When will littlegolem reach 2,000,000 games? General forum
149 replies. Last post: 2019-03-23
Reply to this topic Return to forumReturn to forum
You have 0 new messages
You have 0 games on move.
You have 0 invitations to game.
149 replies. Last post: 2019-03-23
Reply to this topic Return to forumLG is over 1,900,000 games played, try to guess on which we will play the 2 millionth game ?
My guess is on 2nd of October 2018.
Should we agree on a limit date to post answers?
I propose 2018-02-28 23:59:59 as this limit date.
Also maybe we should not accept manipulation by creating tons of new games on proposed date.
Or we could go until last second with a rating system giving more points for earlier answers depending on both distance to correct answer and distance in time for the prediction ?
I don't know what the right formula for this point system would be. It seems like exponentially harder to give accurate prediction in advance, but maybe some estimators become precise after a few months of data and the months left until final time do not matter too much ?
Oh so another parameter to take into account for the rating formula is the precision of the guessing date ;)
WHAT?!
Okay okay, i'm changing my guess. I have done all the maths. the last 100k games took 14 months. Given that there is only 90k to go it has to be approximately in 14*0.9=12.6 months. My guess is christmas 2018.
TIMELIMIT WAS 2018-02-28 23:59:59
UR LATE!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
OK request granted, you have until 2018-03-31 23:59:59
And I change my prediction to 2018-12-05
Current predictions:
2018-06-05 05:06:20.18: mmKall ? on 2017-12-12
2018-08-16: Ypercube on 2018-03-01
2018-12-05: Carroll on 2018-03-02
2018-12-25: Purgency on2017-12-12
There is still room for improvement…
It would be cool to continue to play near the time when it reaches 2M.
Anyone can come up with a rating formula giving more points to earlier predictions and accepting late predictions with less points when we approach the certainty?
Where did you get that number? When I open the LG homepage I get a smaller number:1,936,963 Games
I guess The_Burglar referred to the game with highest game_id and forgot that there have been some gaps in the id sequence.
I propose as the score sqrt(days between date estimate issued and date of estimate) - (days between date of estimate and date of actual).
I accept Bill's proposition to have an inverse square law for predictions errors, with this law, here are the prediction budgets (in days) for each prediction received:
- mmKall for2018-06-05 05:06:20.1: 13.23
-Ypercube for2018-08-16: 12.96
-Carroll for 2018-12-05: 16.67
-Purgency for2018-12-25: 19.44
-The_Burglar for2018-08-08 08:08:08: 15.49
The error in days will be subtracted from this prediction budget to get a score.
So my prediction error has to be less than 19.44 in order for me to beat the dude that leaves his estimate on the very day that the gamecount hits 2 million, hence finishing with a score of 0?
Yep. I was assuming that at least one one competitor would get a score greater than zero. This system is probably not optimal for determining 2nd place, etc. Perhaps for second place finisher, the formula should double the bonus (meaning that you'd only need to be within 38.88 to beat that player).
Well, some theoretical work to backup formulas, constants would be nice.
I think the distribution of games played through time can be estimated.
The mean is easy, I have not tried to compute the standard deviation.
Knowing these figures we can compute how far an estimate may lie from the real date, 3 days before or ten days before or now and so change the bonus coefficient so that it is not way easier to get a positive score just the day before it reaches 2M…
As the distribution of games through time is a Poisson law (of varying parameter lambda, depending on new tournaments…), both the mean and the variance are lambda…
The probability distribution with Poisson l law that n games have been played on day t is given by the formula:
p = l.exp(-l.t)(l.t)^(n-1)/(n-1)! which is a Gamma law.
With my estimate of l, that yields a 95% confidence that it is between d-27 and d+27, with d best estimated date.
So the current budget is too low and I accept to multiply by 2:
mmKall for 2018-06-05 05:06:20.1: 26.46
Ypercube for 2018-08-16: 25.92
Carroll for 2018-12-05: 33.33
Purgency for 2018-12-25: 38.88
The_Burglar for2018-08-08 08:08:08: 30.98
Looks like 2018-06-05 05:06:20.1 might not be the winning date, but perhaps if I create lots of accounts and games…
;)
gamesorrys guess of 2019-01-09 01:01:09 is the clear winning candidate right now
Thanks for reminding me of Gamesorry,
So the current day budget is ( 2*sqrt(delta)).
