The REAL Beta Numbers Thread

As everyone knows, there is a lot of confusion and misinformation getting thrown around about the numbers of accounts and beta members. In this post, I'm going to list facts about the invite system and its history. My next post will have estimated amounts based off the dates and timer intervals, to the best of my knowledge.

Each date is a link to the appropriate announcement thread. Any missing information from timelines is appreciated, but please have a link for reference.

At the time of this post:
133,99x accounts
Unknown number of accounts with beta access
10,798 Default League characters
2,696 Hardcore League characters

17 October 2011 Edit:
146,97x accounts (~618 accounts created per day since above numbers)
14,552 Default League characters
3,227 Hardcore League characters
It appears they have fixed the search issue with accounts, so we can now see an accurate number instead of rounding.

7 November 2011 Edit:
152,502 accounts (~260 accounts created per day since last edit)
16,689 Default League characters
3,465 Hardcore League characters

17 November 2011 Edit:
154,596 accounts (~209 accounts created per day since last edit)
4,323 Default League characters
830 Hardcore League characters
As the new patch is only 3 days old, characters per league is not yet up to numbers that are representative of the beta tester pool.

1 December 2011 Edit:
162,192 accounts (~543 accounts created per day since last update)
14,497 Default League characters
2,499 Hardcore League characters

15 December 2011 Edit:
168,497 accounts (~450 accounts created per day since last update)
23,278 Default League characters
3,609 Hardcore League characters

15 January 2012 Edit:
179,217 accounts (~346 accounts created per day since last update)
13,798 Default League characters
23,914 Legacy League characters
2,455 Hardcore League characters
3,109 Hardcore Legacy League characters

1 February 2012 Edit:
188,026 accounts (~518 accounts created per day since last update)
22,203 Default League characters
3,617 Hardcore League characters

14 February 2012 Edit:
197,124 accounts (~700 accounts created per day since last update)
30,700 Default League characters
4,707 Hardcore League characters

2 March 2012 Edit:
207,970 accounts (~638 accounts created per day since last update)
41,617 Default League characters
6,021 Hardcore League characters

24 March 2012 Edit:
221,000 accounts (~592 accounts created per day since last update)
52,835 Default League characters
7,663 Hardcore League characters

1 April 2012 Edit:
254,588 accounts (~4,199 accounts created per day since last update)
The public stress test created thousands of new characters that are not indicative of beta accounts. Until a Legacy migration occurs, characters per league will no longer be tracked.

11 April 2012 Edit:
281,270 accounts (~2,668 accounts created per day since last update)
88,029 Default League characters
8,790 Hardcore League characters

25 April 2012 Edit:
301,912 accounts (~1,474 accounts created per day since last update)
106,370 Default League characters
11,489 Hardcore League characters

29 May 2012 Edit:
370,620 accounts (~2,021 accounts created per day since last update)
95,675 Default League characters
7,082 Hardcore League characters

14 June 2012 Edit:
382,739 accounts (~758 accounts created per day since last update)
106,546 Default League characters
7,894 Hardcore League characters

10 July 2012 Edit:
412,142 accounts (~1,131 accounts created per day since last update)
133,955 Default League characters
9,719 Hardcore League characters

25 July 2012 Edit:
433,433 accounts (~1,419 accounts created per day since last update)
151,660 Default League characters
10,658 Hardcore League characters

31 July 2012 Edit:
471,636 accounts (~6,367 accounts created per day since last update)
221,335 Default League characters
12,885 Hardcore League characters
Many new characters are due to the public stress test and are therefore not indicative of the true number of beta accounts.

