Author |
Topic: Essay by Christian Freeling on inventing games (Read 539994 times) |
|
MarkSteere
Forum Guru
Finite games rule
Gender:
Posts: 289
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #135 on: May 24th, 2010, 10:05am » |
|
I was seeing links to this discussion every day in the iggc chat. When I finally checked it out I had to answer the desperate cry for a reality check.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #136 on: May 24th, 2010, 10:48am » |
|
Here's an amusing contrast. One the one hand, I have played and studied Arimaa for about six years, and I think it is not a stretch to say that I am one of the world's foremost experts on Arimaa. Nevertheless, I am well aware that I can't know the future of Arimaa if it is someday played by folks who are significantly better at it than I am. Indeed, a portion of this very thread is filled with my skepticism of Christian Freeling's claims that he (or anyone) can know in advance of playing and developing expertise at a game what will become of that game in the long run. I take the "So what?" part of Mark's comments very seriously. Thus, even now, my most forceful predictions about Arimaa (including the one he quoted) are still qualified by words like "probably" and "unlikely". Indeed, even from the paragraph he quoted, Mark trimmed my qualifier, "but one never knows". On the other hand, Mark Steere, who is only superficially familiar with Arimaa, claims to know "exactly" what will happen in its future, although he advances neither argument nor evidence to support his claim. I expect I would dispute the argument or the evidence if it were there, but since neither is present we have only his expertise to dispute (i.e. his right to hold an opinion with no justification). In addition to the lack of substance, Mark also threw in inflammatory comments. This is classic trolling behavior. Mark, I understand that there are communities in the world which would benefit from a "reality check", but you have provided nothing of the sort to the Arimaa community. So far you have added heat while shedding no light. I invite you to choose between (A) backing up your claims in less incendiary form than your first two posts, or (B) leaving the Arimaa community in peace.
|
« Last Edit: May 24th, 2010, 10:54am by Fritzlein » |
IP Logged |
|
|
|
MarkSteere
Forum Guru
Finite games rule
Gender:
Posts: 289
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #137 on: May 24th, 2010, 11:06am » |
|
I'm going to leave you in peace. I hold court in rec.games.abstract, a moderator free zone. I recently started a new topic there, "Arimaa grandmaster society".
|
|
IP Logged |
|
|
|
christianF
Forum Moderator Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #138 on: May 24th, 2010, 11:42am » |
|
on May 24th, 2010, 10:48am, Fritzlein wrote:Indeed, a portion of this very thread is filled with my skepticism of Christian Freeling's claims that he (or anyone) can know in advance of playing and developing expertise at a game what will become of that game in the long run. |
| If that was my claim I'd share your scepticism. What will become of a game is different from how a game will behave in the long run. Some excellent games would behave perfectly, in that players can keep discovering new ideas and strategies with increasing insights, not unlike say Chess or Go. Nevertheless nothing will ever 'become' of them. Arimaa might be in that position, and without doubt the fast majority of my games too. I'd be glad (albeit dead) if two or three can still be found in wiki, half a century from now. So I never predicted Hanniballs 'future'. I just predict that it will behave properly in the long run, gamewise. And I gladly admit that some finetuning has been going on as the result of playtesting, and that I'm not out of the woods yet. Playtesting has only just begun and my neck is still on the block.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #139 on: May 24th, 2010, 1:39pm » |
|
on May 24th, 2010, 11:42am, christianF wrote:So I never predicted Hanniballs 'future'. I just predict that it will behave properly in the long run, gamewise. And I gladly admit that some finetuning has been going on as the result of playtesting, and that I'm not out of the woods yet. Playtesting has only just begun and my neck is still on the block. |
| It may be that the only thing we disagree about is what you claim to be able to know about a game prior to playtesting. As the evidence from playtesting Hanniball trickles in, I keep interpreting it as proving you wrong, whereas you keep interpreting it as proving you right! Perhaps I simply don't know what you mean for a game to "behave properly in the long run". Also I am not sure what you mean by "finetuning" rules, presumably as opposed to making fundamental changes. My thought about what makes a rule change superficial or fundamental is the extent to which it would invalidate the expertise of grandmasters. Let me use chess to illustrate. An example of a trivial rule change would be changing between a 50-move drawing rule and a 100-move drawing rule. If the Russians all played with one rule and all the Chinese played with the other rule and their respective champions got together to play a match, the better player would win regardless of which rule was in force. A moderately significant change would be randomizing the opening setup, as chess960 does. The top players will be similar in strength, but a specialist in one would likely gain a couple hundred rating points of "home-field advantage" over a specialist in the other. A fundamental change would be replacing the queen on each side with a second king and requiring double checkmate for victory. Since I haven't played Hanniball myself, it isn't clear to me which of your several rule changes were "finetuning", except for one. Changing from three actions per turn to four actions per turn was fundamental. If all the Russians played 3-step Hanniball and all the Chinese played 4-step Hanniball, and their respective champions played each other, the winner would be whoever got to play with familiar rules. It wouldn't even be close. So what does that prove? It might be that your introduction of a fundamental rule change proves nothing because you didn't need to make the change. (I believe you implied this earlier.) But suppose for a moment that 4-step Hanniball is playable at an expert level whereas 3-step Hanniball does not behave properly in the long run. Would that not disprove your ability to know in advance of playtesting how a game will behave? At some point, I will no longer be able to fence with you verbally about Hanniball without becoming good at the game myself. At that point I will have to retire from the debate even if I am not persuaded of your thesis. Nevertheless, regardless of how you and I variously interpret the data, I salute you for inventing Hanniball out in the open and reporting candidly on its progress. That is courageously done.
