Author Topic: Design Considerations for Very Large Games?  (Read 161 times)

joejoyce

  • Newbie
  • *
  • Posts: 38
  • Location: New York
    • View Profile
Re: Design Considerations for Very Large Games?
« Reply #15 on: April 25, 2018, 01:43:58 am »
Hi, HG, nice to be discussing things with you again! You've got a lot here, so I'm going to italicize your comment and intersperse with this font.

I don't like multiple moves per turn very much as a matter of principle. Turns are an artifact in the first place; in real life time elapses continuously, and adversaries do their thing simultaneously. Alternating single moves is the closest approximation you can get to that if you don't want the game to emphasize dexterity, rather than mental ability.

I appreciate what you are saying, and argue there is another way to get a close approximation to 'real life', by allowing all the pieces on one side to move a little bit, then allow the other side to move all its pieces a little bit, and continue alternating this way. I do it by using (very) short range pieces and restrictive 'leader/brain unit/activator' rules to keep those pieces close together, played on (very) large boards. I'd argue this is a better approximation of reality for what I've been trying to do, also. Now, I don't see why it wouldn't translate to games maintaining a chesslike character, but I have yet to prove that. Grandlem and Granderlem are my 1st and 2nd test vehicles. 

Games on large boards tend to 'factorize' in nearly independent sub-games anyway, and it makes very little difference if you would play these sub-games simultneously (doing a move in all of them in a single turn), or just play them out one after the other. The latter is more orderly. If I would make a Chess variant that is just 8 orthodox Chess games in parallel, where both players are playing as if it were a simul, would that really add anything? If the players were allowed only one move on a board of their choice, before the opponent could do the same, it seems much more interesting. Forcing players to think about many unrelated things at once is just annoying.

Okay, I see, understand, and pretty much agree with your analysis, as far as it goes. What I'm doing is deliberately 'over-factorizing' the game/gameboard. This addresses your objection, imo, by deliberately setting up 'tight quarters' on the board, where it can be advantageous for neighbors to help each other reciprocally. The restraints are set up so that one zone can pretty much only help the zones immediately adjacent to it. And it is engaged in its own zonal chess game, where it must protect its 'leader' piece, prevent enemy promotions or flanking maneuvers on neighboring zones...

Castling is a multi-move, but a 'coherent' multi-move, which serves a single plan. You could also say that from Pawn double-pushes, which are basically two moves with the same piece. Such moves speed up the game, without creating a distraction. So I think such moves are OK. This is why I proposed a 'transporter piece' in the 50x50 thread, which could carry a bunch of slow pieces all at once over a large distance, again something that has a well-defined coherent goal. But a big time saver in achieving such a goal.

For the purposes of advancing the game, I'm all in favor of group moves. Discussed the idea of transporters with Larry Lynn Smith for a while some years ago, and with others, but never quite liked how it would seem to work out in practice. Instead, I stumbled upon "activators", pieces which other pieces have to be near to move. They have proved so flexible and given me so many things that I never bothered to investigate large multi-move 'pure' chess games. Until now. And making a huge game playable in a reasonable time forces 'bulk moves', I think. The trick is how to organize them to preserve chessness. And I think that means to a great extent, the 'advance into contact' and 'attacking/defending' moves must be 'singular', that is, not supported by another piece moving in the same turn. I'm guessing 70% of the time would be enough to keep the game sufficiently chesslike while allowing some really dirty creative cross-zone attacks. I think it partakes a little of the feel of Crazyhouse rather than being a series of simultaneous non-interacting games.

I am not sure if large games necessalily have to be games of attrition. Look at Modern Tenjiku Shogi. Because the jumping generals can check there, and the King starts badly suffocated, breakthroughs there can occur from the very beginning, and in fact are very common. I guess the trick is basically that Tenjiku Shogi is a game of 'levels', higher ranked generals living in more sparsely populated levels where they fly over all lower-ranked pieces. This way the board is both sparsely and densely populated at the same time, and the breakthroughs can take place on the sparse level. When the game progresses without a breakthrough, the sparse levels become empty levels, and the next-most populated level becomes the sparse one, and breakthroughs are now possible there. I guess this idea could be implemented in a more balanced way as in Tenjiku, where there are only 4 levels, and the upper two levels are only populated by a single piece (per player).

In a way this reflects the dynamics in orthodox Chess, where pieces do not really fly over each other, but where sliders 'fly' in between the other pieces. Presence of Queens provides a good opportunity for an early mate. Once the Queens get traded, and the board population thins, Rooks start to dominate the show. When these cannot finish off the game, it is up to the minors to battle it out, and finally the Pawns.