27 October 2012 Edit:
564,780 accounts (~1,047 accounts created per day since last update)
18,797 Default League characters
3,711 Hardcore League characters


Invite Timer Timeline:
  • Aug 10
    Invite timer begins selecting accounts at midnight (PST) on a 20min interval; reduced to 5min within the first hour.
  • Aug 10
    Forum posters get a 10x boost to their chances to be selected by the timer.
  • Aug 10
    Timer interval increased to 10min (was 5min).
  • *Aug 11*
    Stealth timer increase to 15min with no announcement.
  • Aug 12
    Veteran Timer is activated. It selects accounts that were active posters before the beta began. Both timers select 1 account each on a 30min interval (was 15min for normal timer).
  • Sep 2
    Veteran timer is removed; normal timer remains at 30min. Friend Invites start: beta members are selected randomly to receive a Friend Invite key. This selection is at a faster interval than the Veteran timer (30min), but exact interval is unknown (at least 1 key/30min).
  • Sep 8
    Timer reduced to 25min (was 30min).
  • Sep 19
    Friend Invites selection rate is doubled, exact interval still unknown (is at least 2 keys/30min).
  • Sep 30
    Chris announces version 0.9.3 with the promise of "ramping up" invites when it goes beta. The exact changes are still unknown.
  • Nov 6
    Daylight Savings Time throws the timer into a fit, causing it to iterate nearly instantly and invite around 100 extra accounts in a short amount of time.
  • Nov 13
    With the beta release of 0.9.3, the timer is reduced to 5min intervals.

  • *Sometime after Sep 8, the timer was increased to 35min, then down to 25min.*


    Known Keys Distributed:
    • 20 Aug
      Ladder Race, top placers received keys.
    • 26-28 Aug
      PAX Prime, GGG hands out 2,400 DVD's with beta keys.
    • 3 Sep
      Ladder Race, top placers received keys.
    • Nov 14
      GGG sends out an email notifying beta members of the new patch and invites them to come back and test it out. The email was erroneously sent to around 2,100 non-beta members. These accounts were granted beta access. The email also states that 10,000 friend invites will be given out over the next 2 weeks (until the 21st, presumably). This breaks away from the previously-assumed trend that friend invites were being sent out at a rate approximately twice as fast as the invite timer.

    *Numerous keys have been given out via Twitter, Facebook, and random forum posts. Keys have also been given to streamers and gaming sites. These promotional keys number in the thousands, but exact data is unknown.*


    Non-invite Announcements:
    • Aug 9
      Chris posts the Beta Invite Email so the community can see the beta testing policies.
    • Aug 10
      Chris suggests that users refrain from useless posting as spam doesn't increase chances of entering beta. States that there are 392 beta members.
    • Aug 15
      Chris posts a very informative thread dealing with the complaints about the invite system. States that there are 1,250 testers after 5 days of invites.
    • Aug 15
      GGG starts a 3-day auction for 10 beta keys on ebay, with all proceeds going to the Child's Play charity. Ebay cancels the auction due to policy.
    • Nov 3
      Chris reiterates that beta invites will be increased once 0.9.3 goes beta, and says he wants to keep 500+ concurrent testers online.
    • Nov 29
      Chris states that 13,000 friend invites have been given out since the 0.9.3 announce email (Nov 14).
    • Apr 10
      Chris announces a press release stating that over 13,000 accounts have purchased support packages. For the purpose of beta key tracking, I'm going to use 1 key per purchase.
    • Jul 25
      Chris states that 50,000 accounts have pre-purchased. That's 37,000 more than the last known number.
  • Closed Beta/Alpha Tester back after a 10-year hiatus.
    First in the credits!
    Last edited by WhiteBoy#6717 on Oct 27, 2012, 5:43:07 PM
    The Numbers
    I've take each segment of the timeline and broken them down to get a rough calculation on the number of timer invites and friend invites that have been distributed. For short periods of time (less than 2 days), I've calculated to the hour, rounded. For longer periods, I've calculated to the day. Take these with a grain of salt; they are most likely far less than the actual values due to rounding and many unknown variables.

