Welcome, Guest. Please Login or Register.
Mar 29th, 2024, 7:57am

Home Home Help Help Search Search Members Members Login Login Register Register
Arimaa Forum « Christian Freeling on inventing games (part 2) »


   Arimaa Forum
   Arimaa
   Off Topic Discussion
(Moderators: christianF, supersamu)
   Christian Freeling on inventing games (part 2)
« Previous topic | Next topic »
Pages: 1 2 3  ...  11 Reply Reply Notify of replies Notify of replies Send Topic Send Topic Print Print
   Author  Topic: Christian Freeling on inventing games (part 2)  (Read 226591 times)
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Christian Freeling on inventing games (part 2)
« on: Mar 3rd, 2013, 6:27am »
Quote Quote Modify Modify

It's a good moment to switch to a new thread because the other one is quite long enough, and I've got some new games for starters.
Please note Omar's request to fellow inventors:
on Mar 2nd, 2013, 9:21am, omar wrote:
You are most welcome to post it here. However, I would suggest starting a new thread for any new game announcements rather than appending it to this thread.

The periods in which I invent appear to be scattered, but if they happen it is usually in the approach to winter. The recent wave started late november and was triggered by Luis Bolaños Mures' extraordinary game Ayu. "Dynamic connection" plays a role in Symple, but as a means to an end, not as the object itself. Ayu takes it for its very object, and does so in such a simple and generic way, with each rule necessary and sufficient, that I had little trouble recognizing it as a truly significant game. It made me realize that I had grossly neglected this theme, which in my mind more or less had boiled down to Lines of Action. A very nice game, no question about that, but clearly bot fodder. Symple had just led me to taking the theme more seriously when Ayu arrived and confirmed what I was already beginning to suspect: I had clearly missed something.
Other than LOA, Ayu is highly organic. You can literally imagine the stones having a mind of their own and organizing themselves into unity in the most effective way against equal yet alien opposition. I'm fairly sure I'm not the only one having this sensation when looking at a game of Ayu.
 
A new opening protocol
And then it so happened that I couldn't stop thinking about it. Such periods are characterized by a certain discrepancy between what happens on the outside and what happens on the inside. Virtually I'm sitting in the center of my head, looking at what's going on in IMAX. Physically I'm in zombie-mode, doing everything (except taking care of the animals) on automatic.  
 
I knew Luis had captured the essence of the theme, but it was a big prey and I was out for a chunk - it's a war out there, in case you hadn't noticed Grin . Of course I was after another implementation of the theme, not after the mechanics Ayu uses. So I started to focus on leaving out a fixed initial position, in favor of an opening array that would start out on an empty board and result naturally from one or the other protocol. Something that would spead out evenly, and cover about half the board, as in Ayu. Strangely enough I almost immediately found one, and the process brought me so close to a territory game that at least would work (that much I could see) that I decided to use it to store the protocol. That was Triccs and though nothing is actually wrong with it, it lacks drama. No reason to dismiss it, but not much reason to play it either. Shortly afterwards I noticed that I had the order of moves in the new protocol the wrong way around, but that was an easy fix.  
 
One sticking, one free
I coined the protocol "one sticking, one free" and it turned out to be a combinatorial game in itself. Here's how it goes. White starts by placing one stone on the empty board. From that point on players take turns to:
  • Place a stone on a cell adjacent to the last stone placed by the opponent, and ...
  • ... place a stone on a on a cell that has only vacant cells as neighbors.
When the player to move can no longer make the second placement, then his turn ends, as does the opening phase, and his opponent may start the second phase.
 
On square boards we're talking 'orthogonally adjacent', as usual. If you think about the implications, you'll find that when the protocol has reached its end, then
  • the number of black and white stones will be equal, and ...
  • ... the total number of placed stones is set between a (as yet to be proven) minimum and a (as yet to be proven) maximum, and ...
  • ... there's no telling who's turn it may be.
Regarding the second, I'm not a puzzler, but it can be established for any grid in any form or size. Regarding the third: trying to be the first (or second) to move in the next phase is a combinatorial game in itself, serving as a balancing mechanism for the second phase. Of course you'd have to know what is best, and that again would depend on the game. And in case you should know, then getting it on the board in an actual game isn't all that easy either. It's kind of a new light shining on the turn-order balance issue.
 
The hunt for games fitting the protocol
But above all: it's a generic protocol, free to use for anyone who cares, and for that moment for me in particular. So I had the general concepts of LOA and Ayu revolving around the kind of initial positions that would result from the protocol. It's often a somewhat dreary process till a clue presents itself. The first clue arrived in the LOA realm. In LOA a piece may move (in principle) the number of squares that are on the row or diagonal it moves on, so one move changes a lot of things:
  • It reduces the number of stones on every straight row, column and diagonal it leaves by one, accordingly reducing the range of all pieces on those line, and ...
  • ... it either increases the number of stones on every new straight row, column and diagonal it ends its move on by one, accordingly increasing the range of all pieces on those line, or ...
  • ... makes a capture, reducing the number of pieces of the opponent by one.
The most notable aspect of this protocol is that it is a totally arbitrary criterion to decide a piece's range, or at least in my view far more so than for instance taking its height, like Sid Sackson did with Focus. The second most notable aspect is that it works very well. And I'm a sucker for things that are totally arbitrary and work well.
 
