Author |
Topic: Essay by Christian Freeling on inventing games (Read 522276 times) |
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender: 
Posts: 804
|
 |
Re: Essay by Christian Freeling on inventing games
« Reply #90 on: Apr 30th, 2009, 12:25pm » |
|
I said the game would behave properly, I'm not sure about the program 
Its Lion is now under direct threat (and not merely an exchange), so it may be time to get rid of the ball.
Zillions (bèta) - Christian
position after black-9
(download Zillions HanniBall bèta 0.3)
|
1. Hh4-f5/Hf4-g6/Hb4-d5
2. Hf5-f7/Hd4-c6
3. Hf7-e9/H*e9-c5
4. Lc2-d4/*c5-d4/L*d4-c4
5. Hc5-d7/*c4-e3/Ke1-e3
6. *e3-b6/Lc4-b6/*b6-c6
7. *c6-b6/Hd5-c7/*b6-c6
8. *c6-b6/Hc6-b8/L*b6-a6
9. L*a6-c5/Ef3-f4/L*c5-a6
* = ball
rules
|
Hh14-f10/Hb14-d13
Eh15-f13/Hd14-e12
Lg16-e11
Hf10-e8/He12-d10/Ed15-c14
Hf14-f10/Ec14-b13
Eb13-b10
Hd13-c10/Lc16-d14
Ef13-d11/Le11-d9
Hc10-e6/Ed11-c10
^ = ricochet
history
|
|
|
 IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #91 on: May 1st, 2009, 12:00am » |
|
It's digging in - this is where a mix of Elephants and Horses comes in handy.
Zillions (bèta) - Christian
position after black-10
(download Zillions HanniBall bèta 0.3)
|
1. Hh4-f5/Hf4-g6/Hb4-d5
2. Hf5-f7/Hd4-c6
3. Hf7-e9/H*e9-c5
4. Lc2-d4/*c5-d4/L*d4-c4
5. Hc5-d7/*c4-e3/Ke1-e3
6. *e3-b6/Lc4-b6/*b6-c6
7. *c6-b6/Hd5-c7/*b6-c6
8. *c6-b6/Hc6-b8/L*b6-a6
9. L*a6-c5/Ef3-f4/L*c5-a6
10. La6-c5/Lc5-a6/Hd7-c5
* = ball
rules
|
Hh14-f10/Hb14-d13
Eh15-f13/Hd14-e12
Lg16-e11
Hf10-e8/He12-d10/Ed15-c14
Hf14-f10/Ec14-b13
Eb13-b10
Hd13-c10/Lc16-d14
Ef13-d11/Le11-d9
Hc10-e6/Ed11-c10
Ec10-c8/Ld14/c12
^ = ricochet
history
|
|
|
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #92 on: May 1st, 2009, 2:03am » |
|
After *c5, H*e4, white can ricochet off the Keeper, so I'll have to cover the right flank to a degree.
Zillions (bèta) - Christian
position after black-11
(download Zillions HanniBall bèta 0.3)
|
1. Hh4-f5/Hf4-g6/Hb4-d5
2. Hf5-f7/Hd4-c6
3. Hf7-e9/H*e9-c5
4. Lc2-d4/*c5-d4/L*d4-c4
5. Hc5-d7/*c4-e3/Ke1-e3
6. *e3-b6/Lc4-b6/*b6-c6
7. *c6-b6/Hd5-c7/*b6-c6
8. *c6-b6/Hc6-b8/L*b6-a6
9. L*a6-c5/Ef3-f4/L*c5-a6
10. La6-c5/Lc5-a6/Hd7-c5
11. Lg2-h4/L*a6-a4
* = ball
rules
|
Hh14-f10/Hb14-d13
Eh15-f13/Hd14-e12
Lg16-e11
Hf10-e8/He12-d10/Ed15-c14
Hf14-f10/Ec14-b13
Eb13-b10
Hd13-c10/Lc16-d14
Ef13-d11/Le11-d9
Hc10-e6/Ed11-c10
Ec10-c8/Ld14/c12
Lc12-f9/Ld9-d8
^ = ricochet
history
|
|
|
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #93 on: May 1st, 2009, 2:51am » |
|
Now Kec3 *a4-c3^d4 leaves the Horse on d4 open to capture, but the result is immediate loss.