I think lengthy games get very boring when there is no chance to finish the opponent off quickly after having acquired an obviously decisive advantage.


I've looked briefly at Tenjiku Shogi, but need to see/play a game to begin to appreciate just how it works. It's got a 16x16 board for 256 squares, a starting piece density of just over 60%, with 100 open squares at start & 78 pieces/side. It's roughly comparable to the introductory/training CaM scenario or Macysburg in size. Both those games finish in about 35 turns. How long does Tenjiku take? ;) Yes, that's a cheat. It's possible in Macysburg to make 1000 - 2000 individual piece moves during the course of the game, for each player. But there are only 5 kinds of pieces in the CaM games, making it easy to move numbers of pieces each turn, especially because they have simple chesslike moves: a 1-step king, a 2-step king, a knight, and finally 1, 2, or 3 square leaping bishops and rooks. Since capture is by replacement and every piece might move every turn, armies can melt away in 2 - 3 turns of intense combat. So decisive blows are possible. And there are 3 different levels of victory in Macysburg, with both players able to achieve victory points for achieving geographical and military objectives. So a long, close game can provide interest and enjoyment right up to the end. Or, halfway through the game, someone might get smashed trying to control Macysburg for a tactical victory, then run their army off the board for an operational defeat to avoid the destruction of their army on the board for a strategic defeat. And end the game in 90 minutes - 2 hours.

How long does it take, time and number of moves, to play Tenjiku? Now, suppose you split it into 2 8x16 zones, and allowed 1 move in each zone per turn? Appoint a specific piece leader of the non-king zone. King or leader must be in a zone for a piece starting the turn in that zone to move. What does that do to the game?

HGMuller

  • Global Moderator
  • Full Member
  • *****
  • Posts: 121
    • View Profile
Re: Design Considerations for Very Large Games?
« Reply #16 on: April 25, 2018, 04:00:26 am »
About Tenjiku Shogi:

As it happens I am currently participating in a Tenjiku-Shogi correspondence competition, on the gamerz.net PBeM server. (Time control: 120 days sudden death.) I have heard that the average length of all games played there is 6 moves. This of course is mainly because player strength wildly differ, and most people do not know any opening theory. I won one game there in 2 moves (3 half-moves), as the opponent (a novice) failed to deal with the mate threat I brought to bear with my first move (although this is the most-common opening move). This is a bit like fool's mate in Chess, but the difference is that for fool's mate the opponent has to do two really poor moves out of many, while in Tenjiku you just have to not do the only move that will prevent it.

All games can be seen on http://www.gamerz.net/pbmserv/List.php?TenjikuShogi ; For a more typical game between players with some experience, see game 2010 and 2015. (Unfortunately even the 'western' representation of the pieces is awful; a Rook is not even depicted as a Rook, but as some silly military insigne.) Those games were won by checkmate in 14 and 10 (full) moves, respectively. The opening phase of Modern Tenjiku Shogi is like a mine field; in many positions there are only 1 to 3 playable moves, and any other move loses the game quickly. Even experienced players make mistakes; in my game against 'kokosz' (last year's champion), game 2039, we both overlooked a mate-in-3, and then a mate-in-2 threat until kokosz finally spotted it, and I was mated on move 24. Between good players, which do not grossly blunder, games can be long. Last year kokosz took 115 moves to beat my AI (game 1903), and 63 moves to beat me (game 1885). This year's game between kokosz and my AI (game 2047) is already going on for 45 moves, and although the AI thinks it is leading, by about two Queens, the game is long from being decided. (The pieces that can cause breakthroughs, i.e. Fire Demons and jumping generals, have all been traded, so now the main breakthrough event will be promotion of one of the Water Buffaloes to Fire Demon.)

About multi-moving:

I see what you mean with the short-range pieces, an I am of course aware of your great invention 'Chieftain Chess'. But if the range of the pieces and activators really factorizes the game into inependent sub-games, it really makes very little difference whether you allow a player to move in all of the sub-games, or just in one at the time.

Perhaps this is just a matter of taste, but I don't consider factorization into sub-games a very attractive trait. Why have a large game if they are really just a number of indepenent small games? Furthermore, trying to force the factorization by board zoning fails in the presence of long-range sliders, and factorization through activator pieces (as in Chieftain Chess) seems to encourage contracting the entire army to be in range of all activator pieces at once, so that in practice it becomes an unrestricted multi-move game. (Perhaps the need to defend a promotion zone can counteract this.)