Argon
Looking at a diagram like the one here, showing a possible position at the end of the opening protocol on a square board, a totally arbitrary criterion for determining a piece's range presented itself: make it equal to the number of vacant points adjacent to the departure point. What does a single move do?
  • It increases the number vacancies of every neighbor of the departure square, increasing their respective ranges by one, and ...
  • ... it reduces the number vacancies of every neighbor of the target square, reducing their respective ranges by one, and ...
well, I hadn't considered capture yet. But this is very reminiscent of the of mechanism of LOA, yet different enough to be worthy of its own game. Besides, capture wasn't all that difficult. In LOA a difference in material may or may not be an advantage in any given position. The opening protocol I put at the basis of the game emphatically ends with equal material, so I thought "let's keep it that way". You have to unite your pieces, and both will always have the same number, though the total may dwindle - it had the appeal of elegance.
 
Thus it came together quite easily and I named it 'Argon' because I happened to see the word at the bottom of a web page. It is easy to see that we have a game of visual dexterity and tactical opportunism here. It's also easy to see that it is bot fodder, I have no illusions there. If there's a less than opaque strategy possible, it will emerge towards the endgame, because the move-protocol requires precise steps, and options to alter a piece's range by interaction with other pieces get more limited with less pieces to work with.

So here was the result of approaching via the LOA angle. Very nice and all, albeit not exactly what I had in mind, more of a side effect. But the Ayu angle seemed annoyingly covered by the rigid logic of that game. Groups crawling towards one another under the umbrella of one generic rule governing movement, try to beat that. But the clue came, and as so often, it came by a shift of perspective. It occured to me that the number of groups in Ayu cannot increase, not so much by a rule to that effect, but as the result of a move mechanism that prevents groups from splitting up. But if the rules simply forbid the number of groups to increase, then there's nothing that prevents pieces to travel between groups. And that's a whole new ballgame.
 
Inertia
Here, again, is a diagram of a position at the end of the opening phase. It's hex for a reason: the protocol on a square board allows fairly 'dense' packing if players cooperate to that end. It seemed to interfere somewhat with a smooth flow of proceedings, using the mechanics I envisioned. Argon is indifferent to fluctuations in density, but this game was not. Argon on the other hand has a simple rule for a piece reaching the board's edge. The same simplicity isn't provided by the hexgrid. So both games seem to have a 'natural' grid.
 
Regarding the move to be used to hop between groups, the hexagonal rookmove came naturally. However that seemed a bit ... superficial. So I decided on a special rookmove, one that would start with the option to pass over any number of own pieces, including zero. So it would include the normal rookmove, but not be limited to it, and it would give individual groups a limited move option without the need to split up. All in all it was the perfect move for the generic rule that doesn't allow the increase of the number of one's groups.
 
The final touch was provided by the very definition of 'inertness', linking it to the existence and only to the existence of an open path between groups. The length of the path is irrelevant. Here again is an arbitrary criterion that works well Smiley .
 
What made the game especially interesting is that it allows forced cycles as well as situations where the player to move has groups that are not inert, but cannot move without violating the increase rule. I expect both situations to be extremely rare, but 3-fold is a draw and for the second situation the applet does provide a pass option that is dimmed by default and only is activated if the position is such as described.

Inertia can be played base-5 to base-7 at mindsports and I gladly invite anyone interested to give it a try. For the time being it has found shelter in the Pit, but it is clearly a game that belongs in the ArenA.
« Last Edit: Mar 3rd, 2013, 2:51pm by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #1 on: Mar 4th, 2013, 5:11am »
Quote Quote Modify Modify

In between ...
there were a few other alleys of thought. One of them involved Nick Bentley's Produto, Luis Bolaños Mures' Alpha and Néstor Romeral Andrés' Omega. Nick, by the way, pressed by marketing considerations, has renamed his game Ketchup to "Catchup". But that aside.
 
Multiplicity
Allow me to be frank, but I consider Multiplicity simpler, more logical and more generic than the games that led me to this particular vein of thought. If there was a clue to spark it, it was an idea I've put forward more than once: applets widen the possibilities of abstract game design. I'm not even going into the differences with the others, nor into their mutual differeces for that matter.
 
Multiplicity has a simple goal: get the highest score. The score, at any given moment, is the product of the sizes of a player's groups. A single stone is a group. Players can read their scores in the applet throughout the game. Placement is compulsory, so in the end the board is full and the applet displays the endscore.
 
It couldn't get much simpler than that, but of course there's a problem with turn-order balance and a pie won't solve it. The only balancing mechanism that has any grip on it is the one embedded in the Symple move-protocol. And it's embedded indeed, so you implicitly have to use the move protocol. It does more than just solving the balancing issue: it creates more diversity in the group sizes. It makes draws, not likely to begin with, even more unlikely.
 
It's cold out here ...
Since having many smaller groups beats having a few large ones, compulsory placement makes for a very cold endgame, in which players more than anything try to prevent having to connect their groups. It will be very dramatic, I can assure you. I can't show you yet, but an applet is in the pipeline.
 
So, for that matter, is a square version.

 
Another alley of thought was actually a shortcut. Before I get to that, here's the diagonal cross-cut problem:
 
On the left, is this a connection or not?
On the right, depending on the answer, here both have a connection or neither.
The first answer gives the impression of a race, the second of a block.
 
Many inventors encountered the problem, in particular in square connection games. In Slither, after any move, any diagonal pair of like colored stones must have a mutual like colored neighbor, otherwise a move is illegal. What the inventor maybe didn't realize is that the condition he gives is in fact a special case of a more generic condition, that not only covers all cases, but allows the diagonal connection if it cannot lead to any of the problems concerned.