Zillions (bèta) - Christian
position after black-12
(download Zillions HanniBall bèta 0.3)
|
1. Hh4-f5/Hf4-g6/Hb4-d5
2. Hf5-f7/Hd4-c6
3. Hf7-e9/H*e9-c5
4. Lc2-d4/*c5-d4/L*d4-c4
5. Hc5-d7/*c4-e3/Ke1-e3
6. *e3-b6/Lc4-b6/*b6-c6
7. *c6-b6/Hd5-c7/*b6-c6
8. *c6-b6/Hc6-b8/L*b6-a6
9. L*a6-c5/Ef3-f4/L*c5-a6
10. La6-c5/Lc5-a6/Hd7-c5
11. Lg2-h4/L*a6-a4
12. Ed3-e4/Hb8-c6/Hg6-f8
* = ball
rules
|
Hh14-f10/Hb14-d13
Eh15-f13/Hd14-e12
Lg16-e11
Hf10-e8/He12-d10/Ed15-c14
Hf14-f10/Ec14-b13
Eb13-b10
Hd13-c10/Lc16-d14
Ef13-d11/Le11-d9
Hc10-e6/Ed11-c10
Ec10-c8/Ld14/c12
Lc12-f9/Ld9-d8
He6-d4/Ld8-e6/Lf9-d8
^ = ricochet
history
|
|
| « Last Edit: May 1st, 2009, 3:40am by christianF » |
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #94 on: May 1st, 2009, 4:10am » |
|
White 13 is effectively a pass, but the 'change required' rule isn't yet implemented, so for now we'll have to accept the 'move'. But it is rather stupid to hold on to the ball because the Lion can be captured now.
Zillions (bèta) - Christian
position after black-13
(download Zillions HanniBall bèta 0.3)
|
1. Hh4-f5/Hf4-g6/Hb4-d5
2. Hf5-f7/Hd4-c6
3. Hf7-e9/H*e9-c5
4. Lc2-d4/*c5-d4/L*d4-c4
5. Hc5-d7/*c4-e3/Ke1-e3
6. *e3-b6/Lc4-b6/*b6-c6
7. *c6-b6/Hd5-c7/*b6-c6
8. *c6-b6/Hc6-b8/L*b6-a6
9. L*a6-c5/Ef3-f4/L*c5-a6
10. La6-c5/Lc5-a6/Hd7-c5
11. Lg2-h4/L*a6-a4
12. Ed3-e4/Hb8-c6/Hg6-f8
13. L*a4-b6/*b6-a4/Lb6-a4
(effectively a pass)
* = ball
rules
|
Hh14-f10/Hb14-d13
Eh15-f13/Hd14-e12
Lg16-e11
Hf10-e8/He12-d10/Ed15-c14
Hf14-f10/Ec14-b13
Eb13-b10
Hd13-c10/Lc16-d14
Ef13-d11/Le11-d9
Hc10-e6/Ed11-c10
Ec10-c8/Ld14/c12
Lc12-f9/Ld9-d8
He6-d4/Ld8-e6/Lf9-d8
Hc4xa4
^ = ricochet
history
|
|
| « Last Edit: May 1st, 2009, 4:12am by christianF » |
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #95 on: May 1st, 2009, 4:55am » |
|
The program behaved less than properly . As for the game, the Keeper should not be able to hold the ball indefinitely, because of its range - an attacker would be forced to cover the midfield while the ball isn't actually 'in play'. I'll think of a fix there.
Other than that I'm not entirely sure how to interpret this game. It ended in 14 moves, but against a generic bot of doubtful qualities, at least in a game with a tree density as high as HanniBall.
If someone argues that such a result means twiddlytwit, I'd have to agree.
To know how the game will behave eventually, many games will have to be played. My prediction stands, as well as my qualification: a tactical game with a strategy that is at the same time 'fairly obvious', in terms of pieces having to work together, as 'fairly obscure', because the action has a 'shifting' local focus point in the ball, without having a more permanent strategical framework such as for instance the pawn structure in Chess.
But then, I never quite know what comes out if a game 'autoshapes'. I am happy to have been given an unexpected and unsought for opportunity to illustrate how it happens. To the sceptics: it's not a provocation, but if the truth means offending someone's lack of imagination  then so be it.
