Skip to content
Little GolemMake your next move.

When we'll Reach 1000000 Game??

Public forum discussion.

General

#51 · MichaeI X

26 Nov 2007

<br/>                                 guess       posted<br/>-------------------------------- yyyy-mm-dd<br/>kpato .......................... 2008-10-27  2006-10-27<br/>Pawel Grabowski ................ 2007-08-11  2006-10-28<br/>Theo van der Storm ............. 2008-12-26  2006-10-28<br/>Alan Hensel .................... 2009-03-12  2006-10-28<br/>Michal Piszczalka(piszczyk) .... 2007-08-30  2006-10-29<br/>Gyorgy Csizmadia ............... 2008-10-01  2006-10-29<br/>Michael X ...................... 2008-09-10  2006-10-30<br/>Robin .......................... 2008-11-06  2006-10-30<br/>Hjallti ........................ 2008-02-04  2006-11-01<br/>Nagy Fathy ............ ........ 2008-06-10  2006-12-02<br/>bloke .......................... 2008-07-15  2006-12-03<br/>Ed Collins ..................... 2008-10-19  2006-12-05<br/>Dvd Avins ...................... 2008-03-05  2006-12-06<br/>ypercube ....................... 2008-07-24  2006-12-06<br/>Eduard ......................... 2008-08-23  2006-12-06<br/>Abigail ........................ 2008-08-08  2007-09-01<br/>FatPhil ........................ 2008-10-08  2007-09-04<br/>slaapgraag ......................2009-02-02  2007-09-04<br/><br/>                        order by guess<br/>Pawel Grabowski ................ 2007-08-11<br/>Michal Piszczalka(piszczyk) .... 2007-08-30<br/><br/>Hjallti ........................ 2008-02-04<br/>Dvd Avins ...................... 2008-03-05<br/>Nagy Fathy ..................... 2008-06-10<br/>bloke .......................... 2008-07-15<br/>ypercube ....................... 2008-07-24<br/>Abigail ........................ 2008-08-08<br/>Eduard ......................... 2008-08-23<br/>Michael X ...................... 2008-09-10<br/>Gyorgy Csizmadia ............... 2008-10-01<br/>FatPhil ........................ 2008-10-08<br/>Ed Collins ..................... 2008-10-19<br/>kpato .......................... 2008-10-27<br/>Robin .......................... 2008-11-06<br/>Theo van der Storm ............. 2008-12-26<br/>slaapgraag ..................... 2009-02-02<br/>Alan Hensel .................... 2009-03-12<br/>

Feel free to join (or review) the bet on start of the game with gid=1000000

The discussion about how to honor early vs. exact bets is still open

(and perhaps not necessary, if kpato's "i think..in 2 years" makes it anyway)

#53 · Theo van der Storm

26 Nov 2007

Eternal fame, as always :-)

Guessing right is already much less difficult than it was 13 months ago.

It should not be allowed to change one's guestimate and IMHO we should close the registration by 31-12-2007 otherwise it will become silly in the last few weeks.

Put in your bet in 2007!

#55 · Abigail

27 Nov 2007

It's not hard to write a script to make sure we will have reached 1000000 games before 2008-08-08 11:59:59. ;-)

#56 · MichaeI X

28 Nov 2007

Abigail,

it's not hard for ypercube neither...

My bet is: You won't run such a script in the first days of Aug-2008

Be aware of the eternal fame ;)

#59 · Adam Dzedzej

3 Dec 2007

April doesn't have it's guess so I'll go for Apr 30, 2008

Maybe there will be no scripts, but there may be increase of games connected with increase of guesses in this thread.

#61 · Ed Collins

5 Dec 2007

May 2008? June 2008? July 2008? You guys are way off. All indications still point to late 2008/early 2009.

Number of games one year ago: 623,479

Number of games today: 808,062

Number of games played in past year: 184,583

Number of games played each day in past year: 506

Number of games needed to reach goal: 191,938

Number of days before goal is reached at 506 per day: 379

Today's date: 12/5/2007

379 days from now: 12/18/2008

My original estimate on October 28, 2006: December 4, 2008

My revised estimate on December 5, 2006: October 19, 2008

My revised estimate today: December 18, 2008

Note: As far as determining a "winner", any guesses posted late last year should be considered completely separate from any guesses posted recently.