|
|
IP Logged |
|
|
|
christianF
Forum Moderator Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #140 on: May 24th, 2010, 2:50pm » |
|
on May 24th, 2010, 1:39pm, Fritzlein wrote:It may be that the only thing we disagree about is what you claim to be able to know about a game prior to playtesting. As the evidence from playtesting Hanniball trickles in, I keep interpreting it as proving you wrong, whereas you keep interpreting it as proving you right! Perhaps I simply don't know what you mean for a game to "behave properly in the long run". Also I am not sure what you mean by "finetuning" rules, presumably as opposed to making fundamental changes. |
| To get an idea across you have to exaggerate to some degree. Grand Chess is a good example of a game that supports my claim. Some opening weakness might have emerged that I missed, but I saw none. Given that why would Grand Chess not behave similar to Chess? HanniBall is another matter because the mechanics are quite new and the interaction is very complicated. That makes its long term behaviour much harder to predict. But I can predict that if no fatal flaw is found the game wil be rich in strategies and tactics. Outlines of both the possible flaw and at least the tactical richness have by now emerged. The possible flaw is that the 'catenaccio' approach might prove to strong. But it's not even clear which player has the better chance of getting the ball more or less savely surrounded into his own ranks in the opening stage. But you're right that I didn't get it 100% right. on May 24th, 2010, 1:39pm, Fritzlein wrote: Since I haven't played Hanniball myself, it isn't clear to me which of your several rule changes were "finetuning", except for one. Changing from three actions per turn to four actions per turn was fundamental. |
| Yes, that was less than a week after its genesis. Good fundamental change! on May 24th, 2010, 1:39pm, Fritzlein wrote:But suppose for a moment that 4-step Hanniball is playable at an expert level whereas 3-step Hanniball does not behave properly in the long run. Would that not disprove your ability to know in advance of playtesting how a game will behave? |
| Or would it 'prove' it? It's not provable is it, but it might suggest it: I changed it because the four moves version is the smallest that supports grabbing a ball, shooting it to an opponent, capturing the opponent and getting rid of the ball, in one turn. Kind of the essence of HanniBall capture. on May 24th, 2010, 1:39pm, Fritzlein wrote:At some point, I will no longer be able to fence with you verbally about Hanniball without becoming good at the game myself. At that point I will have to retire from the debate even if I am not persuaded of your thesis. Nevertheless, regardless of how you and I variously interpret the data, I salute you for inventing Hanniball out in the open and reporting candidly on its progress. That is courageously done. |
| Thank you, it's always a pleasure to read your comments, and I'd sure welcome you at iGGC .
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #141 on: May 24th, 2010, 4:46pm » |
|
on May 24th, 2010, 2:50pm, christianF wrote:Or would it 'prove' it? It's not provable is it, but it might suggest it: I changed it because the four moves version is the smallest that supports grabbing a ball, shooting it to an opponent, capturing the opponent and getting rid of the ball, in one turn. Kind of the essence of HanniBall capture. |
| So, it might suggest that your intuitions are highly informed by minimal playtesting. This puts you at the opposite end of a spectrum from those who can't see flaws in their favorite games despite a mountain of evidence. But even if so, the ability to intuit game behavior with no playtesting whatsoever would be off the end of the spectrum. Quote:To get an idea across you have to exaggerate to some degree. |
| Quote:Thank you, it's always a pleasure to read your comments, and I'd sure welcome you at iGGC . |
| I, too, enjoy reading your comments (when you aren't manipulating your audience through intentional exaggeration ), and enjoy trying to glean principles of good game design from you. Your writing is always lively and laced with insight. Although I insist that no one can know how a game will behave prior to playtesting, I also don't go to the opposite extreme of saying that generating random rules and playtesting is likely to create good games. Obviously people like yourself somehow manage to generate more sound game ideas than other would-be game inventors, and there must be a reason for it. Thank you for sharing your wisdom and experience.