Zillions (bèta) - Christian
position after black-14
(download Zillions HanniBall bèta 0.3)
|
1. Hh4-f5/Hf4-g6/Hb4-d5
2. Hf5-f7/Hd4-c6
3. Hf7-e9/H*e9-c5
4. Lc2-d4/*c5-d4/L*d4-c4
5. Hc5-d7/*c4-e3/Ke1-e3
6. *e3-b6/Lc4-b6/*b6-c6
7. *c6-b6/Hd5-c7/*b6-c6
8. *c6-b6/Hc6-b8/L*b6-a6
9. L*a6-c5/Ef3-f4/L*c5-a6
10. La6-c5/Lc5-a6/Hd7-c5
11. Lg2-h4/L*a6-a4
12. Ed3-e4/Hb8-c6/Hg6-f8
13. L*a4-b6/*b6-a4/Lb6-a4
(effectively a pass)
14. Hc5xa4/Ee4-d4/H*a4-c3
* = ball
rules
|
Hh14-f10/Hb14-d13
Eh15-f13/Hd14-e12
Lg16-e11
Hf10-e8/He12-d10/Ed15-c14
Hf14-f10/Ec14-b13
Eb13-b10
Hd13-c10/Lc16-d14
Ef13-d11/Le11-d9
Hc10-e6/Ed11-c10
Ec10-c8/Ld14/c12
Lc12-f9/Ld9-d8
He6-d4/Ld8-e6/Lf9-d8
Hc4xa4
Le6-d5xc3/*c3-d1/goal!
^ = ricochet
history
|
|
| « Last Edit: May 2nd, 2009, 1:28am by christianF » |
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #96 on: May 3rd, 2009, 5:20am » |
|
on May 1st, 2009, 4:55am, christianF wrote:| The Keeper should not be able to hold the ball indefinitely, because of its range - an attacker would be forced to cover the midfield while the ball isn't actually 'in play'. I'll think of a fix there. |
|
So I did, and the solution that presented itself was not to restrict the Keepers time with the ball explicitly.
On the contrary:
The Keeper combines the options of the 'Horse' and the queen in Chess, but may not leave the goal area (except for the goal itself). A Keeper in possesion of the ball, may not let go of it other than by shooting it (that is, he may not move and leave the ball behind). A Keeper shoots the ball up to six squares away, queenwise. Direction and distance are the player's choice, but the ball must land outside the goal area.
The darkblue parts are new.
Note that the rule feels in agreement with actual soccer, and the time restriction is implicit: a Keeper in possesion of the ball risks capture like any other piece, and 3 moves per turn opens ample opportunities to threaten it.
I've also made '3-moves per turn' the main variant and pushed the other two a bit further to the back.
christian freeling
PS. I'm playing the Zillions program again, this time with 10 minutes or more per move (half an hour or more per turn). The 'or more' is due to the fact that Ed must keep an eye on the timer himself: Zillions has no automatic timer above 3 minutes.
Nevertheless it needed only seven turns to lose its first piece.
|
| « Last Edit: May 3rd, 2009, 5:24am by christianF » |
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #97 on: May 3rd, 2009, 6:45am » |
|
This time Zillions has 10+ min. per move (30+ min. per turn)
Links in the notation table go to the position after the black move.
The new Keeper rule is implemented, swap and obstruction not yet.
@ black-30: this is tricky, Zillions is winning a piece ... mea culpa!
Disregarding the fact that I'm an absolute beginner, I think it's clear that Zillions tends to blunder in positions with many pieces, as was to be expected, but improves as the number of pieces is diminishing, to the point of tricking an inexperienced humanoid out of a piece  .
Meanwhile, in this game, none of the preperceived 'problems' materialized. Obstruction needed a formal definition and a formal procedure, but as an actual problem not even a shadow of it has emerged on the horizon.
That's not to say these problems might not emerge, although I don't feel they will. I trust the game to be what I said all along: an intelligent tactical game, and a nice and unexpected addition to my oeuvre, but not quite a 'mental sportsweapon'.
finished game ...
|
| « Last Edit: May 15th, 2009, 2:48am by christianF » |
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #98 on: May 5th, 2009, 4:37am » |
|
Since Zillions is obviously no match, this game may be considered to be the first serious HanniBall game.
