Is this a true quality game?

I've just started playing Einstein and think it is great fun, but I reckon I've pretty much sussed out the optimum strategy and calculating the odds for each individual move is MUCh simpler than another similar type of game, Backgammon.

Is this really a quality game for LG or just a bit of fun like Street Soccer? I think once a player has more or less worked out the optimum strategy you've got about a 50/50 chance to win, whereas in Backgammon, very skillful players retain a, say, 55/45 advantage over less experienced players which they use over long matches to win.

IMHO, the board either needs to be extended to have 10 or even 15 stones or you need to play minimum 9 games matches to make it a truer test of skill on a par with the high quality games here like Chess/Go.

You can find some theoretical stuff on "EinStein wurfelt nicht" in the report

http://www.minet.uni-jena.de/preprints/althoefer_06/rockNroll.pdf

written by computer science student Andreas Schaefer.

Currently best computer programs and best humans play

EinStein-w-n at about equal strength. Recently there

was a 161 [in words "onehundred-and-sixty-one"] game match

on the server www.inetplay.de between top human Luise

Reinhardt (=luiser) and top computer program

MeinStein_c (written by Dutchman Theo van der Storm).

Luise won the match by 84-77.

Both made their moves typically within seconds. The whole

match took less than two and a half hours.

The notation of all 161 games may be found in the

forum

http://forum.inetplay.de/viewtopic.php?topic=1808&forum=23&1

in the thread

MeinStein - Luiser games (please ...)

by Theo van der Storm.

(Observe that in this thread a different notation from the

one on LittleGolem is used.)

Ingo Althofer.

Thanks for that response. I'll have a look at the website tomorrow.

Typically computer programs play games that are exceedingly complex (like Go) or statistical (like backgammon) rather poorly (though naturally this is becoming less true with time).

If computer programs already play this very new game on equal strength with humans and this is a game of statistical analysis and gut instinct (similar to backgammon), then does that support my argument that Einstein is a clever and fun game, but would not be suitable for, say, a serious tournament involving prize money?

> If computer programs already play this very new game

EinStein is not completely new. It was designed in

August 2004 and came to market early in 2005.

The first computer program was developed by

Andreas Schaefer during the winter semester 2004/05

at Jena University. See for a report (in German) at

http://www.computerschach.de/index.php?option=com_content&task=view&id=123&Itemid=111

In April 2005, the game was presented on the (German)

game server www.inetplay.de and has attracted a lot

of attention there.

> on equal strength with humans

The story is somewhat different:

In Summer 2005 a little handful of

computer programs existed, and many people in the

scene thought that computers were superior to human

players.

And then, inspired by an EinStein tournament in

November 2005 in public, Luise Reinhardt and a few

others found ways to turn the table (or at least, to

bring it back into balance).

> and this is a game of statistical analysis and gut

> instinct (similar to backgammon), then does that

> support my argument that Einstein is a clever and

> fun game, but would not be suitable for, say, a

> serious tournament involving prize money?

I agree concerning the aspect of price money.

However, a serious 5-hour tournament (without

price money) has been played, see

http://www.computerschach.de/index.php?option=com_content&task=view&id=365&Itemid=205&limit=1&limitstart=1

Ingo Althofer.

Hi Ingo,

I've had a read of the students' project to write the computer program. Very intersting and informative indeed! It was only after starting to read that thesis that I realised I was actually conversing with the games' inventor!

Congrats on a clever and fun game. I've written elsewhere in this forum about what I consider to be an easily explained and easily understood "optimum" strategy in just 3 points. I was wondering if you had any comments on that and on my suggestion of:

1) Playing Einstein matches over longer runs (.ie. 9 or 11 games, say, instead of 5);

2) Having a stone placement phase where you each place one stone until all 6 are placed.

Does having 10 or 15 stones on a 6x6 or 7x7 board fundamentally change the games as well? Someone has suggested it might actually make the game more lucky with a larger board and another has suggested that the game would lose some of its simplicity and "fun factor" as well. Interesting...