#62 · Theo van der Storm

7 Dec 2007

New estimate is not possible.

You wrote:

"December 5, 2006 (today) + 684 days = October 19, 2008

That's my final estimate."

#63 · Ed Collins

8 Dec 2007

Ha ha! You're not gonna hold me to something I said a year ago, are ya?

Someone should come up with a formula to determine how well of a prediction everyone made. Obviously, the formula should take into consideration WHEN how far in the future) the prediction was made.

For example, assume the winning date is December 1st, 2008. If two full years earlier Player A predicted December 5th, 2008 and Player B, in late November of 2008, predicted December 3rd, even though Player B was closer, (by two days) Player A's prediction, because it was so much further in advance, is much more impressive.

The formula would then automatically determine for me if my estimate of 10/19/2008, made a year ago, was a better prediction than Wednesday's estimate of 12/18/2008.

#64 · Theo van der Storm

8 Dec 2007

I would, actually :-)

first introduce an offset of 0.5 day, i.e.

the right day is 0.5 off

one day wrong is 1.5 off

two days wrong is 2.5 off.

divide "off" by the age of the posting at the right day.

Lowest number wins.

#65 · MichaeI X

8 Dec 2007

That's about the algorithm I had in mind as well:

Compare the relative errors:

If game 1,000,000 starts on 2008-12-01,

Ed's bet 2008-10-19 posted on 2006-12-05 has an error of 43/727 = 5.9%

An entry 2008-12-18 posted on 2007-12-05 has an error of 17/362 = 4.7%

Theo's offset would not change too much now, but assure that a correct guess two days before, would at least have an error of 0.5/2 = 25%.

Challenge for the higraders: when is the optimal time to post the probably correct time, as a function of Theo's offset currently proposed to be 0.5

Anyway, after the "million games day", we can still try to find scoring rules looking rather fair but assuring oneself a good rank ;)

#66 · Theo van der Storm

8 Dec 2007

> Theo's offset would not change too much now,

> but assure that a correct guess two days before,

> would at least have an error of 0.5/2 = 25%.

Good! A guess two days in advance is worth shit - excusez le mot.

Moreover, the formula avoids negative aspects of eternal fame.

Probably you mean:

What is the optimal time to post for winning according to Theo's formula?

However, lacking an agreed model for the difficulty of the guess the answer remains undefined.

#67 · MichaeI X

10 Dec 2007

J'excuse le mot, of course, and I agree in general.

Any scoring rule, which would not give a huge bonus to kpato's still valid bet "I think in two years" would not be accepted by me personally anyway.

"negative aspects of eternal fame" can never be avoided, IMO.

Probably I do not understand your post, Theo.

Assume player A guesses the correct date, 100 days in advance, and player B is one day off,but posted much earlier. How much earlier should B have posted, to receive the golden medal? Now the same question with errors of 1 and 2 days?

Once you decided that "What's fair" question, and with my deeply buried math "knowledge" I guess one needs two parameters and a nice formula to represent that fairness relation. Extrapolation of that formula to shorter times (like the "2 days before" in my previous post) should exclude such posts as well. Other extrapolations could show if the formula type is adequate or if there are even more parameters required.

#68 · Dvd Avins

10 Dec 2007

Even if you find your biparametric formula inadequate, a very ddiferent biparametric formula may serve. There are mnay operations one could use to combine the parameters.

#69 · FatPhil

10 Dec 2007

We need to decide if a linear or geometric scale makes more sense.

If precisely 1 year before game 1M 3 people each make their guesses as follows, which of the over-estimators should be considered to have attained an equal accuracy to the under-estimator: 0.8y, 1.2y, 1.25y.

Linearly, 0.8 and 1.2 are both 0.2 years out in opposite directions; but geometrically 0.8 and 1.25 are both a ratio of 4/5 away in opposite directions.

Personally, I think the geometric approach is the vastly better one, as it's scale invariant (the difference between 0.1y and 0.3y is huge, the difference between 5.0y and 5.2y is tiny).

#70 · Theo van der Storm

11 Dec 2007

Michael,

You can answer your own question as an exercise, because my formula is explicit.

FatPhil,

How do you resolve the issue of one or two day old correct guesses?

What's your formula?