I've rejected to swap white's first move.
Links in the notation table go to the position after the corresponding move.
@ Black 9: overlooked the Horse on c10, now I'll have to do without a Keeper  .
@ Black 11: my last move initially was E*f13-f12 which was stupid because the Horse on e9 can capture in two moves and thus win a piece. Adanac kindly allowed a redo (E*f13-g13).
It's the flipside of the medal. When inventing, my mind moves effortlessly, when playing it stumbles from one tactical oversight to the next, making it hard to do justice to any game, including my own.
The game, I'm glad to say, behaves as expected, and so, I'm sorry to add, does the inventor  .
@ Black 27: White aims at the ricochet La87/L*a7b5/*b5c3 ^...
The horse was moved to b5 to prevent that, for the moment.
Should it be captured, the Lion on d8 can recapture in two moves and use the third to get rid of the ball, possibly even using a ricochet off the white Keeper.
@ Black 34: In terms of material I could capture 34 ... Hg9-h11/*h11-g11/Eh12xg11. White can recapture, but he cannot risk the second possible capture of the Horse on g10, because his Lion would remain in possesion of the ball and be captured in the rebound.
However, he can instead move 35. Lh9xg11/L*g11-f13/*f13-e15. And he would certainly do that because it's the point of moving the Elephant to d15 in the first place. Since I must manage without my Keeper, the capture of the Elephant on g11 at this point would actually lose the game.
@ Black 36: Here White has only the chosen option, Lh3, *i5 or the same with *i2 or *i3.
On the chosen option 37... Hg9-i5/*h5 seems also possible, but I'm a bit worried about 38. Lh35, *i7 because I can't reach the ball in one move there.
@ Black 39: Since my breakthrough @36 Adanac has had to defend very carefully, and now he threatens to ricochet off to midfield.
Exchanging a Lion doesn't help, because the Lion on d5 recaptures in 2 and thus can still employ the same ricochet. So I'll have to 'bury the ball' and summon reinforcement. I'm curious how the game will handle the ensuing cornerfight ... .
@ Black 40: White 40 is a dirty trick  and a pattern that should be kept in mind for cornerfights. I can't capture the Elepant because I would lose a Lion. Meanwhile the threat is to pick up the ball with it, return to h2 and ricochet off the Keeper. So I'm forced to retreat, politely but decisively.
@ Black 43: White 43 threatens to capture the Lion on h3 with the Elephant on g3.
The exchange of an Elephant on h6 looks disadvantageous for black because it would set off the g5 Lion towards the wide open spaces in front of my keeperless goal.
@ Black 48: My idea in the last couple of moves was to sneak up on the solitaire Elephant on d15 and try to trap it, but events take a somewhat unfavorable turn  .
@ Black 49: Black resigns.
|
| « Last Edit: May 25th, 2009, 3:04am by christianF » |
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #99 on: May 25th, 2009, 5:03am » |
|
First things first: Adanac, congrats with your victory in the first ever serious Hanniball game. Thanks for the game and I hope you enjoyed it as much as I did .
Of course I'll never be able to prove any claims with regard to the game's behaviour at a hypothetical grandmaster level, because there will be no 'grandmaster level'. When a game 'autoshapes' I never quite know what will emerge until it's there, and HanniBall turned out a nice recreational game among a multitude of nice recreational games.
Strategy games have 'advantageous sub-goals' in their bag, such as for instance promotion in Chess of Draughts. Hanniball has only material gain as a sub-goal, so I'd classify it as a tactical game.
With regard to its behaviour, one aspect isn't quite clear to me. Fritzlein touched on that when he wrote:
on Apr 14th, 2009, 10:46am, Fritzlein wrote:| Given the obvious advantage of massed pieces that mutually protect each other, there needs to be a compensating advantage of spreading out pieces. In Go, for example, there is the strength/territory tradeoff where players are continually torn between playing thinly and playing thickly. Will well-played Choccer be a slow-moving game because the pieces stay clumped of necessity, and therefore you have two mobs slowly pushing each other forward and back? |
|
I think it's not quite that simple because pieces do not capture the usual way and thus can't protect one another in the usual way either. If a piece is in possession, any opponent's piece that can reach it in three moves can capture it, but only at the risk of countercapture. Leave a piece in possesion that can be reached in two moves or less, and it can be captured without risking countercapture (because the capturing piece can separate itself from the ball after capturing). In terms of capture and exchange this leads to a way of thinking that is quite new and various consequences have emerged that were not immediately obvious. Playing so far suggests that clustering is less of a problem than feared. True, the action is where the ball is, but the ball goes round rather quickly.