What I mainly worry about is how non-factorize multi-moving affects tactics; there seems to be no viable defense against many pieces ganging up on you. This woul make the game degenerate into a 'capture as much as you can' game, as on the next turn you would lose the material anyway. This doesn't really require much planning, and loses the subtleties normally involved in playing Chess. You have this in Progressive Chess, which really is a poor game IMO.

Perhaps multi-move games work well if the moves are non-captures, and making a capture ends your turn. (Just as delivering check in Marseillais Chess ends your turn.)
« Last Edit: April 25, 2018, 05:24:54 am by HGMuller »

John_Lewis

  • Newbie
  • *
  • Posts: 47
    • View Profile
Re: Design Considerations for Very Large Games?
« Reply #17 on: April 25, 2018, 10:11:59 am »
Yes, my variants are also obscure. I don't mind. ;D

joejoyce

  • Newbie
  • *
  • Posts: 38
  • Location: New York
    • View Profile
Re: Design Considerations for Very Large Games?
« Reply #18 on: April 25, 2018, 01:49:20 pm »
Yes, my variants are also obscure. I don't mind. ;D
I'm proud to say I'm obscurely famous, known by tens worldwide!
Like Like x 1 View List

joejoyce

  • Newbie
  • *
  • Posts: 38
  • Location: New York
    • View Profile
Re: Design Considerations for Very Large Games?
« Reply #19 on: April 25, 2018, 03:15:07 pm »
HG, thanks for the Tenjiku info  - yeow! That sounds just a little too upgunned a game for my taste - Jeremy Good and Carlos Cetina would love it. I will have to look at the games in the tourney. Grin, I don't think that game needs speeding up. Disregard my 2 moves/turn suggestion, or try it... how much faster could the game be over, after all?

As for how combat works in a chess variant where all the pieces move each turn, here's a link to my (very) unfinished website showing the end of each player turn for the first 'day and night' or 1/3 of Macysburg. The game runs for as long as a weekend getaway package, 3 days and 2 nights, 38 turns. You see the initial elements on board at start. A few turns go by, then reinforcements enter the game. After several more turns (12 turns total, 24 player turns) night falls, the armies finish up current combat, separate for the night, and finally leaders 'rally'(bring back on board) 1/3 of the day's casualties, where the slow animation ends. Then day 2 starts. http://anotherlevel.games/?page_id=193 You might have to wait a bit to see anything, and what you will see is just the opening. Each player gets 2 more groups of reinforcements on day 2 equal in size to the 2 groups on day 1, plus more rallied units during night 2. Roughly 100 pieces/side, on a 32x32, 1024 squares, and it's all used during the game, if only for entry at the blue squares. Yes, combat is bloody, but you get more troops during the game to make up for it, and this is the only wargame I know where your armies feel fragile.

Oh, yeah, Macysburg is an abstract riff on the American Civil War Battle of Gettysburg.

Fast rules:
capture by replacement
must be within 2 of adequate leader to move, except skirmishers
leaders rated by maximun number of units they may order to move each turn
skirmishers are self-activating and may freely move each turn

Terrain
trees, hills, buildings
must start turn next to aany terrain square to enter it, and stop upon entering that terrain square
leaving terrain is at the normal movement rate, unless entering another terrain square
Cannon may never enter tree squares
Cavalry may never enter hill squares

Leader-cannon "castling"
 an adjacent leader and cannon may change places as one action, though neither may otherwise move that turn, and the leader can activate other units too

Victory points:
 1 - sole possession of Macysburg
 2 - chasing enemy army entirely off board
 3 - "destroying" enemy by reducing enemy army to less than 20 units on day 3
Both players can get victory points. Victory points are cumulative. Subtract loser's from winner's for level of victory.

Level of Victory:
 1 - Tactical
 2 - Operational
 3 - Strategic
 4,5 - Decisive
 6 - Overwhelming ;) And that, along with the specific leader rules and how each piece moves, cover the basic rules of all CaM games. Specific scenarios have special rules, like where and when pieces enter the game or Macysburg's night rules, but the entire game is essentially described by the series name, Command and Maneuver.

