Thought experiment: How to figure out the players we have

Discussion in 'General Entropia Universe Discussion' started by RAZER, Dec 7, 2011.

  1. RAZER

    RAZER Custom title ... uh ...

    Thought experiment

    Well we all know that last week MA changed the friend list and we could see the amount of people actually playing the game (about 300k). The same day Ma disabled that feature, probably because it freaked them out that we could actually see that data (well good testing on their part again). Today they re-enabled the feature, but with a few changes. We can no not see all the players and just do a search which generates a maximum of 10 pages of people (so max. 100 people). So, is there any way we can use this to generate the real amount of players we have.

    Because this is a thought experiment we can use some things that might not be entirely 'legal' like some script that does searches and OCR's the number of pages for each search.
     
  2. Tass

    Tass Administrator

    Hm...

    We need search terms that produce less than 10 result pages. Then check how many of the them are online. Leading to results like 2.37% of all "Tass" are online. Doing that over and over again and we'll come to an average value. Like in 100 searches an average of 3.4% have been online. The more searches the more accurate this value will become. Since we know the total of ~300.000 we can assume a number of players online.

    Just a try right now:

    There are 78 "beta", 1 of them online.
    25 "gamma", 0 online.
    19 "epsilon", 0 online.
    57 "start", 0 online.
    65 "tass", 1 onine.
    43 "razer", 0 online.
    81 "giga", 0 online.
    76 "peta", 1 online.

    Well, you see... with less than 100 results... it seems to be hard to do (and that's probably one of the reasons MA why have limit it to 100 results). I guess less than 1% of the active accounts are online at a time = less than 3000.

    EDIT: Ok, a little bit more elaborated. I searched "Jack A" to "Jack Z" and "Jacka" to "Jackz", some of the searches produced more than 100 results. However, counting more than 100 as 100 + all the searches with their correct result values there must at least 2650 "jack". 8 of them are online = ~0.3% of the "jack". 0.3% of 300.000 = 900. I think we are getting closer to the correct value...
     
  3. RAZER

    RAZER Custom title ... uh ...

    Nice idea Tass, but I was talking about how to figure out how many registered players we have. Sure we know the ~300k right now, but that will change I guess and that is what I wanna track some how (well hypothetically speaking of course, this might need the use of not so legal stuff according to the EULA).
     
  4. Tass

    Tass Administrator

    Ok, that would need to be done the other way round. Similar to the "jack" experiment I did above. You count the number of results for 100 search terms producing less than 100 results and then you sum up. Maybe you get 7510. In a month you are doing the same with the same search terms and maybe you'll get a result of 7535. And then you apply the change in % to the total number of ~300.000 we have right now. ;)
     
  5. RAZER

    RAZER Custom title ... uh ...

    That would indeed give us an very rough estimate. So to do this easy is to some how build a program that enters the search terms and captures the amount of people and puts that in a list for all the 100 searches every month.

    The 8 terms you used in your first post should do I think, there just need to be more.
     
  6. Search for every combination of 4 letters (including spaces), then record the names you get. I think you'd be able to find all entries this way. Obviously, you'd need a program to do this and obviously it would be against the EULA :)
     
  7. Tass

    Tass Administrator

    Uhm... even if there were no blanks, numbers or special characters but only the letters of the alphabet that would already require 4.503.599.627.370.496 searches. MA would not be happy. Then clicking through result pages, OCR reading results put them all in a database respectivly checking whether the name is recorded already. Mh... considering that process would just take 1ms each time it would still need ~142808 years until it's done.

    Am I wrong? :headscratch:
     
  8. RAZER

    RAZER Custom title ... uh ...

    So we better get started then, I have 3 weeks of from from work starting next week ;)

    and yes Neil against the EULA, but that's why this is a thought experiment, just to see what we can come up with. Nobel prizes have been won from stuff though up in one of these experiments.
     
  9. Tass

    Tass Administrator

    Ok, let's improve.

    - searches with randomized search terms
    - if 10 result pages, skip, next search
    - if less than 2 result pages, skip, next search
    - if less than 9 result pages, record # of result pages, next search
    - after R=10.000 records, sum up the values, multiply the result by 10, substract 50.000 (for 10.000 only half (=avarage) filled result pages), final result=number of players (not the total of course, just a base value to compare with future results and see how it changes)
    - the higher the value for R the more accurate the result will become
    - with R=10.000 and each record needing 3.6 seconds the whole process will be done in 10 hours, which means it can be done once a day
     
  10. I think it would be only 450,000 searches for letters only. It's not case sensitive, so that helps.
     
  11. Tass

    Tass Administrator

    Lol, you are right, I was wrong.

    We can become more realistic.

    Assuming 26 letters, 10 numbers, and 4 special characters like blank, -, * and @ we would have 2.560.000 searches, each search and processing would need just 100ms, the whole process would need 3 days.
     
  12. Couple of thoughts.

    First, I was not convinced by the speculation that there were 300,000 active accounts. It was too close to a round number to be probable in my mind.