Adanac wrote:
on Apr 25th, 2009, 8:41am, Adanac wrote:There are 2 weaknesses, but I’m optimistic that both can be solved quite easily.
1. The lions are far too powerful relative to the other pieces. While an attack with only a few horses and/or elephants is very easy to stop, lions are quite dangerous. However, lions are also extremely powerful defenders covering a full radius of 5 squares plus about half the squares 6 steps away. One player might choose to keep both lions back leading to…
2. A defensive player can create an iron curtain by keeping 2 lions back, supported by a few other defenders. My game against Christian (note: the previous one) was very offence-focused because we both used lions on the attack. I’ve tested a few scenarios and it seems impossible to score against a talented and determined defender that keeps both lions back. Obviously, if perfect defensive play makes it impossible to score then this will be a fatal weakness in the design. |
|
I think this is still an open question, and no doubt Adanac is a good defender. But a crowded defense opens possibilities for exchange (in our game several of these were declined), and exchange reduces material. And this touches on an aspect of HanniBall that has been underexposed. It concerns 'decisiveness', as defined by Mark Thompson in his leading article Defining the Abstract, that I highly recommend. Here's a quote:
Quote:| But in addition to drama, a game must also have decisiveness: it should be possible ultimately for one player to achieve an advantage from which the other player cannot recover. Abalone has been criticized as lacking decisiveness: there appears to be a strategy which a weaker player can adopt (clumping his pieces together and never extending them, even to attack), which makes it impossible for the stronger player to win. |
|
That sounds not too far off: Hanniball might suffer from indecisiveness, almost the same way soccer does. Two strong defensive players, familiar with the subtleties of its tactics, might have the ball going round and round, without the scales tipping one way or the other.
Mark Thompson remarks this about Chess:
Quote:| Even Chess at the highest levels is becoming drawish; in matches between world championship contenders, dozens of games are played and most end in draws. Imagine how unsatisfying it would be if contestants for the world championship played fifty games and the victor won 3-2 with 45 draws; one could not help but wonder whether, if the match had been ten games longer, the other player might have been champion. |
|
He might have said the same about Draughts, or even Arimaa, if some comments are to be believed.
And these games have 'pawn structures' of a progressive nature, enabling the 'sub-goal' of promotion (although in Arimaa this promotion has been 'promoted' to the actual goal). HanniBall doesn't. Wouldn't that make the problem even worse?
I'm not sure about HanniBall's measure of 'decisiveness', but it differs in one important aspect: where in the above games, mutual reduction of material will eventually lead to more drawish positions, in Hanniball it's the reverse: the less material, the greater the chance of an attacker escaping the ranks of the defenders. Even with a mutually balanced reduction of material, the game becomes progressively more 'unbalanced' and more inclined to tip one way or the other.
Alas, we'll never know, because it would require a miracle to get enough high lever players for a game that is nice enough between a multitude of nice enough games. It's fun, but it's certainly not a 'mental sportsweapon'.
In terms of sticking my neck out by predicting the behaviour of this highly unusual organism, with no past references to build upon ... I think my head is still attached .
|
| « Last Edit: May 26th, 2009, 12:19am by christianF » |
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #100 on: May 25th, 2009, 6:16am » |
|
A more general remark about 'future behaviour' - Fritzlein wrote:
on Mar 26th, 2009, 10:24pm, Fritzlein wrote:It is rational behavior on the part of gamers not to put their faith in a game that hasn't withstood the test of time. Why waste effort on a new game when it is very likely to prove flawed in the long run? I also quite sympathize with masters of established games not wanting to walk away from the thousands of hours they have invested honing their skills at one game if their reward is to be a beginner at another game.
Admittedly it creates a nearly-closed circle of great games, because one needs gamers to prove that a game is great, and one also needs to prove that a game is great before it will attract gamers. |
|
I totally agree, but there's an underlying premise that the question of predicting future behaviour is devoid of rational arguments.