#71 · FatPhil

12 Dec 2007

Trivial - just clip. No guesses after a certain date. That and rounding up to include the endpoints (such that the extra day that costs you becomes more and more expensive as time goes on).

#73 · MichaeI X

4 Jan 2008

Now, after Theo's deadline:

<br/>                                 guess       posted<br/>-------------------------------- yyyy-mm-dd<br/>kpato .......................... 2008-10-27  2006-10-27<br/>Pawel Grabowski ................ 2007-08-11  2006-10-28<br/>Theo van der Storm ............. 2008-12-26  2006-10-28<br/>Alan Hensel .................... 2009-03-12  2006-10-28<br/>Michal Piszczalka(piszczyk) .... 2007-08-30  2006-10-29<br/>Gyorgy Csizmadia ............... 2008-10-01  2006-10-29<br/>Michael X ...................... 2008-09-10  2006-10-30<br/>Robin .......................... 2008-11-06  2006-10-30<br/>Hjallti ........................ 2008-02-04  2006-11-01<br/>Nagy Fathy ............ ........ 2008-06-10  2006-12-02<br/>bloke .......................... 2008-07-15  2006-12-03<br/>Ed Collins ..................... 2008-10-19  2006-12-05<br/>Dvd Avins ...................... 2008-03-05  2006-12-06<br/>ypercube ....................... 2008-07-24  2006-12-06<br/>Eduard ......................... 2008-08-23  2006-12-06<br/><br/>Abigail ........................ 2008-08-08  2007-09-01<br/>FatPhil ........................ 2008-10-08  2007-09-04<br/>slaapgraag ..................... 2009-02-02  2007-09-04<br/><br/>Ruben Berenguel ................ 2008-05-15  2007-11-26<br/>Hayashi Minoru ................. 2008-07-30  2007-11-28<br/>Marcel Peperkamp ............... 2008-06-23  2007-11-29<br/>Adam Dzedzej ................... 2008-04-30  2007-12-03<br/>silexu ......................... 2009-01-27  2007-12-03<br/>Ed Collins ..................... 2008-12-18  2007-12-05  (3rd rev.)<br/> <br/><br/><br/>                       order by target time<br/>Pawel Grabowski ................ 2007-08-11<br/>Michal Piszczalka(piszczyk) .... 2007-08-30<br/><br/>Hjallti ........................ 2008-02-04<br/>Dvd Avins ...................... 2008-03-05<br/>Adam Dzedzej ................... 2008-04-30<br/>Ruben Berenguel ................ 2008-05-15<br/>Nagy Fathy ..................... 2008-06-10<br/>Marcel Peperkamp ............... 2008-06-23<br/>bloke .......................... 2008-07-15<br/>ypercube ....................... 2008-07-24<br/>Hayashi Minoru ................. 2008-07-30<br/>Abigail ........................ 2008-08-08<br/>Eduard ......................... 2008-08-23<br/>Michael X ...................... 2008-09-10<br/>Gyorgy Csizmadia ............... 2008-10-01<br/>FatPhil ........................ 2008-10-08<br/>Ed Collins ..................... 2008-10-19 (2nd)<br/>kpato .......................... 2008-10-27<br/>Robin .......................... 2008-11-06<br/>Ed Collins ..................... 2008-12-18 (3rd)<br/>Theo van der Storm ............. 2008-12-26<br/>silexu ......................... 2009-01-27<br/>slaapgraag ..................... 2009-02-02<br/>Alan Hensel .................... 2009-03-12<br/>

Of course I'm still happy with my original estimate of 2008-09-10,

but a result about 2008-12-20 would ensure hottest discussions ;)

Let's hope not to forget this thread until then ...

#74 · FatPhil

4 Jan 2008

I implied what the formula should be, it's geometric, so it's to do with ratios.

score = |guess-actual| / (earlier(guess, actual) - date of guess)

#76 · Theo van der Storm

4 Jan 2008

Sorry, too late Jordi. Michael X's list is final.

Closed, over and out, over en sluiten, finito.

#77 · Greck

4 Jan 2008

that is what a deadline is all about :P

#78 · Jordi

5 Jan 2008

Ooooh!

I can't win the big prize now!

Happy new year!

#79 · MichaeI X

5 Jan 2008

Of course you still could win, Jordi (if your guess were correct ;)

Theo will definitely ignore your claim, but the big prize of eternal fame just becomes bigger the more it is disputed.