Here is the generic restriction rule (a 'group' is orthogonally connected):
  • At the end of a player's turn any two diagonally adjacent like-colored stones must be part of the same group.
Note that the restriction in Slither is a special case of this. Under the generic rule, diagonally adjacent like colored stones are possible, if and only if the stones are part of the same group. Why? I think here the picture indeed is better than words.
 
To give credit where it is due, the generalization was suggested by Luis in our correspondence about Scware. Luis has put more thought into square connection games than I care to reflect on (but the possibility of using the Symple protocol had eluded him). Smiley

Scware
Scware was invented on autopilot. Benedikt Rosenau, who has been deplorably obstructed by work lately, without any time to spend on his creation, had used the Symple move protocol in Hex. Benedikt is very good at Hex. I suck at Hex. Benedikt says Symple-Hex is a very good game. I can relate to that because ...
  • ... the protocol is generic and I can see the kind of decisions that a player might face if it is applied to Hex, and ...
  • ... he's much better at Hex than I am.
So I thought, "well if it can be done with Hex, then it can be done with Square", only what is "Square"? Hex is virtually one very obvious game, but square connection games are manifold. So I thought, "well, let's take Slither and see what happens".
 
I chose a larger board because of the intended multi-move protocol, and added the Slither restriction and called it "Scware", which I still like very much. Then Luis pointed to a possible impasse and suggested the generic restriction rule, that he probably wouldn't have found without Scware, and so deepened the insight in the nature of the diagonal cross-cut problem and simplified the solution.

So this was a summary of this season's batch. Now the brain is pleasantly empty, the sun is shining and spring is in the air, at least here in the Netherlands. Applets are in the making and there's some time for that because I don't expect coming down with creativity again, any time soon. Smiley
« Last Edit: Mar 4th, 2013, 10:05am by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #2 on: Mar 5th, 2013, 2:36am »
Quote Quote Modify Modify

Copied from part 1:
Quote:
So I was at the presentation of CodeCup winners at the University of Twente to deliver a small speech. The point of it was that when game programming began, we were looking for games that were easy for computers, whereas now we're looking for games that are easy for humans.
 
Take a game like Explocus. In the late eighties a student at that same university wrote a program that ate itself through three ply in twenty minutes, using alpha-bèta pruning and a very simple evaluation function. It was unbeatable because no human player could manage more than a partial two-ply evaluation. Imagine a CodeCup contest now. The worst program would beat the best human ten out of ten. Whatever would happen in games between programs, would be beyond any human comprehension. It would prove that programs are better than humans at Explocus - and little more. It would be utterly uninteresting. That's why we should be looking for games that are easy for humans.
 
Humans need games that allow strategic understanding, the making of plans and the achieving of calculable sub-goals. I still beat the best Havannah bot seven out of ten. My score against the winner of the Symple contest is 24 wins out of 35. These games may not be more difficult to program, but they give humans a firm handle and, for the time being, a fair chance. That's more interesting for humans, and certainly not less interesting for programmers.
 
I've made a suggestion, therefore, to include this aspect in the procedure to select games, and to hold a yearly match between the winning program of the previous year, and the best human player (to be determined in some form of competition, possibly at mindsports, in the year following the win of the program).
 
The game for 2014 has been selected, but remains a secret for now, so I've got no clue. For 2015 I've suggested Luis' Ayu, because contrary to some people's perception I'm not after a one-man-show and Ayu is easily the best game I've seen in the past year. It also fits the requirements of a game wherein humans have a chance against digital opposition.  
 
I hope the idea of a yearly follow up match between man and machine finds some fertile ground. In that case, there may be an additional Symple match next to the new CodeCup Challenge, in about a year.

 
on Mar 2nd, 2013, 9:13am, omar wrote:
Thanks for sharing this Christian. If there's a video of the lecture, I would love to watch it; but maybe it's not in English. It's always interesting for me to see your perspectives on game design, since you've been at this for many years.  
 
I like your idea of having a yearly man vs machine match for the AI resistant games. I hope it fuels more interest in this area of AI. I really think there is a lot to be discovered in this area. But it's also a bit scary to think that eventually we may get to a point where there is no overlap between the games that are interesting to humans and games that are difficult for computers.

Hi Omar, I had missed your reply. The lecture was in Dutch indeed, with an audience of some 50 new students. Of course one cannot be all-embracing in 15 minutes, so I focused on the technogogy driven changes that are taking place in the world of abstact games. As for your fears, I fear they are fully justified. Humans will lose the battle, eventually, but that doesn't mean we shouldn't engage (do we have any choice, really?) or that the battle itself cannot be interesting. The fun is not in the outcome but in the journey. Smiley
 
On second thought ...
Maybe I'm jumping to conclusions here. It may be a bit more compiclicated. We're at the point now where some less than trivial games like Checkers, Awari and Fanorona have been solved (I'm not distinguishing, in this context, between strongly and weakly solved). Humans may still enjoy them, but they're playing in the shadow of perfect play. The question would be if this fate threathens all games, and the answer, for the foreseeable future, is still "no".
 
First of all, will there be any incentive to kill more games? We know it can be done with certain games, and we know that we're not anywhere near where others are concerned. Schaeffer et al did something that wasn't done before, at least not with a game of Checkers' standing. But follow-up kills will increasingly be duly noted. It will be like stretching the realm of the possible in the full realization that it is embedded in the impossible. If there were any profit in it, things might be different, but what would that be? Recognition for something that has been done before? Better programming techniques? I have some doubts there.
 
I don't think you can rule out the human factor. If a 16x16 Othello Computer Challenge were organized, then there would be a winner. Who cares? The programmers maybe, but the field to me seems hardly the most likely area for great innovations in programming techniques.
 