Good luck with marketing the game.

I already commented on 2) stone placement phase in the "first move" thread.

Regarding strategy: There is (already) more to this game than you think!

Anyway, I take this game seriously eventhough the luck factor could be reduced by increasing the size. Would it be as much fun, is the real question.

Now I should say I appreciate the introduction of this game to this and perhaps also other websites very much. I'm sure it takes a lot of programming effort to support new games. The concept of a tournament is not present on the http://www.inetplay.de website. It is on LittleGolem: excellent!

Generally speaking: The stronger both players the more games you need in a match. This is something to consider for playing the final of a tournament.

A match between two players should always have an even number of games, because it is an advantage to start a game.

When playing qualifying rounds it is important to keep all statistical game information contained in the calculations until the decision is made who goes on, otherwise the luck factor will dominate the result. Winning 3-0 is much better than winning 3-2. So it's best to add up all game results in a qualifying round, i.e. not scoring 0-8 won matches, but scoring 0-40 won games for each player. There are several tie-breaker rules possible.

Theo

A reply to Alain:

> ... It was only after starting to read

> that thesis that I realised I was actually

> conversing with the games' inventor!

> Congrats on a clever and fun game.

Thanks for the flowers.

> I've written elsewhere in this forum about what

> I consider to be an easily explained and easily

> understood "optimum" strategy in just 3 points.

> I was wondering if you had any comments on that

Your proposal is rather far from optimal. My

impression is that 5x5-EinStein is not well

solvable in general by the current computer

generation.

> and on my suggestion of:

> 1) Playing Einstein matches over longer

> runs (.ie. 9 or 11 games, say, instead of 5);

Good idea. On www.inetplay.de some

hardcorers prefer 161-game matches (right, Theo?).

> 2) Having a stone placement phase where you

> each place one stone until all 6 are placed.

This was my original formulation when I designed

the game in August 2004. In the meantime I learned

(mainly by the inetplay esperiences) that the

random setups give the game more substance.

> Does having 10 or 15 stones on a 6x6 or 7x7

> board fundamentally change the games as well?

I think so. Recently I made an extended computer

autoplay experiment with 6 stones vs 6 stones.

On board size 5x5 strong players typically beat

(in the average) the same number of own and

foreign stones.

On quadratic board size 4x4 or 3x3 beating the

opponent happened much more frequent.

On quadratic board size 6x6 or larger beating

own stones happened much more frequent than

beating the opponent's stones.

So, 5x5 seems to be an optimum concerning the

balance between self-capture and opponent-

capture (and this balance makes a great part

of the game's quality in my eyes).

> Someone has suggested it might actually

> make the game more lucky with a larger board

maybe.

> and another has suggested that the game would

> lose some of its simplicity and "fun factor"

> as well. Interesting...

Years ago I was a serious tournament chess players

- with lots of games that took 5 or more hours.

Now, when I have become old I prefer quickies.

> Good luck with marketing the game.

Thx for the wishes - and have fun with

EinStein on LittleGolem.

Ingo.

Ingo wrote: "On www.inetplay.de some

hardcorers prefer 161-game matches (right, Theo?)."

Some info for the littlegolem people:

In on-line EWN play people usually play between 10 and 30 games

at a pace of one move in 4 seconds average (Luise R. under 3s!).

I prefer to play matches between 50 (against top players) and 20 (against others).

Surely a 161-game match is an absurdity as you probably meant :-)

Theo

First, thanks for your compliments, Ingo and Theo. But how come you consider me best human player... :-)

I like that game very much because it is quick and there is yet no perfect strategy. On Inetplay there are very good players and each plays different.

I hope some of you joins us on Inetplay because it is great fun to play 20 quick matches in a row, or 161 or even 300. :-D

I am not sure if LG is the right stage for this game...

But I am very interested how the programs perform with more time to calculate. I hope Meinstein_c and fraggle_c can play here.

Luise

Luise,

I don't remember ever writing that you are the best human player :-)

We don't know.