#81 · Dvd Avins

6 Jan 2008

I hope no one delays the last game of a championship to affect this 'contest'.

#82 · slaapgraag

10 Jan 2008

i forgot there was a monster goiing on when guessing...

#83 · Robin

8 Mar 2008

@ Hjallti: you're out!

@ Dvd Avins: you're out too!

Nineteen people are still in.

Are we really going to let Ed Collins compete with two guesses? I don't think so! :)

#84 · MichaeI X

9 Mar 2008

According to Theo, it's too late for Ed to withdraw one of his two entries. So he must keep both entries and will never be an undisputed winner: If his "2nd rev" wins, he won't accept it himself, if his "3rd" one wins, don't ask Theo ;

)

Big part of the fun is NOT to have an authorised referee.

(Like in a real life streetsoccer match)

And the other interesting question is, if no one hits the date exactly, who's best ( in math ).


Just for the record: game 854000 does not exist yet...

#85 · slaapgraag

8 Jun 2008

i just found out i'm playing game 8888888

#86 · Alan Hensel

8 Jun 2008

slaapgraag: you're in game 888888, and have advance notice that you'll be in game 8888888? Wow 8-) ... Anyway, slow down! Silexu is on track to win (2009-01-27). The last 5 still have a chance, unless things speed up unexpectedly, but silexu is right on top of where it seems to be converging.

#87 · slaapgraag

11 Jun 2008

typo! beiing in this game is a vicory on its own, right?

#88 · Alan Hensel

11 Jun 2008

Yes, of course it is. In fact, split vertically, it's twelve o's, which we won't see again for a very long time.

#89 · MichaeI X

11 Jun 2008

Congrats to kuku as well! You're seldom alone in a game, here at LG.

Yes, seems nobody ran that (Abigail 2007-11-27) "game starter" script to quickly advance to 999999

#91 · Ricardo (Santos)

24 Jun 2008

... and the Day when 1000000 games will be reached is going to be 30.12.2008 (I would have said 26.12.2008 if not Theo already had choosen that day)

#93 · Alan Hensel

24 Jun 2008

Well, extrapolating from 800000 to 1000000, with 900000 as the midpoint, puts 1 million at 2009-01-28, which would mean silexu is still gliding right to the bulls-eye (way to go, Dan!), but of course that assumes that the next 100K games will go by at the same rate as the last 100K. Is LG getting more popular, or less popular? Or, since the last 100K went by in about 7 months, maybe seasonal variations are more important than overall site popularity? Then again, it would be a winter-to-summer period versus a summer-to-winter period, and what difference would spring versus autumn make?

Anyway, the betting is closed for 1M, but it looks like we have our first bet for 2M :-)

#94 · KPT

26 Jul 2008

the end is coming!!!

#95 · slaapgraag

27 Jul 2008

kpato, i know you're a christian, but that's off-topic! ;-)

#96 · Alan Hensel

27 Jul 2008

The 1 millionth game is a sign of the Apocalypse?! Play more slowly, people!

#97 · KPT

28 Jul 2008

nice one LB hehe

#98 · Ed Collins

25 Sep 2008

slaapgraag's guess of 2009-02-02 is looking best to me right now, based upon Alan's comment that 900,000 was reached on June 24th. silexu is a very close second.

Speaking of Alan, he may end up winning it, and notice that his guess was about eleven months earlier than slaapgraag's!

It's clearly obvious by now that nearly EVERYONE undershot the mark, some players by more than a year! Interesting.

#99 · Alan Hensel

25 Sep 2008

It's obvious that there were a lot of optimists (a.k.a. "suckers" ;-)

It looks like both silexu and slaapgraag assumed the same rate of playing and did a simple linear extrapolation. Or made lucky guesses :-) Just about everyone else apparently assumed LG would get more popular, and I was the only pessimist, figuring that people would be distracted a bit by new technologies and sites and play less on LG. It now seems that the linear assumption was best.

#100 · Greck

26 Sep 2008

According to I.M. Annedict in his article "Golemoholics and long-term exposure to turn-based servers" in Trans. Amer. Psic. Soc. (3) 2008, few of us got ever distracted from our "duty" (a bit down in holidays, but that's all) :)

Last page