The interesting areas are those where humans can compete with bots on the highest level. The problem with new games is precisely that: they lack the broad base that is needed for a high peak. My suggestion for a follow-up match between last year's winner of the CodeCup and the best human player tries to fill that gap. Humans will not within a year be able to reach the 'highest level' in the kind of games that qualify, but neither will bots, I presume.
 
Another interesting area is the one of generic bots like Zillions. Not so long ago it was deemed impossible to have 5.000 songs in a matchbox, with all sorts background information provided. I think we may soon be able to have 500 abstract games at our disposal, allowing one to play live or turnbased with anyone, anywhere, anytime, complete with rules and examples, and a generic bot that has a 'reasonable' level of play in all. And a lot of them may belong to a new category of abstracts that are dependent on features provided by the app technologies.
 
Programs
In the realm of computation one may well imagine there's still huge progress to be made in terms of speed. But what are we calculating? Deterministic evaluation functions for new games will never be perfect (nor are they for many old games). Does calculating faster at a fuzzy evaluation function help? Monte Carlo evaluation basically cares for the result of a statistical outcome, so it doesn't have this particular drawback of requiring 'knowledge' (though it usually helps). MC would appear to profit more from increased calculation speed. But so far as I can see, not all types of games are equally suited for the method. Infinite games in particular pose some inherent problems in random play-outs. Moreover, pressing the limits costs money. We're not talking simple programs to suit humans in intelligent recreation. In any case, players of abstract games will have a few interesting decades ahead of them. Smiley
« Last Edit: Mar 5th, 2013, 10:35am by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #3 on: Mar 7th, 2013, 1:46pm »
Quote Quote Modify Modify

I've played a couple of games now, and I kept losing against Jos so I switched to a smaller board to get some more grip. I understand Jos' strategy, albeit usually too late. Positions do not exactly lack details to consider.
 
Christian Freeling - Jos Dekker (0-1)
Here for instance it's black to move and he cannot just connect at the bottom in two steps, because then red is faster. So he must defuse red's impending inertia by opening at the top. Lucky for Jos he could do so without splitting the group. Now red has no chance.
 
A few moves earlier I tried to escape with a forced cycle, but I didn't come anywhere near. The jury is still out on the frequency of their appearance, but I'm fairly confident they will not, and not even eventually, turn out to be problematic.

« Last Edit: Mar 7th, 2013, 1:48pm by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #4 on: Mar 12th, 2013, 5:41am »
Quote Quote Modify Modify

I've finally won my first game against Jos, which gives a confidence boost of sorts. Smiley
 
Jos Dekker - Christian Freeling (0-1)
In the final position, the threat is of course 23...E8B5.  
 
On 23.D47 Black moves E8I8 and Red cannot block at H7 (because of the increase rule).
 
On 23.D87 Black moves ED8 and Red cannot block with C37 (increase) while GC7 is met by 23...EH8.  
 
Note:
The resolution in our ability to read tactics has increased to the point that the emergence of forced cycles should become more apparent. And yes, we've seem the occasional shadow, but no actual emergence so far. I'm still not saying that Inertia is out of the woods, merely "so far so good".

« Last Edit: Mar 12th, 2013, 6:09am by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #5 on: Mar 12th, 2013, 3:01pm »
Quote Quote Modify Modify

As I write, I'm playing a game of Inertia against Luis in which he presents a position in the preliminary assumption it might be a draw. You can see it on the left.
 
Not only do I owe Luis because Ayu was so inspirational, but I also appreciate and actually need the scrutiny with which he goes into a game's workings.
 
However ...

 
... it's not a draw. Here is the final position after a remarkable black win, in which the pass rule actually takes effect.
 
If, on move 23, Black would move E5-D4?, then after Red's pass he would have to move 24.D4-E5, losing the initiative to Red, and reversing the whole circus, actually allowing a similar Red win*. Similar to the one here because Black of course moves 23...D74 and wins as shown.
 
The rules cover 3-fold and pass adequately, and a position in which both player's cannot move (despite the absence of inertia) has yet to be found. If anyone can it will be Luis Smiley
 
* Edit:
Ah, the role of stupidity in game inventing, tell me about it. Need I say who pointed out the red remark is wrong? Of course Red must pass after 23.E5-D4 but Black can simply proceed with 23...DE7 and still win in a similar matter as the one shown under the link at the top.

So it's a simple black win, where black has even some room to manoeuver. So much the better I'd say.  
« Last Edit: Mar 14th, 2013, 1:08pm by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #6 on: Mar 14th, 2013, 1:08pm »
Quote Quote Modify Modify

Ludotopia
It's hard nowadays to find a name that isn't already taken, so I wasn't really surprised to find that Ludotopia is actually a gamesite "... for all gamers who don't care about winning or loosing as long as we play all kinds of games: boardgames, videogames, cardgames, miniature games, etc...". Close enough, so I hope they don't mind that I use it in the context of abstract strategy games for:
  • "Ludotopia" - a parallel reality where game-X has been subject to the scrutiny of thousands of dedicated players over several decades, engaging in tournaments and competitions and amassing a load of theory and literature.
One can only speculate about Ludotopia and the opportunity is seldom left unused. But how much is there to go on?
 
on Nov 22nd, 2012, 9:33am, MarvinSpellbinder wrote:
Remember in Jurassic Park when Jeff Goldblum says "Life will find a way"? Same thing with draws. Draws will find a way. No one has to "envision" them. In Luis' case, he was far from "pleased" when he found out Ayu is draw susceptible.