The CaM games are wargames, but a lot of what I do there can translate to chesslike large variants, and then maybe the typical tournament game would last into double-digit turns. :)

joejoyce

  • Newbie
  • *
  • Posts: 38
  • Location: New York
    • View Profile
Re: Design Considerations for Very Large Games?
« Reply #20 on: April 27, 2018, 03:36:46 pm »
HG, your comments in italics, mine interspersed

...About multi-moving:

I see what you mean with the short-range pieces, an I am of course aware of your great invention 'Chieftain Chess'. But if the range of the pieces and activators really factorizes the game into inependent sub-games, it really makes very little difference whether you allow a player to move in all of the sub-games, or just in one at the time.

Perhaps this is just a matter of taste, but I don't consider factorization into sub-games a very attractive trait. Why have a large game if they are really just a number of indepenent small games? ...


The answer has to be that the larger 'short-range pieces game' is NOT just an unconnected series of battles that happen to occur more or less near each other. Instead, there is a slow flow of pieces around the board, maybe going in fits and starts, but nonetheless, pieces are transferred between neighboring sectors. A good example of this is the introductory CaM scenario, A Tale of Two Countries:Intro. (Interestingly, if you trace the world-lines of the pieces game after game, you see elements of mathematical chaos, with 'strange attractors' that 'eat' piece worldlines, and neighboring pieces often having similar worldlines, but with some pieces wandering far across the board. I've been curious for a while about the mathematics behind the CaM series.)

...Furthermore, trying to force the factorization by board zoning fails in the presence of long-range sliders, and factorization through activator pieces (as in Chieftain Chess) seems to encourage contracting the entire army to be in range of all activator pieces at once, so that in practice it becomes an unrestricted multi-move game. (Perhaps the need to defend a promotion zone can counteract this.)

In Grandlem and Granderlem, both zones and activators are used. The zone boundaries go along or between files and extend from end to end of the board. Only 1 activator in a zone can activate pieces, no matter how many activators are in that zone. An activator may only activate pieces that start the turn in that activator's zone. Yes, you can crowd all the pieces into a smaller and smaller portion of the board, but you lose moves each turn if you empty out zones.

There's also a limit to how often you can fire pieces into the center, because at some point, all your long range pieces will be in the middle, where all of them together only get 1 or 2 moves/turn, so you don't have an opportunity to send infinite sliders back to the other zones, because effectively you're using your move or two in the center to capture own pieces, as while your opponent is (probably) attacking, you have to remove a powerful piece from the area of combat and send it away. Yes, it can come back later, but helping your opponent remove defenders from the area of your king on your turns is not a winning strategy.

I will say in GLS and GdrLS that the pawns cannot really change zones, and this could be a problem... first thought to 'correct' this is to make the pawns forward sideways wazirs which maintain the pawn's diagonal capture mode. I don't know if this mkes much if any difference to the game play, but in the Granlem 3-mover playtest, the pawns on the wings got destroyed (15 of 16 captured & 1 blocked) but the redesign is meant to fix that.

What I mainly worry about is how non-factorize multi-moving affects tactics; there seems to be no viable defense against many pieces ganging up on you. This woul make the game degenerate into a 'capture as much as you can' game, as on the next turn you would lose the material anyway. This doesn't really require much planning, and loses the subtleties normally involved in playing Chess. You have this in Progressive Chess, which really is a poor game IMO.

Perhaps multi-move games work well if the moves are non-captures, and making a capture ends your turn. (Just as delivering check in Marseillais Chess ends your turn.)

I hope I've at least partly answered some of your comments, and will get you to rethink some of your objections. I totally agree multi-movers need to be very carefully designed so they don't go chaotic, which leads to your 'undefendable mass attack' scenario among other things. Controlling chaos is the reason for rules. As things get more complex, you need tighter rules. It's an interesting balance; too tight, and you're right, each area of the board might as well be its own separate game.; too loose, and you've got pieces flying all over the board, preventing players from being able to forecast future game states with any real accuracy or success.

Finally, thanks for the comment about my "great invention, Chieftain Chess". I'm taking it as a compliment! ;) It was a gift from the Muse if there every was one! That game is a sport, a saltation, a mutant, a 'hopeful monster'. I've gotten a lot out of it. Too bad nobody else has! :D I suspect the main reason is that I pushed the boundaries too far with it. Activator chess (or more properly, activator shatranj) is a different game than chess, and it allows chess to do something 'easily' that chess has found very difficult, play well-organized multi-move games, including giant massively multi-move games that are simple simulations of things from war to microscopic life in a puddle. I like to explore new areas. But these areas aren't really chess as we think of it, but chess plus something else.

I think I can keep the essential chessness of structured multi-move variants intact while speeding them up without letting things get out of hand. I do need a playtest or so of the 4-mover variant, if anyone is interested.