|
|
IP Logged |
|
|
|
christianF
Forum Moderator Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #142 on: May 25th, 2010, 2:20pm » |
|
on May 24th, 2010, 4:46pm, Fritzlein wrote:So, it might suggest that your intuitions are highly informed by minimal playtesting. This puts you at the opposite end of a spectrum from those who can't see flaws in their favorite games despite a mountain of evidence. But even if so, the ability to intuit game behavior with no playtesting whatsoever would be off the end of the spectrum. |
| I surrender myself to the mercy of the court. Let's go for the minimal playtesting then. My latest 'finetuning' concerning the restrictions on the knight's move made the game slightly less positional and slightly more tactical. Not the way I would have gone naturally, but it would have been no problem ... if it had solved the problem. But it hasn't. So there's no need to divert from the previous version. The problem is catenaccio, because it is both boring and successful. It doesn't violate any rules and it hasn't led to draws, even to a substantial percentage of wins, but it's not the way the game wants to be played: victory comes like popping a pimple, a huge mass progressing ever so slowly, and then suddenly poof. I've been pondering this for a couple of hours now, looking for a generic solution without arbitrary parameters and I think I've found one. But I'm going to sleep over it to make sure.
|
|
IP Logged |
|
|
|
christianF
Forum Moderator Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #143 on: May 26th, 2010, 4:18am » |
|
on May 25th, 2010, 2:20pm, christianF wrote:But I'm going to sleep over it to make sure. |
| I slept over it. There's a Shogi proverb "if you find a good move, look for a better one" and it applies to rules too. The whole problem was centered around the ball and the definition of 'obstruction'. I once praised Greg Magne's definition as 'perfect' and it was indeed, if perfect is defined as 'without the currently known defects'. The unknown defect being catenaccio. So a new obstruction rule emerged that implicitly includes the current one. If Fritzlein would ask me how, my explanation would suffer from retrospectively applied logic. In hindsight every step makes sense. Of course not every step in the actual process did. Maybe I just got lucky. For the new obstruction rule it is necessary to define the ballsquare as 'the square where the ball is at the beginning of any given turn'. The squares of the goals cannot be ballsquares (because there would be no next turn). Around the ballsquare, there are rings: The first ring consists of the squares one king's move away from the ball. The second ring consists of the squares two outward king's moves away from the ball. The third ring consists of the squares three outward king's moves away from the ball. ... The 'N'th ring consists of the squares 'N' outward king's moves away from the ball. Goalsquares are excluded, implicitly as ballsquares, explicitly as ringsquares. The obstruction rule: If at the beginning of his turn a player finds that on any one ring (the first, second, third ...) the number of opponent's pieces is more than half the number of squares of that ring, then the opponent commits obstruction and the player is entitled to send one of his pieces off the board. Barring the keeper, this may or may not be one of the pieces causing the obstruction. If a non-obstructing piece is sent off, then the opponent, on his next turn, will have to undo the obstruction himself, or risk having another piece sent off next turn. The most important difference with the current obstruction rule is that it is defined in terms of rings around the ball rather then in terms of access by an 'open' route, given enough steps to get there. In the new definition, the number of defenders that is allowed, shrinks with closer proximity to the edges or the corners. Centerfield there's no problem: with four defenders on the first ring a player doesn't even have enough pieces left to violate the second, unless the ballsquare is on the b- or h-column, where the second ring has at most eleven squares. A ball in the corner has three squares on the first ring, five on the second and seven on the third. That's one, plus two, plus three possible defenders. It is not possible to distribute these over the three rings in such a way that an opponent's access to the ball is completely obstructed. Along the side it's a similar story. The best columns to push catenaccio would still be the b- and h- column. Along the left and right center, you can put four defenders on the first ring, five on the second. But this still is far to 'loosely packed' to deny an opponent access the way the current style catenaccio can. Implicitly the old style obstruction test has become obsolete. Though generically defined, the actual test a system has to perform would be on the first three rings, or four if you want to cover every eventuality (a player could put 5 pieces on the 4th ring of a cornersquare).