Here a number of issues come together. Are games "alive"? Before you answer the question, look at a game of Ayu in a one-minute play-out. It looks like two self-assembling organisms, each with its own intent and plan. It's not Ayu specific of course. Go, Oust, Symple, Catchup (Nick renamed it), Slither, Draughts, Emergo, Othello and countless others would give a similar impression.
 
On the other hand, games are mathematics. If a game is solved (the distinction between 'strongly' and 'weakly' is not relevant here), then the truth of every possible position is known: win, loss or draw. I've more than once linked to MiniMancala because it shows a complete tree, including the truth of every possible position (onmouseover). So you can simply count the number of positions that are a draw. This is a very small game, but the argument extends to any game that can end in a draw, whether or not we have an 'overview' of the complete tree:
  • Mathematically the 'margin of draws' of a game is the number of drawn positions in its tree, set against the total number of positions.
In other words, its a percentage. So the above "life will find a way" metaphor should be at most "players will find a way", and the percentage of drawn positions obviously plays an objective role in that. Draws are not something that 'grows' in a tree.
Another thing is whether the margin of draws is known or not. In the latter case we must rely on speculation and estimates based on the best available arguments.  
 
on Nov 22nd, 2012, 4:35pm, NickBentley wrote:
I don't envision them, but I figure if we're really designing games to last for the long haul, we're designing them to be played at a level of which we're mostly ignorant (or at least I am; I don't want to be presumptive). Anyway that's what I try to aim for.  
 
Once players start to see how draws are possible, players can increasingly "play for the draw" to avoid losses, and this can compound the draw problem way beyond what one might expect it to be. See Chess. So, it may be that a single drawn Havannah game creates new ways of looking and new opportunities for players who want to play competitively, and a few years down the road, draws are way more common than we ever expected. For this reason, it's important to me to find out whether forced draws are possible.  
 
Don't get me wrong though: I really like Ayu and Havannah. It's just that my admiration is conditional, as it is for my own games that carry the same risk.

In the first paragraph Nick refers to Ludotopia and I can only agree with his purpose. But the second starts with a statement that I dare to question, because it doesn't address the objective aspect of a game's margin of draws: the fact that it's a percentage. We may not know this percentage for any particular game, but its size is a factor in whether or not the doomscenario Nick paints will actually take place. So his argument boils down to "better safe than sorry" and as such I think its valid, but also restrictive. Let's look at a number of games that either may end in a draw or are as yet labeled 'inconclusive'.
  • Reversi or Othello if you like. It's easy to see it can end in a draw, but it is not hard to estimate the margin. And there's been a lot of time to do that, without a lot of change in the initial estimate: "Games in which both players have the same number of disks their color at the end (almost always with a full-board 32-32 score) are not very common, but also not rare, and these are designated as 'ties' and scored as half of a win for each player in tournaments."
    wiki - reversi
    So in Reversi a tie isn't that big a deal. You certainly can't 'play for it'. How would you?  
     
  • Havannah has seen one recorded draw in thirty years of play. The conditions one has to fulfill in constructing a drawn position, like dividing corners and sides and knitting it al together with no two corners connected by one color, and no three sides, avoiding rings in the process, are so limiting, that the total number of drawn positions on a base-8 board would seem calculable, as opposed, at the moment, to the total number of possible positions (though estimates are calculable). Here's where the objective side kicks in: the margin of draws is, by any serious estimate, extremely low. And 'visualizing' a draw is far more difficult than visualizing the actual object (if indeed at all possible). So the suggestion that an increased level of play would automatically lead to an increase in the number of draws is in my opinion at least questionable.
     
  • Ayu has no history at all, so how would it fare in Ludotopia? As it happens both the game and its inventor support my case. Luis does because he seriously hunts for cycles in the game. Where he failed initially, others were successful in finding cooperative cycles. Here's the thing about cooperative cycles: they don't matter. So Luis' announcement of "bad news" was in my opinion a peculiar way of looking at it. If it had been a precursor of possible forced cycles, now that would have been another matter, though even then a game might not necessarily die from it. But there's been no indication of that. The two or three cooperative cycles that have been found are not even game positions because they have unequal and incomplete forces. Ayu's objective margin of draws, that is, the percentage of drawn positions in the gametree, is not known, but anyone can see that it must be extemely low. Sure there may be loops in the tree, caused by cycles, but they will never materialize unless in a mutual catatonic cycle between dedicated members of the Church of Cyclophobia, determined to demonstrate what everybody can see in the first place: a cooperative cycle and a refusal to win by at least one of the players. Now imagine that: players that abhor draws playing to draw.
     
  • Inertia has no history either, or it would be that it was inspired by Ayu. Does it support my case? I'm confident, but that's private, don't tell anyone. There's no question that Inertia has forced cycles. Also, it is not that easy to estimate the objective percentage of draws in the gametree. So the jury is out on this one. For now forced cycles in Inertia are a draw, but not necessarily a problem.
The 'evolutionary' argument
on Nov 22nd, 2012, 10:30pm, SpeedRazor wrote:
Very concise, elegant, and poignant - (shouldn't this be 3 of the x criterion for an abstract game?). Thnx Nick.
 
Should games be bequeathed to the World before they are completely evolved/vetted? I say yes ... show us your [current] best stuff. Otherwise there never would have been a 'chess', for instance. Chess is a conglomerated consesus from myriad geniuses from around the world through the millennia, and it ain't even done ... . Put the game out now, but continue to fine-tune - "envision"/ troubleshoot - problems. I could be wrong, here, though ...  (I'm not a game designer).