    Invites Over Time:
    • Aug 10 2am-10pm
      1 invite every 20min for less than 1 hour = 2-3 invites
      19 hours = 1,140 minutes
      1,140min / 5min interval = 228 intervals
      1 invite/interval * 228 intervals = 228 invites
    • Aug 10, 10pm - Aug 12, 1:45am
      Timer gets changed from 10-15min at unknown time, average of 12.5min.
      ~28 hours = 1,680 minutes
      1,680min / 12.5min interval = ~134 intervals
      1 invite/interval * 134 intervals = 134 invites
    • Aug 12 - Sep 2
      2 invites every 30min
      21 days = 504 hours = 30,240 minutes
      30,240min / 30min interval = 1,008 intervals
      2 invites/interval * 1,008 intervals = 2,016 invites
    • Sep 2 - Sep 8
      6 days = 144 hours = 8,640 minutes
      8,640min / 30min intervals = 288 intervals
      1 invite/interval * 288 intervals = 288 invites
    • Sep 8 - Sep 26
      Timer gets changed from 25min to 35min to 25min, average of ~30min.
      18 days = 432 hours = 25,920 minutes
      25,920min / 30min interval = ~864 intervals
      1 invite/interval * 864 intervals = 864 invites
    • Sep 26 - Oct 17
      Update for such a long time span. Timer still at 25min.
      21 days = 504 hours = 30,240 minutes
      30,240min / 25min interval = ~1,210 intervals
      1 invite/interval * 1,210 intervals = 1,210 invites
    • Oct 17 - Nov 7
      Update for another 3-week time span. Timer still at 25min.
      21 days = 504 hours = 30,240 minutes
      30,240min / 25min interval = ~1,210 intervals
      1 invite/interval * 1,210 intervals = 1,210 invites
    • Nov 7 - Nov 13
      6 days = 144 hours = 8,640 minutes
      8,640min / 25min interval = ~346 intervals
      1 invite/interval * 346 intervals = 346 invites
    • Nov 13 - Nov 17
      Timer reduced to 5min intervals for 0.9.3 beta release.
      4 days = 96 hours = 5,760 minutes
      5,760min / 5min interval = 1,152 intervals
      1 invite/interval * 1,152 intervals = 1,152 invites
    • Nov 17 - Dec 1
      14 days = 336 hours = 20,160 minutes
      20,160min / 5min interval = 4,032 intervals
      1 invite/interval * 4,032 intervals = 4,032 invites
    • Dec 1 - Dec 15
      14 days = 336 hours = 20,160 minutes
      20,160min / 5min interval = 4,032 intervals
      1 invite/interval * 4,032 intervals = 4,032 invites
    • Dec 15 - Jan 15
      31 days = 744 hours = 44,640 minutes
      44,640min / 5min interval = 8,928 intervals
      1 invite/interval * 8,928 intervals = 8,928 invites
    • Jan 15 - Feb 1
      17 days = 408 hours = 24,480 minutes
      24,480min / 5min interval = 4,896 intervals
      1 invite/interval * 4,896 intervals = 4,896 invites
    • Feb 1 - Feb 14
      13 days = 312 hours = 18,720 minutes
      18,720min / 5min interval = 3,744 intervals
      1 invite/interval * 3,744 intervals = 3,744 invites
    • Feb 14 - Mar 2
      17 days = 408 hours = 24,480 minutes
      24,480min / 5min interval = 4,896 intervals
      1 invite/interval * 4,896 intervals = 4,896 invites
    • Mar 2 - Mar 24
      22 days = 528 hours = 31,680 minutes
      31,680min / 5min interval = 6,336 intervals
      1 invite/interval * 6,336 intervals = 6,336 invites
    • Mar 24 - Apr 1
      8 days = 192 hours = 11,520 minutes
      11,520min / 5min interval = 2,304 intervals
      1 invite/interval * 2,304 intervals = 2,304 invites
    • Apr 1 - Apr 11
      10 days = 240 hours = 14,400 minutes
      14,400min / 5min intervals = 2,880 intervals
      1 invite/interval * 2,880 intervals = 2,880 invites
    • Apr 11 - Apr 25
      14 days = 336 hours = 20,160 minutes
      20,160min / 5min interval = 4,032 intervals
      1 invite/interval * 4,032 intervals = 4,032 invites
    • Apr 25 - May 29
      34 days = 816 hours = 48,960 minutes
      48,960min / 5min interval = 9,792 intervals
      1 invite/interval * 9,792 intervals = 9,792 invites
    • May 29 - Jun 14
      16 days = 384 hours = 23,040 minutes
      23,040min / 5min interval = 4,608 intervals
      1 invite/interval * 4,608 intervals = 4,608 invites
    • Jun 14 - Jul 10
      26 days = 624 hours = 37,440 minutes
      37,440min / 5min interval = 7,488 intervals
      1 invite/interval * 7,488 intervals = 7,488 invites
    • Jul 10 - Jul 25
      15 days = 360 hours = 21,600 minutes
      21,600min / 5min interval = 4,320 intervals
      1 invite/interval * 4,320 intervals = 4,320 invites
    • Jul 25 - Jul 31
      6 days = 144 hours = 8,640 minutes
      8,640min / 5min interval = 1,728 intervals
      1 invite/interval * 1,728 intervals = 1,728 invites
    • Jul 31 - Oct 27
      89 days = 2,136 hours = 128,160 minutes
      128,160min / 5min interval = 25,632 intervals
      1 invite/interval * 25,632 intervals =25,632 invites