The final score will be \|budget minus prediction error\|
mmKall for 2018-06-05 05:06:20.1: 26.46
Ypercube for 2018-08-16: 25.92
Carroll for 2018-12-05: 33.33
Purgency for 2018-12-25: 38.88
The_Burglar for2018-08-08 08:08:08: 30.98
Gamesorry for 2019-01-09 01:01:09 on2018-03-02: 35.38
Hopefully everyone (or most) will play all (or most) tournaments in August to celebrate the roman empire
Here are the new budgets errors in days ( 2*sqrt(delta)):
mmKall for 2018-06-05 05:06:20.1: 26.46
Ypercube for 2018-08-16: 25.92
Carroll for 2019-02-08: 30.79
Purgency for 2019-02-01: 30.33
The_Burglar for 2018-08-08 08:08:08: 30.98
Gamesorry for 2019-01-09 01:01:09 : 35.38
4 hours to go on my misprediction would need 10 new games every second, maybe too much for the server to handle
Need a new game every 20 seconds to reach mine (2018-08-16). Time to write a bot perhaps ;)
2019-01-19 20:19:01 is my new guess based on number of games (not game number)
Here are the new budgets errors in days ( 2*sqrt(delta)):
mmKall for 2018-06-05 05:06:20.1: 26.46
Ypercube for 2018-08-16: 25.92
Carroll for 2019-02-08: 30.79
Purgency for 2019-02-01: 30.33
The_Burglar for 2019-01-19 20:19:10: 25.53
Gamesorry for 2019-01-09 01:01:09 : 35.38
will 9 seconds earlier make a difference? :01 instead of :10 (I spot mistakes quickly, maybe from computer programming in the 80's with z80's)
I don't take the time of the day into account for the budget as I use dateutils.ddiff 2018-08-09 2019-01-19, but when we do \|budget-error\|, we can take it into account…
Well I could:
dateutils.ddiff -f “%Hh, %Mm, %Ss” -i “%Y-%m-%d:%H:%M:%S” 2018-08-09:00:00:01 2019-01-19:20:19:01
3932h, 19m, 0s
So your new budget would be 25.6005, but I don't know the time where the other predictions were made.
1973846 highest game number just 26154 more to game number 2 million
1970446 just 29554 more tooooooooooooooooooooooooo reach 2 million
Oh, it was more about 8k, I checked wrong. So 30/8 = 3.75. Still alot, if the pace is kept it would finish late december
@purgency i clicked on the game you linked - and looks as if someone is obviously running a bot v bot there…
That seems to be the case, also 800 games monthly just by those 2 players/bots
The discrepancy is mentioned in some other thread, but I don't really have the time to dig it out right now. IIRC, there was a server problem which caused quite a few game numbers to be skipped over.
geez, 9 days ago i wrote the number of games was 1971198.
How did this increase by 4k in 9 days.
At this pace it will be done in under 2 months
i was pretty sure that i did my prediction before.But since i can't find it.
My prediction is Dec 12
Here are the new budgets errors in days ( 2*sqrt(delta)):
mmKall for 2018-06-05 05:06:20.1: 26.46
Ypercube for 2018-08-16: 25.92
Carroll for 2019-02-08: 30.79
Purgency for 2018-11-05 on the 22nd of September : 13.27
The_Burglar for 2019-01-19 20:19:10: 25.53
Gamesorry for 2019-01-09 01:01:09 : 35.38
Kpato for 2018-12-12 on the 22nd of September: 18
2 new rule proposals:
First, people who specified only dates may either choose a time of day (sometime before September ends), or they will be assumed to have guessed noon on the day they selected.
Also, a clarification – all times given in predictions are treated as being in the server time zone.
i choose 13:37
Did you forget the second rule proposal or was that the clarification? :p
My prediction was standard (UTC and 00:00:00) although it doesn't matter much since it has already gone.
Yes, I suppose the constant was 2 \sqrt(Day).
And, yes, the clarification was the second rule proposal.
@Burglar right, K may be confusing. It may mean 103 = 1000 or 210 = 1024. I should have clarified.