This was in reply on Nick's quote above. I don't necessarily agree on the "myriad geniuses". The transformation to what basically would become the modern game took place in a very tight timeframe:
 
"The queen and bishop remained relatively weak until between 1475 AD and 1500 AD, in either Spain, Portugal, France or Italy, the queen's and bishop's modern moves started and spread, making chess close to its modern form."  
history of chess - wiki
 
This change was more of a revolution than a gradual evolution, and it may well have started with a small group of 'modern thinkers' deciding to rethink the whole concept.  
 
More importantly, the example also doesn't quite fit the argument. I've previously commented on the difference between 'organic' and 'mechanic' design. The latter is characterized by clever assembly rather than by invention or discovery. Go to BGG-abstract and you can find lots of threads about 'assembly' games, with various different pieces and various movement- and capture mechanics, serving quite a number of different goals. Often they are presented as provisionary, and posters are invited to comment on them, or playtest them and suggest improvements. There are always lot's of things to tamper with along the fringes, and every now and again something more fundamental is addressed.
 
These types of game are more prone to tampering and thus more prone to evolutionary changes than the four discussed above and Chess is there to prove it. But the only possible tampering with Reversi led to Othello, Havannah is a lucky and very specific merger. Ayu is quintessential, that is: there's nothing to tamper with. Inertia allowed one obvious choice: the movetype. These games are discovered rather than put together and they are not "tamper friendly". So Luis wasn't hunting cycles to somehow "correct" them, should they emerge, but just to see if they were possible. Because if you can't rule them out, you'll have to rule them in.  
 
Draughts would actually serve SpeedRazor's argument better. In 1947 Roozenburg scored 37 out of 40 in the first postwar tournament in the Netherlands and in 1948 he became "Sportsman of the Year". A Draughts player no less! Nowadays matches for the worldchampionship start with ten obligatory draws, and then the 'speed' and 'blitz' protocol takes over to eventually decide the winner. Draughts is definitely an 'organic' game, though exceptionally prone to tampering. The huge number of variants bears witness to that. At the time the rules of the international 10x10 game were established, the overwhelming combinatory power of the game blinded the ones in charge for the large percentage of drawn positions in the gametree, because they weren't looking with unbiased eyes, nor with 'modern' eyes. I've long heard the argument, even by top players, that the number of draws is dependent the attitude of the players. But the draws can only be found because they are there in the first place, and sometimes too abundantly so. So Draughts and Chess support the 'evolutionary' argument, but not because it holds for all games, but because they are "tamper friendly" and have a large enough margin of draws to make them subject to evolutionary pressure.
« Last Edit: Mar 16th, 2013, 9:32am by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #7 on: Mar 15th, 2013, 11:11am »
Quote Quote Modify Modify

Of small changes & big effects
One of the arguments Nick gives for thoroughly playtesting a game before passing judgement is that small changes can have a big impact. The butterfly effect. It's good to realize that it concerns sensitive dependence on initial conditions. Put the same butterfly in the center of the storm it caused, amid flying rooftops, cars and cows, and its flapping one way or the other won't matter much. So it's true, yes, but only for small changes at the core of a game. Although many would like to have it otherwise, tampering at the fringes usually results in a slightly different game.  
 
The Tampertown assembly line
Two thousand published chess variants cannot all have a big impact, now can they? Then why are there so many? Because it's so easy to assemble a chess variant, so easy to try, adjust, replace and try again. Its great fun and you can call yourself a game inventor. A guaranteed 99% successrate!
 
Or you let yourself inspire by Arimaa and the Jungle Game and come up with something along those lines. It's easy to visualize several concepts, all dependent on the nature and shape of the grid ('special' squares or hexes are popular) and the choice of pieces. Then the result is posted at BGG and discussions more often than not follow lines like: "I've given the Hippo the power to [-make your choice-] and it works fine and even helps to [-make your choice-] in the endgame.". It's all very social and workshop-like, but most of it is about game-assembly and tampering along the fringes. In those cases small changes usually don't have any major effect.
 
I've spend my early years as an inventor in Tampertown, and it was all very social and workshop-like and, yes, fruitful. A fair amount of chess variants gives evidence of that. But Grand Chess was invented a quarter of a century ago, and since then I've had two Tampertown hiccups, Hanniball and Cyclix. The first one illustrates my very point in that Arty Sandler had to come to its rescue by tampering it into a reasonable behaviour, the second one was based on a piece roulation system that made a perfect completion of the "Atlantis Triplets", three miniature chess variants with shrinking playing areas. Apart from these short visits, I've not seem Tampertown in a long time. The other games of the last two creative waves are all 'uniform' systems.
 
The discovery trail
Can one tamper with Ayu? Yes, actually Luis is doing just that. It started with my suggestion for a 'circular' hexhex set-up for HexAyu. It has one drawback: odd-sized boards would require a 7-cells vacant section in the center.
 
Luis in turn started to consider this set-up for the square board. In the current set-up odd-sized boards result in opponents each occupying opposing edges, while even-sized boards result in opponents each occupying adjacent edges. 'Circular' set-ups would not have that drawback, though slightly different set-ups would be required for sizes modulo (4k), (4k+1), (4k+2) and (4k+3).

 
So here's the set-up for (4k+3) boards. Eventually the applet at mindsports will feature all options, but for the time being, circular set-ups on square boards (as well as any set-ups on the hexhexboard) are still in the pipeline.
 
The reason is of course to compare the various initial positions in terms of gameplay, but I don't expect any major differences to emerge. This is a clear case of tampering along the fringes. The essence of Ayu remains the same.
 