In judging Grand Chess, can't one rely to a fair degree on the behaviour of Chess?
There are differences, sure, otherwise it wouldn't be Grand Chess, but there are ample similarities too. Grand Chess has by now proven to suffer no obvious imbalance from its opening array. That being the case, isn't the behaviour of Chess a reliable indicator of Grand Chess' behaviour?
As for specific strategies, let's for the sake of argument assume different openings will eventually emerge in Grand Chess, and that they will have names. I cannot predict which main alleys will emerge, only that main alleys will emerge. And so, it would appear to me, can anyone.
Why Grand Chess would "very likely to prove flawed in the long run" is not clear to me, unless one holds to the idea that inventing games is a human activity excluded from progress.
And why would that be?
The same argument holds for Dameo.
The framework on which it is build is well known and well tested. It is capable of the same mindboggling combinations that can be found in International Draughts, but has a much smaller margin for draws, quite independent of the level of play. It is more flexible in terms of pace than any other draughts game.
This is what Benedikt Rosenau writes at BoardGameGeek's A call to arms to all ABSTACTS fans!:
Quote:| Dameo - from the ever mutating Checkers family, a very testbed of game design. Simple rules based on historical versions. It keeps all that was good in Checkers, adds to it, and drops what was bad. The aspect that men can jump their own for movement influences both strategy and tactics. The fascinating combinations of Checkers are still there, while the option for increased speed results in new plans. Interestingly, the people I showed it to liked it beginning with the first game. |
|
So is it really that difficult to predict how it will behave at top level? Isn't the problem, really, that players do not trust their own judgement?
|
| « Last Edit: May 25th, 2009, 6:25am by christianF » |
IP Logged |
|
|
|
Adanac
Forum Guru
Arimaa player #892
Gender:
Posts: 635
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #101 on: Jun 1st, 2009, 9:29pm » |
|
Thanks again for an interesting game, Christian. The Keeper for Horse trade created quite an unusual imbalance. On the one hand, I knew that I could score easily if I could advance the ball deep into your territory. Unfortunately, with one less piece on the board, I had great difficulty getting control of the ball.
Overall, I’d have to say that this second game has reinforced many of my beliefs from the first game:
- Lions are far, far more powerful than the other pieces and virtually every successful attack & defence in our 2 games have relied heavily upon Lions.
- I find Horses to be the least useful pieces. They’re good at long-range threats against enemy ball-carriers and, of course, they can be effective as part of a wall of pieces. But they seem to be near-useless as attackers because it’s often so awkward for them to pick up the ball, and having done so they might have to immediately kick the ball away at a range of only 1 square.
- It’s extremely difficult to score. Our last game went on for 49 moves and I only won by getting a Lion in deep against a Keeper-less net. Intuitively, I still believe that scoring against a determined stonewaller with defensively-positioned lions is near impossible.
I still believe that increasing the kicking distance of the Horse & Elephant would simultaneously solve many of the problems that I’ve noted in previous posts. Extra kicking power it would allow many more tactical options on each move, allow more interesting long-term positioning of pieces across a larger area of the board, reduce the clumping of pieces, eliminate/reduce those long sequences of play where pieces are afraid to get too close to the ball, reduce the enormous power imbalance between Lions and other pieces & greatly increase the number of scoring chances.
I’m of the opinion that HaniBall has a little bit of commercial potential if it’s targetted at youths/teenagers when they’re just getting into organized sports & strategy games. This game has a really fun & appealing theme and a lot of strategic and tactical options on every move. However, I find the game overly defensive at this point and some small rule tweaks are likely needed. I have no idea whether a simple change to the kicking rules would be the magic-bullet fix to all the little issues, but I would be very interested to see it tested.
Finally, Christian may have stumbled onto a great anti-computer game, without even trying to do so! HaniBall is both tactical & strategic, yet doesn’t easily allow material imbalances. That’s an unusual feature in a tactical game and it implies that humans may be able to dominate this game even if talented bot-programmers took up a “HaniBall Challenge”. I’d love to see it, but I’m not necessarily saying that I’m brave enough to risk $10,000 of my own money like Omar did for Arimaa! 
|
|
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #102 on: Jun 3rd, 2009, 6:44am » |
|
Thanks Adanac. I've gone over your comments and I agree largely with your diagnosis, less so with the suggested remedies.