    Second, is it really important to know the actual number of registered accounts? I don't think so. I think measuring these numbers over periods of months is more important:

    1) rate of increase/decrease of registered users
    2) rate of increase/decrease of the fraction of registered users who are online (daily/weekly average)

    Now these two things are relatively simple to find out and would not violate the EULA, as far as I know. Use 100 carefully chosen search terms similar to what Tass mentioned at the top of the thread. It would be most straightforward to use searches that return less than 100 names (and don't pick "fad" names). If you want to know the rate of inactivation of accounts, you'd have to record not just the overall number of results returned, but the actual names, and see how many new ones have been added vs how many old ones were dropped each time period. And of course you can look at the percentage that is online. If you wanted to get really involved, you could even keep track of each avatar's online status and make estimates like .1% of players are responsible for 10% of the online time.
     
  13. RAZER

    RAZER Custom title ... uh ...

    well Neil it was not 300.000 exactly, Tass posted a screenshot somewhere and it was kindof a random number to me, so it could be that number. The most important thing about the actual number of players is that MA keeps telling us it is 1.000.000 and now we have seen just 300.000 and there is no explanation what so ever about that number we got.

    Anyway what you say could be done if we make it a team effort. Every month (?) we do a search for those names and record them in an excel sheet so we can do the math on them. Problem is we would have a % increase without a starting number (unless the 300k is indeed true).
     
  14. Hi,

    think you'll not come far other then OCR it, storing all player names in a database and "wash" dupes regularly. And repeat this often. Add time stamps, and after a while doing this you'll have a quite clear number of active participants.

    300K? Guess this ways overestimated number doesn't reflect the active population. Maybe with all the second avatars that log on once a month ...
    I'd guestimate 10 - 20K that log on at least 3 times during a week, regularly.

    Have a good time!

    Edit: Corrected a misleading sentence.
     
  15. RAZER

    RAZER Custom title ... uh ...

    Well Xandra the 300K is not something I made up, that was shown to us in the one day we had the player registry before they took it down to change it again. Remember that those 300k are not active accounts, just accounts. But as I said what happened to the 700k account to get to the million MA is advertising with all the time??
     
  16. Why can't those "1 million" simply be registered accounts like MA says? We have no idea how many people registered accounts but never made avatars, nor how many avatars with registered accounts have been inactivated over the years. Furthermore, the 1 million accounts refers, I believe, only to accounts created on Calypso. I don't understand the importance of trying to find the 700,000 "lost" players because we don't even know that there were ever that many avatars to begin with. In any case, all of these numbers far exceed the actual number of real players, and any realistic assessment of EU's playability has to look at the number of real players instead. I think Xandra's numbers are in the ballpark.
     
  17. RAZER

    RAZER Custom title ... uh ...

    Well Neil it is very easy why I wanna know where the 'missing' 700k players are. I wanna know because MA is advertising with 1.000.000 registered players and if that turns out to be just 300k or that those people never made an avatar for whatever reason they have been falsely using that number for advertisement. Everyone seeing that number would assume that there are 1.000.000 players playing the game. I know that even the 300k is way above the actual amount of people playing, but now for the first time ever (well we knew back in the early days) we have at least some numbers and they are not 1.000.000, so what I want is MA to tell me what the 300.000 means and where the 1.000.000 is coming from. I might want to send in a support case, but I know already what that will tell me.

    And to be honest, I do not think that 400.000 people made an account but NEVER ever made an avatar, why register in the first place. And sure maybe that 1.000.000 also has the inactive accounts and the banned ones, but I would like MA to tell us where the numbers come from.
     
  18. Tass

    Tass Administrator

    And that's exactly the point. They want 6 Million Dollar for 25% of Calypso's revenues, they want to be trusted but still they work like a totalitarian regime: Maximum non-transparency, censorship, nepotism and other vile practices all the way.

    They only deflect from these policies if and as much as it serves their interests. There can be only one respectively three conclusions: they are either villainous or they are simply not capable of running their business halfway ethically or they are both.

    Advertising 1 Million+ accounts is only another record in a long history of misconducts. It's like 'other' subscription games would not advertise the number of their current subcribers but the total number of people who ever had a subcription or a 14-days-free-trial account, it's like a car company would not advertise the number of sold cars but the number of people who showed up for a test drive.

    Severely unethical business. Not more, not less.
     
    • Like Like x 1
  19. Hi Razer,
    Sorry if I formulated poorly, in no way it was my intention to doubt you, nor to call it "made up". I understand that this number was shown in game, and I'll try to interpret it further below.
    Again, please excuse, I'm not a native English speaker and even if I try my best, sometimes I'm failing at actually writing what I want to say ... (Above post edited). Thx!

    Let me try to interpret these numbers:
    • the "Million accounts" is quite old, not sure, but I think this number was used already when the CE2 was introduced.
      I'd say it was the number of accounts ever made until then. Accounts made and never played, accounts made, played, left soon again, and, for sure, accounts still active.
      Back in my head I seem to remember to have seen even larger numbers, 1.25 millions of accounts, or such - not sure.
    • the 300K might be the accounts made (see above) and/or played since CE2. At this time they might have taken the opportunity to update their old databases, purging it of orphaned entries.
    • This way the lacking 700K would be accounts that didn't make into the EU after CE2.
    Speculation, for sure.

    Have a good time!
     
  20. RAZER

    RAZER Custom title ... uh ...

    No problem Xandra, I did not see that as an 'attack' on me in any way, so no worries.
    And about your theory, well that could be it, or is might be completely wrong, my point is that we should get this info from MA and not speculate about what all the numbers are.
     
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.