The essence of Ayu
Ayu's invention was object driven. Luis may correct me if I'm wrong, but I think he was considering an organism of uniform but divided composition, intent on unification. The concept of a group comes naturally, but the mechanism may have been elusive, till he suddenly saw (!) the consequences of a simple mechanism for the 'intent' as well as a very simple restriction to keep it in check. To put it loosely "a group moves as a whole but must approach the nearest friendly group". 'Nearest' in terms of single steps.  
 
Mechanics versus intent
Many inventors' attempts are mechanics driven. Mechanics are manifold and distracting. Concentrating on intent is more demanding, but it gives focus. You can have a group on a board, and pick up one man and place it elsewhere adjacent to the same group, and the mind will wander of in every possible direction without getting anywhere. But look at it in the context of intent, as Luis did, and suddenly a whole new concept emerges!

The generic character of the core rule of Ayu is brilliant. It is applicable to grids of every size, shape and nature and there is an essential consequence in the restriction: If a group is isolated by the opponent, i.e. if there is no open orthogonal path between it and any other group, then it cannot move because there is no nearest group to move to. A shift in the object from 'unification' to 'immobilization' solves this most elegantly, and in fact makes it less rigid: you can win a game of Ayu while having more than one group left.
 
A small change at the core
In Ayu the core rule supports unification: the number of groups can never increase. It was a simple reversal that sprouted Inertia: forbid the number of groups to increase, and see what mechanics it suggests. This is a change at the very core of the game, the type of change Nick's argument that small changes may have big effects, refers at. The big effect is a new move protocol: pieces can now move between groups without increasing their number. The mechanics came last!
 
Ayu and Inertia have exactly the same object, though Ayu's 'self-immobilization' is called 'inertness' in Inertia. The games are are amazingly similar in structure, each employing one of two slightly different governing restrictions. But the resulting mechanics are so different that they lead to very different games. Ayu is the cleanest game in that the rules seem necessary and sufficient, not to mention self explanatory once the core idea is understood. Inertia features a move choice that is very good, but admittedly arbitrary, so it is not quite 'self explanatory'.  
 
The possibility of cooperative cycles doesn't harm Ayu at all, and anyone who harbors the opinion that "draws will find a way" may try to find even one such position. But I'm also confident now, after playing a reasonable number of games, that Inertia will most likely never have a draw problem in Ludotopia either. Of course all Ludotopia related statements are speculation. But for draws to be a problem you need obvious examples. For instance, Draughts' draw problem emerged after a century or so, but there was no lack of obvious examples of draws before that, now was there? In the absence of these, no game is likely to suffer from a draw problem. And I've seen no obvious examples in Inertia yet.
« Last Edit: Mar 15th, 2013, 12:01pm by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #8 on: Mar 24th, 2013, 9:27am »
Quote Quote Modify Modify

Here's an update after a number of increasingly interesting games we played in the exploration of Inertia. In this game you'll find some hopefully useful comments on strategy.
 
Jos Dekker (ger) - Christian Freeling (nl) (running)
IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #9 on: Mar 25th, 2013, 11:46am »
Quote Quote Modify Modify

Ingo Althofer sent me this mail:
"Now two reports on your Havannah match were published in the ICGA Journal.
You can find the table of content of that Journal issue in the appendix."

 
Attached was this jpg:
IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #10 on: Apr 1st, 2013, 6:42am »
Quote Quote Modify Modify

The first draw in Inertia
And an interesting one because it brings some general aspects into focus. The two trapped stones started out as two singles. That means that Red could seal off their enclosure without making all groups inert. The only obligation in such a case is to open it if and when the two connect. So Black had to connect and try to position the resulting 2-group in such a way that one of the stones could jump to the big group. Unfortunately the group is positioned in a way that allows Red to close a threatening sightline (to a vacant cell adjacent to the big group) while opening a path to the other side to avoid making Black inert.
 
The moral for now: don't get more than one trapped in a narrow pathway with a bend if the enclosing group is big enough. This position would have been a black win if the large red group had been 3 stones smaller (and the small group 3 bigger). So there's more to it than a simple 2x2 groups division.
 
It is a situation I of course tried to avoid in our next game.
« Last Edit: Apr 1st, 2013, 8:12am by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #11 on: May 1st, 2013, 8:51am »
Quote Quote Modify Modify

Inertia unwrinkled
So this is what happened, I didn't die but Inertia spawned a second draw and I saw problem arising because of the way it happened. I got the distinct feeling a player could "play for it".
 
I've recently commented on the difference between tampering and making small changes at the core of a game, and in retrospect this is a nice illustration. For starters, Inertia isn't very "tamperable" because it is effectively based on the interaction of one generic governing principle - the number of groups may not increase - and an upgraded hexrook move that allows a stone to start by sliding over any number of subsequent like colored stones (including zero) before proceeding over any number of vacant cells.
 
Here's where I got involved as an inventor. Clearly the governing principle, once it had formulated itself, would allow stones to conditionally jump between groups. The one move that suggested itself above all others is the most basic one: the hexrook move.
But it felt a bit flat. By allowing a stone to start its move by sliding over any number of subsequent like colored stones, groups would have their own limited ability to move and to conditionally jump from the far side of their group. It seemed to enhance tactics without any trade-off with strategy. So much for inventor involvement. Yet, with no tamperable variables available, I had to reconsider the move. "If the system is sound, the rule will be there" has always been one of my leading principles. So I reconsidered the move and saw one possible change that would not compromise the concept: extend the same move with the possibility to capture.
 
Capture emerging
It's not all that uncommon for a rook type piece to be able to capture, so in terms of internal logic there's no compromise involved. But the consequences in terms of gameplay are enormous. That's why I played a fair number of games against Ed van Zon and Jos Dekker before making the game public. Because something fundamental is lost and I wanted to make sure what is won.
 