    Total of estimated timer invites:
    more than 107,096


    Friend Invite Keys:
    • Sep 2 - Sep 19
      17 days = 408 hours = 24,480 minutes
      24,480min / 30min interval = 816 intervals
      1 key/interval * 816 intervals = 816 keys
    • Sep 19 - Sep 26
      7 days = 168 hours = 10,080 minutes
      10,080min / 30min interval = 336 intervals
      2 keys/interval * 336 intervals = 672 keys
    • Sep 26 - Oct 17
      21 days = 504 hours = 30,240 minutes
      30,240min / 30min interval = 1,008 intervals
      2 keys/interval * 1,008 intervals = 2,016 keys
    • Oct 17 - Nov 7
      21 days = 504 hours = 30,240 minutes
      30,240min / 30min interval = 1,008 intervals
      2 keys/interval * 1,008 intervals = 2,016 keys
    • Nov 7 - Nov 13
      6 days = 144 hours = 8,640 minutes
      8,640min / 25min interval = ~346 intervals
      2 keys/interval * 346 intervals = 692 keys
    • Nov 14 - Nov 29
      Chris stated the number of friend invites given in this time period to be 13,000. As this is not in line with previous assumptions of the friend invite system's methods, I will no longer give estimates in the same fashion. This may change if more info is made public.

    Total of estimated Friend Invite keys:
    more than 19,212 (As of Nov 29, 2011)

    Total of all estimated invites/keys plus known distributed keys:
    more than 176,308 possible beta accounts
    (approx. 31% of all accounts)


    *Since "thousands" of promotional keys, non-tracked friend invites, and pre-purchase keys have been given out, the number of possible beta accounts is actually much higher than the total estimate shown above. I would personally estimate these to number in the tens of thousands.*



    How the timer (likely) selects accounts:
    Spoiler
    Skip to bottom spoiler for the simplest, no-math explanation.

    People throw around the term "RNG" (Random Number Generation) quite a bit. I thought it might be useful to explain how the random selection probably works.
    To read up on RNG, check here. The theory is simple: the computer pseudo-randomly picks a number based off an input number (seed) run through an algorithm. Computers are logical, so true randomization is impossible to achieve. The seed is necessary to give the computer something to calculate, and it can be derived from many things.

    To begin, we're going to give all accounts a numerical designation. To keep it simple, we can just sort them in the order they were created, starting with 0 as the first (nearly everything in computer mathematics uses 0 as the first digit). This order has nothing to do with the selection, since the RNG could pick any number equally.
    0 Bob
    1 Joe
    2 Billy
    3 George
    4 John
    5 Michael
    6 Steve
    7 David
    8 Tom
    9 Ed


    Now, we take this list and throw it into our code, which tells the computer to run the algorithm to select a number from 0-9, giving the corresponding account beta access. For each selection, the timer code has to filter out all beta member accounts and assign new numbers to each remaining account. If no new accounts are created, everyone's chances improve with each selection.

    That is the very basic example.