I'd like to disregard the 'behavioural claims' in favor of the game itself, because there's room for improvement indeed.
on Jun 1st, 2009, 9:29pm, Adanac wrote:Overall, I’d have to say that this second game has reinforced many of my beliefs from the first game:
- Lions are far, far more powerful than the other pieces and virtually every successful attack & defence in our 2 games have relied heavily upon Lions.
- I find Horses to be the least useful pieces. They’re good at long-range threats against enemy ball-carriers and, of course, they can be effective as part of a wall of pieces. But they seem to be near-useless as attackers because it’s often so awkward for them to pick up the ball, and having done so they might have to immediately kick the ball away at a range of only 1 square.
- It’s extremely difficult to score. Our last game went on for 49 moves and I only won by getting a Lion in deep against a Keeper-less net. Intuitively, I still believe that scoring against a determined stonewaller with defensively-positioned lions is near impossible. |
|
Horses and Elephants are each others reverse in terms of movement and shooting, while Lions are their union. The resulting set and the 4-4-2 distribution has a certain completeness and elegance that I value greatly. In fact: this is the set I saw playing in my head and I still believe it basically behaves properly. Differences in character appear almost implicitly in such a set, so for me that's less of an argument. At its introduction in the sixteenth century, the Queen in Chess was also largely considered to be off the scale in terms of strenght, and even coined the 'mad queen'.
But I agree with your third point: a player who declines exchanges as much as possible, and concentrates on defense, might keep a game dragging on indefinitely. A lack of 'decisiveness' as Mark Thompson would call it.
on Jun 1st, 2009, 9:29pm, Adanac wrote:| I still believe that increasing the kicking distance of the Horse & Elephant would simultaneously solve many of the problems that I’ve noted in previous posts. Extra kicking power it would allow many more tactical options on each move, allow more interesting long-term positioning of pieces across a larger area of the board, reduce the clumping of pieces, eliminate/reduce those long sequences of play where pieces are afraid to get too close to the ball, reduce the enormous power imbalance between Lions and other pieces & greatly increase the number of scoring chances. |
|
You advertise it so convincingly that I'm almost inclined to follow. Almost, because I think the same improvements can be made without altering the nature of the pieces. It's not simple to maintain the 'duality' of Horse and Elephant, and their 'union' in the Lion, when you want to increase the shooting range, even apart from other side effects that might emerge.
So this is what I suggest:
1. Reduce the keeper to the Lion's move.
That is 'king or knight' instead of 'queen or knight'. Of course its restriction to the goal area remains, as do its shooting abilities.
I had initially given it the queen's move to compensate for its restriction to the goal area, but the Keeper is indeed too strong in the current context.
2. Reduce the board by two rows.
That obviously brings the target nearer. It also implies that a shot by or a ricochet off the Keeper both have a maximum range of 5 instead of 6.
3. Make it 4 moves per turn and forget about '3' and '1-6'.
This needs a somewhat more elaborate motivation.
First and foremost: it will be harder to avoid captures and harder to restrict them to exchanges. That's good, because the less pieces, the more Hanniball tends to tip one way or the other.
Secondly: light pieces one step away from the ball get the opportunity to capture solitary. With 3 moves per turn they can't: they can grab the ball (1), shoot it at an opponent (2) but next they cannot capture, because they move and shoot in different ways. So you always need a combination of light pieces to capture.
With 4 moves per turn however, such a single minor piece can capture on its own, albeit at the price of being in possession of the ball, and thus vulnarable, at the end of the turn. Nevertheless this is an increase in potential to be reckoned with.
Of course the Lion also profits: it can now capture solitary and even get rid of the ball. That's good in another way: it increases opportunities for mistakes that result in the loss of a piece and thus increases 'decisiveness'.