What's lost
What I like most about Ayu is its constructive character. There's no placement, no growth and no reduction, an almost friendly organism trying to arrange and unify itself in the face of equal opposition, governed by the simplest of rules. Inertia was initially based on that very aspect.
That is definitely lost. Inertia trades it for a LOA type of unification where the conflicting organism are effectively reduced by mutual capture. You can actually win with one stone left. It also shares a characteristic dilemma: capture reduces the number of stones that the opponent must unify, so there must be different reasons to do so.
 
What's won
  • A shorter ruleset.
  • No forced cycles. Cooperative cycles are possible, but since they don't qualify as regular play they pose no problem. Formally you'll still find "3-fold is a draw" in the rules.
  • The object is "unification", not "immobilization of which unification is a special case". Inertness is no longer applicable. Ironically it does still lend its name to the game.
  • A move may not increase the number of groups and that means the sum of the number of groups of both players. This rule was already generic, but under the old rules a player could only alter the number of his own groups. Under the new rules a player can cut his opponent's groups, provided the move simultaneously unifies enough of his own groups to compensate for the increase in the number of his opponent's groups. The conclusion is that the already generic rule now also works generically, which enhances tactics by introducing brand new ones.
Unification versus immobilization
Ayu restricts a unit's move as 'towards those units that are at the minimal distance' of the moving unit. Since 'distance' is the defining criterion, there must always be at least two groups involved. If a player has several groups, but all of them are isolated, with no path between any two of them, then he cannot move. If it happens to be his turn at that point, he has won, despite not having unified all his groups. Inertia, under the old rules, needed a similar adaptation. Under the new rules it does not. However, one aspect has been maintained: if a player on his turn has only one group left, he has won. Let's call it 'strong unification' (as opposed to weak unification, as featured in LOA).
Inertia thus joins a set of games where you formally win after an opponent's move. Besides Ayu there are several others, of which I mention Epaminondas and Hexade. In Inertia it means that:
  • Simultaneous unification is a win for the opponent (since he on his turn finds only one group).
  • A unified group may be cut by a 'counter-unification' where the cutting stone simultaneously unites with another like colored group. Such exchanges may go on and they make for interesting endgame tactics, depending on who has the most resources.
The new ruleset has not been made public yet. Actually it hasn't even been written. We've been playing a 'shadow applet' tailored to support the new rules, but it has all taken place outside the ArenA. This weekend or so we'll make the change. Here's an example game between Jos and me. It's the first one I won against Jos, and since his level usually increases without notable dips, it must be one of the best games played yet.
 
And, why not, here's our current game.
« Last Edit: May 2nd, 2013, 5:45am by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #12 on: May 5th, 2013, 10:21am »
Quote Quote Modify Modify

Done. If you go to rules of Inertia now, you'll find the capture version. Same for the players' section. All games of the non-capture version have mysteriously vanished.
 
Inertia and Ayu both result from the interaction of one generic restrictive rule and one specific move. That's all both games require to generate deep strategy, certainly in Ayu's case, and innovative tactics, certainly in Inertia's case.
« Last Edit: May 5th, 2013, 2:28pm by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #13 on: May 15th, 2013, 4:54am »
Quote Quote Modify Modify

How to get more stupid and gain the wisdom not to care
It's easy, get older.  
 
So I missed the obvious once again, and, to my consolation, so did everybody else, including Luis who previously led me to reconsider game concepts or specific rules more than once.  
 
The first implementation of Inertia, the Ayu implementation so to say, showed emphasis on strategy, with tactics in a minor role, as in its parent game.
The second implementation, the one introducing capture, shows the reverse, and more than necessarily so. Here's why: the first and foremost move to present itself under the generic restriction of the game is and was the hexrook move. In the Ayu implementation this appeared a bit flat as I said at the time, and the decision to give the rook a "special" power, namely the option to start a move by sliding over a subsequent number of own stones, seemed appropriate because it widened the range of tactics. I explicitly mentioned this as inventor involvement as opposed to a system explaining itself.
 
When the Ayu implementation revealed a possibly problematic margin of draws, I made the above switch to "Inertia capture", which is a far more tactical concept with a very fluid behaviour and low on solidity. So what made me take along the "special" rook that increases tactics is a mystery. It drives tactics beyond balance. The normal hexrook move that suggested itself so emphatically is tailor made for the game. What was wrong was the absence of capture, not the move. This little fix will be implemented shortly and further simplifies the rules.
 
The restrictiveness of using the hexrook move (as opposed to the "special" hexrook move) can be felt by placing two red and two black stones on four subsequent cells. like this: R-R-B-B. Under the "special" move the outer stones can capture. Under the normal hexrook move only the inner stones can capture ... but they cannot because such a move would increase the number of groups.
« Last Edit: May 15th, 2013, 5:18am by christianF » IP Logged
christianF
Forum Moderator
Forum Guru
*****



Arimaa player #4019

   


Gender: male
Posts: 804
Re: Christian Freeling on inventing games (part 2)
« Reply #14 on: May 17th, 2013, 4:09am »
Quote Quote Modify Modify

Inertia now has hit the bottom in terms of simplification. New games will feature the normal hexrook move.
IP Logged
Pages: 1 2 3  ...  11 Reply Reply Notify of replies Notify of replies Send Topic Send Topic Print Print

« Previous topic | Next topic »

Arimaa Forum » Powered by YaBB 1 Gold - SP 1.3.1!
YaBB © 2000-2003. All Rights Reserved.