The only way to find out is by playing tournaments (on-line or correspondence)

in a closed group with plenty of games.

I would consider humans:

LuiseR, thstorm, Mentor, Fraggle, Kitaktus, Pinsel, Opmp.

Of course Ingo should join us to make the group even :-)

For Ingo's information:

Luise and alias Rorymaus have also lost long matches against MeinStein_c.

Thank you for reminding me, Theo. :-D

But I'm still the fastest! :-))

The players you've named are very strong. Yahweh and uliko too. I don't think you can pick out one who is best.

And I agree that MeinStein is too good for me. At the moment :-)

Luise

I don't know how uliko plays. Not seen enough.

Sorry, Yahweh does not belong in this group of seven (or 8-).

Yo!

Ingo, please note that MeinStein's AMD64 Athlon is glowing with pride.

quoting the rules:

"Special rules on LittleGolem: Game is played as 3 point match."

See my earlier post on the 22nd.

In fact there is another special rule in effect on LittleGolem.

However, this rule is not listed:

"The loser of a gamepoint starts the encounter for the next gamepoint."

It is an advantage to start, so this rule introduces a rating imbalance

compared to the normal rule: "players start games alternately".

IMHO it's a bit childish rule to offer such an advantage.

Games will tend to be closer than justified by the difference in strength.

Hi Theo,

is MeinStein already playing some tournaments? BTW, I plan to let fraggle_c play some games. Perhaps we meet each other... ;-)

Alain wrote:

"I've pretty much sussed out the optimum strategy and calculating the odds for each individual move is MUCh simpler than another similar type of game, Backgammon.

Is this really a quality game for LG or just a bit of fun like Street Soccer?"

Heh :) The skill factor in StreetSoccer is very much higher than in Backgammon. See ratings on other sites with both games.

I don't mean to disagree with you, Cwali, but on what grounds do you argue that "The skill factor in StreetSoccer is very much higher than in Backgammon"? I'm curious as to what data you base this conclusion on, and what analysis is behind suh a statement. Me, I have done no such analysis, so I don't know, but I would very much like to see what is done, as I have a glowing interrest for statistics and probability calculations.

The statement made by FC Cwali isn't exactly true in my opinion. Perhaps if you look at one single game of Streetsoccer compared to one single game of backgammon, I might agree. But backgammon usually is played as a match towards a certain (usually odd) number of points, where each victory gives you one (or more, depending on the doubling cube) point. And this doubling cube and its use is what makes the difference between skilled and not-so-skilled players in backgammon. In my opinion, making really good use of this requires a higher skill level than the skill level needed for playing Streetsoccer. But it's just my opinion, my gut-feeling, and I have no real arguments to back this.

Still, I agree with FC Cwali that alain's remark that 'Streetsoccer is just a bit of fun' is totally bogus! Yes, it's a really fun game, but how can you explain that FC Cwali has won 5 out of 7 championships here if it was not a game of skill?

I've always compared Streetsoccer to backgammon: in both games new players (and not so new!) think it's just about luck, and complain when someone gets "the number he just needed", not thinking that the good players put the pieces on the board so they get the most chances of getting a good number, and minimize the effect of bad rolls.

Thanks cwali for a game that's fun and makes you think!

This link

http://www.minet.uni-jena.de/preprints/althoefer_06/rockNroll.pdf

mentioned in this topic above, appears to now be invalid. Does anyone have another link to this paper?

Hello Ed,

for me the link works. However, it takes a few seconds,

and it starts on page 2 (of 46), which is empty.

Let me know if a second try works with you.

Regards, Ingo

@Ed: Try right-click and "Save link as...": http://www.minet.uni-jena.de/preprints/althoefer_06/rockNroll.pdf

I can't see the pdf in Firefox either but it's saved normally on disk.

Hello,

it opens with my Firefox 4.0.1 (under Windows XP home),

and ypercube's trick with the right-click + save does

work as well.

Ingo.

For some reason, it worked fine for me this time.

Thanks!