    To weight accounts as more likely to be picked, they are simply added to the list more than once. From the list above, if we wanted Bob to have twice the chance everyone else has, we would simply list him twice before assigning numbers:
    0 Bob
    1 Bob

    2 Joe
    3 Billy
    4 George
    5 John
    6 Michael
    7 Steve
    8 David
    9 Tom
    10 Ed

    Bob's chance goes from 1/10 to 2/11, or .1 to .181818. His chances are now twice that of each other person. They each went from 1/10 to 1/11, or .1 to .090909. Keep in mind that we're not attempting to double Bob's chance, only to give him twice the chance as each other person.

    What if we have more than one person getting the same weight? Let's put Joe in the same 2x chance category as Bob.
    0 Bob
    1 Bob
    2 Joe
    3 Joe

    4 Billy
    5 George
    6 John
    7 Michael
    8 Steve
    9 David
    10 Tom
    11 Ed

    Bob's chances are now 2/12, or .166667, along with Joe's. Each other person's chances went from 1/11 to 1/12, or .083333. As you can see, Bob and Joe still have twice the chance as everyone else, but Bob's overall chances have been lowered due to Joe being weighted in the same category.

    Spoiler
    The simplest explanation of the name-selection process, without lists and math:
    • Assume names are being drawn out of a hat.
    • Each account is put into the hat once.
    • Accounts that are "forum users" are put into the hat 10 times instead of just once.

    I used a longer description above to mathematically show that the 10x weighted chance doesn't actually give you a +10% chance (as many people are saying), nor does it give you ten times your base chance before posting on the forum.

    If we knew the number of active accounts and the number of beta accounts, we could apply this method to calculate the actual chances of active accounts versus inactive accounts. Unfortunately, this is unknown. Calculating from assumptions won't do us any good, either.



    Closed Beta/Alpha Tester back after a 10-year hiatus.
    First in the credits!
    Last edited by WhiteBoy#6717 on Oct 27, 2012, 5:48:39 PM
    Hmmm good job
    (╯°□°)╯︵ ┻━━━━━━━━━━━━━━━━┻
    I FLIP WITH THE BIG BOYS
    odds are looking rather good, I'll keep waiting.
    Forgot to add that the invite timer will be getting a shorter interval in the near future, when they push out the 0.9.3 patch. That's good news for everyone still waiting!
    Closed Beta/Alpha Tester back after a 10-year hiatus.
    First in the credits!
    Awesome thread whiteboy, thanks a bunch for these numbers. So everyone can see things aren't so grim.
    These odds are NOT encouraging honestly.

    Barely over 10,000 characters for definitely over 7k keys? That means that people really aren't using their invites as we thought.
    It doesn't take into account any deleted characters, or the fact that many testers may have stuck with or quit after playing one character. It also doesn't consider that of those 7,400+ possible accounts, many of the friend invites and PAX keys may not have been activated yet.

    I would estimate the publicity keys (ones that went to gaming sites and streamers) to total in the hundreds, so the possible accounts might be 8,000+. If we consider that as a safe estimate, then 8,000/134,000=~6% of all accounts could be beta members. That's not really all that bad, since this is being run as a true beta test and not a demo or stress test. GGG says most of the invited accounts are being used, and we have no evidence that suggests otherwise.

    Like I said above, there are many unknown variables involved that could swing these numbers in any direction.
    Closed Beta/Alpha Tester back after a 10-year hiatus.
    First in the credits!
    Last edited by WhiteBoy#6717 on Sep 26, 2011, 3:18:25 AM
    Still a fair few amount of players. Still dont think all the keys sent out via the timer would be used.

    Cant believe Ebay took the auction down. I could understand if it was a normal person trying to sell beta keys and the company asked ebay to take them down. But to be taken down when its the actual company selling them and for a charity as well, seems rather stupid if you asked me.
    Well you must take in common that half of those who received key dont even play. They either dont have time or totaly forgot about the game.
    Keysus- His power is infinite!
    Last edited by Snagod#2442 on Sep 26, 2011, 3:55:28 AM

    Report Forum Post

    Report Account:

    Report Type

    Additional Info