BTW does anyone know, why the number of players on the main site is so tremendously incorrect?
What do you mean incorrect? Do you mean that there is a difference between the number of players shown and the max id of current players?
That's well known, there are some gaps (of around 986 id values missing).
SYSTEM has id = 1
Richard (the webmaster and owner of the site) has id = 1002.
Other players have ids from 1003 and above.
There are some (test) players with ids between 2 and 1001 but not for all id values.
According to the main page, the number of players is 118,092.
On the list of players:
https://littlegolem.net/jsp/info/player_list.jsp?gtvar=&filter=&countryid=&page=1
there are 84 sites with 20 players per site, so the number of players is 1,680 (actually less, as there is no full 20 players on the site no. 84).
So the number on the main page is approx. 116.000 higher (and approx. 70x higher) than the number of actual players.
@GaldianI think that list shows only “active” players. Not sure about the exact definition of “active” but probably has to do with either having games going on or recently finished or having logged in recently.
For example, my friend agoui (from Cyprus) who doesn't have any open game and his last login shows as 2016, doesn't appear in the list (easy to check, no one is listed from Cyprus).
4298 which is 2% less, but should an update be if the % difference is maybe 5, 10 or even 20 (always give 3 choices there's a 50 chance they will pick the middle one {if they did that in politics they would only spend time talking and voting})
3311 which is 43*11*7 (spot 11's divide by =301 spot subtracting 280 (multiple of 7) leaves a multiple of 7)
Do you have any statistics, how each game contributes to the total number of almost 2 million?
Yes.
@Force majeure, no but it could be estimated by taking the number of games of each type of most active players.
Congrats:
Purgency for 2018-11-05 on the 22nd of September : 13.27
you will get between 12.27 and 13.27 points depending when this occurs and no one can do better !
Luck or skill ?
Its mostly a matter of trying to win. Like, I was the only one who changed his predictions whenever the current/previous one seemed unrealistic. Frequently checking how many games were being started in like a week or a month or whatever, calculate the average games started per day and then just divide the number of games left to 2m by that average number of games started per day, to find out what approximately should be the days left until the 2m are hit. And I know that you did something similar too, Carroll, since our first two guesses were fairly close to each other. So I wouldn't call it luck, but also not really call it skill. Just the simple way to do predictions based on current trends and graphs, and then actually doing that more than once to update predictions.
I think it will be today in the US. Do I recall correctly that LG's clock setting is New York time?
There must have been a sudden surge. A couple of hours ago the pace was about 10 or 20 new games per hour. Now we're way over. So much for my plan to get a screenshot at 2,000,000. :(
Here are the final points ( 2*sqrt(delta)) - date error) :
mmKall for 2018-06-05 05:06:20.1: 26.46 - 154.5 = -128
Ypercube for 2018-08-16: 25.92 - 82.5 = -56.58
Carroll for 2019-02-08: 30.79 - 94.5 = -63.71
Purgency for 2018-11-05 on the 22nd of September : 13.27 - 1.5 = +11.77
The_Burglar for 2019-01-19 20:19:10: 25.53 - 74 = -48.47
Gamesorry for 2019-01-09 01:01:09 : 35.38 - 64 = -28.62
Kpato for 2018-12-12 on the 22nd of September: 18 - 36.5 = -18.5
Thanks to all for participating !
Now any prediction for 3 millions games?
oh, nice! we hit it. around 5pm it seems. 2030 seems like a fine guess for 3m, we better keep this place alive till then to see the day :p
Fantastic news! Massive congratulations to LG and everyone who had helped make this happen!
I'm back for the time being, btw. Also am willing to implement Squadro for the site, it would be a great fit. :)
ypercube shows the game
Start time was 2018-11-16 16:22
Next target could be when 2 million games are finished, I would expect alot weren't even started
Yes.
Maybe somebody with database skill could check how many games were concluded by game rules or resign rather than by time. An approximation is the number of games above 20 moves
I predict 3 million game will be reached at 28th February of 2022.
Shall we start a new thread for “When will littlegolem reach 3,000,000 games?“?
Yes please do and select a way to count points.
You did not specify the hour of the day, and sure it is not of the 29th ?