Taken together, these changes don't actually alter the core of the game all that much. Boardsize is arbitrary to begin with, and 4 moves per turn was already an option. The only real change is a slight reduction of the Keepers options, and one that even brings is more in line with the moves of other pieces.
on Jun 1st, 2009, 9:29pm, Adanac wrote:| I’m of the opinion that HanniBall has a little bit of commercial potential if it’s targeted at youths/teenagers when they’re just getting into organized sports & strategy games. |
|
I'm not so optimistic there, because I'd have to convince a manufacturer, and I'm not a merchant. I can't say (though any manufacturer must have heard that a hundred times before in the first place) that I invented a 'great strategy game' and that 'the end of Chess is nigh', because it's not true. But I'm expected to say at least that much, because saying that I've invented a nice tactical game when there's a multitude of nice tactical games out there, just won't cut it either.
on Jun 1st, 2009, 9:29pm, Adanac wrote:| Finally, Christian may have stumbled onto a great anti-computer game, without even trying to do so! HanniBall is both tactical & strategic, yet doesn’t easily allow material imbalances. That’s an unusual feature in a tactical game and it implies that humans may be able to dominate this game even if talented bot-programmers took up a “HanniBall Challenge”. I’d love to see it, but I’m not necessarily saying that I’m brave enough to risk $10,000 of my own money like Omar did for Arimaa! |
|
On this I can comment more generally. It's not too difficult to taylor a game to be manageable for humans and unmanageable for bots. Omar did so with Arimaa, which is not to say that it was an 'easy' game to invent (you should ask Omar for that), but only that this particular aspect is not too difficult. I guess Omar took his inspiration from the Jungle Game and introduced 'multiple moves per turn' to make sure the game tree had an exploding branch density that would choke a program. HanniBall's bot resistance is based on the same, and if you care to take a look at Medusa, you'll discover a mid-game branch density that easily surpasses them both. Medusa also was invented without that in mind (at the time programs were hardly able to play Chess or Checkers and Go was a faint dream). The point being that for the forseeable future there will be human superiority in games, because we can taylor games that way  .
There'll also be computer superiority in many other games. The qualities required for one or the other are very interesting. Havannah's 'unprogrammability' for instance, is not so much based on complexity and an exploding branch density.
I think this is a highly interesting field, and the more programs, the clearer the contours of the possible and impossible, and the more mysterious the pattern recognition abilities of the human mind.
P.S. I've already simplified the rules and 'a late arrival' accordingly, and of course I'd love to play another game for the gallery, against Adanac or anyone willing to give it a try.
P.P.S. On informing Ed of the above changes, he suggested to get around the swap rule (which is difficult to implement on Zillions) by limiting the number of moves for white on his first turn to 'two', instead of 'four'. Since this effectively leads to a very similar result, I think it's an excellent idea - consider it implemented.
P.P.P.S. The game has been updated @ Zillions - thanks Ed!
Download HanniBall for Zillions (bèta 0.6)
|
| « Last Edit: Jun 3rd, 2009, 10:39am by christianF » |
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #103 on: Jun 5th, 2009, 9:35am » |
|
Here's a HanniBall game 'new style' for the Gallery:
@ black-5: A mistake triggering a nice manoeuvre by the Horse on c8, first clearing the way for the white Lion, while preparing for the capture of the black Lion.
@ black-7: Here black already indicated the hopelesness of the position. It would seem that Mark Thompson's decisiveness criterion is sufficiently met under the '4-moves-per-turn' rules.
@ black-9: Hopeless or not, black 9 is a strong defense and I must retreat with one Lion. Both white Lions are offered for exchange after white-10 and the Elephant on d7 covers the center against breakout attempst by the black Lion.
@ black-10: A 'ricochet escape' that white overlooked. It suggests that there's no lack of Mark Thompson's drama criterion either.
@ black-11: I made a mistake in the display (Elephant should be on the h-file, ball on the i-file instead of vice versa - miraculously white's next move still fits as intended). It is corrected now.
|
| « Last Edit: Jun 9th, 2009, 10:18am by christianF » |
IP Logged |
|
|
|
christianF
Forum Moderator
Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #104 on: Jun 9th, 2009, 11:05am » |
|
Well guys, that's it. Ed will make an applet soon, so HanniBall will be accessible to everyone at mindsports. Later maybe at iG Game Center too, but that might take a while because Arty has several shortlists of different weight and for the moment HanniBall isn't on any of them.
I enjoyed the process of invention that unexpectedly started and the reactions of the posters in this thread, and I thank Omar and Adanac in particular, for starting the thread and contributing so enthusiastically to the game's development respectively.
christian
|
|
IP Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Notify of replies
Send Topic
Print
