties accounts together are payment cards, the other, IP address. It’s trivial for CCP to know, but we as players can only speculate.
This is just the current online numbers, the player population (as in number of subscribed accounts) haven’t been publicly officially stated for a very long time, Eve was among the last MMOs to publish it (they all stopped as they killed their games and tried to hide their failures). Now of course they have to make some active metrics to handle the alpha accounts, but as they state they do not actually give out the numbers - another good example of CCP statistics, notice how they say they don’t give out the numbers, and to avoid explaining why, say they never have (DATA AMA video linked above, at 6m07s). But below is a quick find in official statements ( EVE Online Surpasses 500,000 Subscribers Worldwide - EVE Community ):
" EVE OnlineSurpasses 500,000 Subscribers Worldwide
Start of Game’s Second Decade Sees Most Explosive Subscriber Growth in its Rich History
REYKJAVIK, ICELAND – Feb 28, 2013 – CCP Games, the world’s leading independent massively multiplayer game studio, today announced their award-winning, sci-fi MMO EVE Online® had surpassed the 500,000 subscriber mark worldwide for the first time in the game’s history. The milestone continues ten consecutive years of subscriber growth and follows the December 4 release of EVE Online: Retribution, the game’s 18th free expansion and its most successful ever."
I assume you are not looking very hard. Try this page for all the stats. I know you are going to tell me there has never been 133,000 players online in a given day. People come go. I for one, don’t hang out 24 hours a day. There might be 30,000 online but they log off and others take their place. Not certain where they pull results from, but they do track 143 games.
Of course they do, but they will say what suits them best marketing wise. Sometimes that means hiding things and sometimes that means amplifying things.
(Replying to Princess Rose Ivory’s post, but it’s a comment about data validity in general.)
Keep in mind there’s a difference between finding a website that publishes numbers, and finding a website that publishes accurate numbers. MMO-Population uses a formula based on Reddit account activity for that game’s Reddit threads. So it’s pure guesstimate based on an external (to the game) reference. MMO-Population: About
While I check that site myself from time to time to gauge “when interest has gone up and down in a game”, it’s not at all an accurate tracker of actual game activity. For instance it can show a ‘spike’ in activity during something like Nullsec Blackout, when Reddit activity goes up but actual in-game activity goes down.
Even CCP’s best internal numbers will be a (close) guesstimate because they also ‘estimate’ actual unique people from things like payment sources (but one person can use multiple payment sources, or multiple people paid by a single one), emails the account is registered to (some people register all their accounts to one email, others use different emails for different accounts), or IP address (spoofed and combined in multiple ways).
At the end of the day the specific numbers are always going to be a guess. What helps is to use consistent methodology over time to compare patterns and trends, and to keep in mind the limitations of various sources when combining them.
People keep on and on confusing…often it seems deliberate…‘concurrent’ players with ‘active’ players. They are simply not the same thing, especially given that ‘active players’ is generally a longer time frame such as the course of a whole day or more. Concurrency figures really don’t tell you how many people are actively playing the game.
For example…20,000 active players logged in 24 hours a day will give you the sameconcurrency as 80,000 active players logged in for 6 hours a day. The number of ‘active players’ could shoot through the roof, but if they all log in for less and less time then the concurrency could remain exactly the same.
You could even have an increase in concurrency with fewer active players, or a decrease in concurrency with more active players. The vital part of the equation people keep on and on missing is how long people log in for.
All the concurrency graphs that people love to display are really somewhat meaningless…yet people love to display them as if they represent a precise metric. Once you grasp that concurrency falling does NOT necessarily mean ‘less players’ then you realise just how useless those graphs actually are.
Believe it or not. I was contracted by a company that wanted a way to verify botting, multiboxing, spoofing. and other cheating methods. I submitted my content, we tested it on the test server and it caught every bot they tossed at it. However they paid me for my work and told me it wasn’t what they desired at present. I am still under NDA not to discuss the game, but I can tell you how it worked.
Detection
Player enters a dungeon and thinks the other guy is a bot. The human player reports him.
The players screen clicks are too accurate, hitting the same pixel, or too fast and radical, such as auto aim. The player is placed on the watch list now.
Player is active online too long, a marathon of over 20 hours and no sleep. The player is placed on the watch list now.
Testing
I didn’t want to just toss them offline bots will have a reconnect script. Instead we quietly sent them to a testing shard on the server. The walls clearly said, “You are suspected of cheating. Please respond with this code in chat to prove you are human.” The walls were replaced by this graphic and not text a bot could read. If the player didn’t respond playing for a long time, or finishing the fake dungeon, it was enough to prove botting. Otherwise if the code was entered the suspect player would get rewarded and sent to the real dungeon. If reported by a player, we took down their trust level a notch.
Apparently the company didn’t see this as beneficial to them, but I got paid and that is all that mattered to me. Sometimes I wonder, if they really do want the cheaters in the game or not. Detection has hit a whole new level this generation. Cheaters are using hardware to cheat instead of software. The software still exists, it just runs on a dirty laptop playing the game via stream broadcast. But I think about it, my method for detection and testing would even find them, because I wasn’t scanning for software patterns in memory.
Thats extremely easy to work out, each machine can very easily be fingerprinted, watching for patterns in gameplay also fairly easily indicates players as does things like the unique way each individual prefers to fly by looking at specific commands, certain players might ping specific things more often or like to group specific modules in specific ways while playing from the same IP address etc
All too true, I remember when they told people the newer CPU would include a unique identifier and all controversy about it. They can lie about locations, but you can’t spoof the hardware.
Drop to a Windows command prompt with Run → cmd
Then type; wmic cpu get processorid
This line will return your gpu as well;
wmic win32_videocontroller get pnpdeviceid
However there is one other aspect. They know machine X at possible location is playing on 12 accounts, but they don’t know if the machine is shared unless each user logs in with a different user ID. A family with 3 boys for example sharing the same PC with a general guest account is 3 players on one machine. That is just being way too technical in my opinion.
Like I’ve said before, CCP has no way of knowing if 3 machines running at the same location are the same player, or a guy who has 2 room mates. What are they going to do? Start making you send a PDF of your driver’s license just to play a game?