|
« Last Edit: May 26th, 2010, 4:24am by christianF » |
IP Logged |
|
|
|
gatsby
Guest
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #144 on: May 26th, 2010, 6:29am » |
|
on May 26th, 2010, 4:18am, christianF wrote:A ball in the corner has three squares on the first ring, five on the second and seven on the third. That's one, plus two, plus three possible defenders. It is not possible to distribute these over the three rings in such a way that an opponent's access to the ball is completely obstructed. |
| But it is four possible defenders on the fourth ring, and it is indeed possible to distribute the ten pieces of a player over the four rings to commit obstruction: . . . . . . x x x x . . . . . x . . . . . x . . . . x x . . O x x . . . | | | | | | | 4th ring | | 3rd ring | 2nd ring 1st ring O -> the ball in the corner x -> pieces (Pasting the diagram on Notepad will help)
|
« Last Edit: May 26th, 2010, 6:30am by gatsby » |
IP Logged |
|
|
|
Arty
Forum Full Member
Arimaa player #3423
Gender:
Posts: 16
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #145 on: May 26th, 2010, 6:57am » |
|
on May 26th, 2010, 6:29am, gatsby wrote: But it is four possible defenders on the fourth ring, and it is indeed possible to distribute the ten pieces of a player over the four rings to commit obstruction: . . . . . . |
| Didn't you put 7 defenders on the 3rd ring? (at least this is what I see after pasting to Notepad)
|
|
IP Logged |
Only those who attempt the absurd will achieve the impossible.. (Escher)
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #146 on: May 26th, 2010, 7:34am » |
|
I like the "font=courier" and "/font" tags. . . . . . . x x x x . . . . . x . . . . . x . . . . x x . . O x x . . . | | | | | | | 4th ring | | 3rd ring | 2nd ring 1st ring O -> the ball in the corner x -> pieces
|
|
IP Logged |
|
|
|
christianF
Forum Moderator Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #147 on: May 26th, 2010, 7:43am » |
|
Gatsby is right. It would appear that the new rule does not fully include the current one. So some more is required. Suggestions are welcome . Of course demanding that both definitions be used and both tests required is less than elegant. It is worth noting that in Gatsby's example all 10 initial fieldpieces are required. Having one piece captured renders the given example impossible. Also, keeping up such an arrangement may prove quite a puzzle. At the same time this type of arrangement is possible elsewhere too, for instance on the side: In this example the left side is the edge of the board, and only 9 pieces do the trick (2 on the first ring, 4 on the second and 3 on the third). x x x . . . x . . . x . o . x . x x x . . . . . . . . . It's even possible on the on the b-and h-columns (left side edge: 4 on the first ring, 5 on the second). x x x . . . x . . o x . . x x . x x . . One possibility is to change the maximum number of pieces per ring from "not more than half" to "less than half". This would reduce the maximum on rings with an even number of squares with one piece, while not affacting rings with an odd number of squares, Of the three example above, it would only change one though, so this isn't what we're looking for (though it's always worth to keep in mind). So we're not looking for a bandaid solution. We're looking for one definition that covers both previous ones.
|
« Last Edit: May 26th, 2010, 9:56am by christianF » |
IP Logged |
|
|
|
MarkSteere
Forum Guru
Finite games rule
Gender:
Posts: 289
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #148 on: May 26th, 2010, 8:01am » |
|
on May 26th, 2010, 7:43am, christianF wrote:So some more is required. Suggestions are welcome . |
| This is turning into a pretty big "nutshell"
|
|
IP Logged |
|
|
|
christianF
Forum Moderator Forum Guru
Arimaa player #4019
Gender:
Posts: 804
|
|
Re: Essay by Christian Freeling on inventing games
« Reply #149 on: May 26th, 2010, 9:29am » |
|
on May 26th, 2010, 8:01am, MarkSteere wrote: This is turning into a pretty big "nutshell" |
| I described the problem in a nutshell, I never said it was one, whatever that might mean. Systemwise the solution is simple: test on both definitions. One test is already implemented, and the new one is even simpler to implement (according to Arty). Rulewise this is less than elegant, that's why I feel one definition covering both criteria would be preferable.
|
« Last Edit: May 26th, 2010, 9:33am by christianF » |
IP Logged |
|
|
|
|