Recommend
2 
 Thumb up
 Hide
32 Posts
1 , 2  Next »   | 

Abstract Games» Forums » General

Subject: Single disturbing life rss

Your Tags: Add tags
Popular Tags: [View All]
Mark Steere
United States
Palo Alto
California
flag msg tools
designer
publisher
Avatar
mb
A ko position which is not adjacent to all strings of a ko coupling or is adjacent to another string is a single disturbing life if - with each adjacent string to the ko position having as an assumption at least one breath not in the ko mouth - on it a ko stone move-sequence of infinite length could not be played or if it is a ko mouth that consists of 4 board points in a row and has not had on it two ko strings after a move.

Why?
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jonny Lawless
United States
Bountiful
Utah
flag msg tools
Love my girls!
badge
My name is Glenn! Long have I carried Cyrus's hopes and dreams, and now I bear the Masamune as well! Henceforth, I claim them as my own! I shall slay the Fiendlord Magus and restore our honor!
Avatar
mbmbmbmbmb
Huh?
11 
 Thumb up
0.02
 tip
 Hide
  • [+] Dice rolls
Mark Steere
United States
Palo Alto
California
flag msg tools
designer
publisher
Avatar
mb
This is the kludgiest thing I've ever read in my life:

http://home.snafu.de/jasiek/newko.html

Why even have a game (Ing) that's this complicated?
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
James Adrian
Canada
Edmonton
flag msg tools
badge
Avatar
mbmb
The New Ko Rules wrote:
The New Ko Rules are a successor of the Ing 1991 ko rules and furthermore explain them. They are an improvement, because they abandon methodical flaws and use clear definitions and rules.



Sheesh... I wouldn't want to read the original...



EDIT: Sorry. I screwed up my code...
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Mark Steere
United States
Palo Alto
California
flag msg tools
designer
publisher
Avatar
mb
The original:
http://www.usgo.org/resources/downloads/IngRules2006.pdf

As far as I can tell, the Ing system was intended to be a way of preventing a cycle of moves but which was not superko. Superko simply prescribes that you can't ever repeat a board position no matter how long the cycle of moves between repeated positions.

The Ing system was supposed to make Go perfect, but it didn't work. So now they've come up with this extraordinarily complicated amendment that now makes Go perfect.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Russ Williams
Poland
Wrocław
Dolny Śląsk
flag msg tools
designer
badge
Avatar
mbmbmb
Well, yeah; I've never understood why some people seem to have an aversion to superko (simply don't repeat a previous board position) which seems quite natural and elegant, and instead want to produce a huge network of convoluted complex rules full of bizarre special cases. (Japanese and Ing rules being the prime offenders...)


As far as I understand the motivation of those rule authors, it seem to stem from some instinct that certain positions "should" work a certain way, and that superko makes them not work that way, so instead of modifying their intuition about how the position "should" work, they tweak the rules to be more and more complex, but there are always more cases that appear that don't work the way their intuition says they "should", so the rules get more and more complex. It seems far preferable to have a simple elegant set of rules, and then discover how the various patterns really do work. The game is deep either way, and I'd much rather have it be deep with simple elegant rules.

It seems analogous to the old solar system models which presupposed that planetary motions "should" be based on circles, and so made ever-more complex circles within circles to model the observed planetary motions, intead of simply jettisoning the idea that it "should" use only circles, and changing to a more elegant simple model with ellipses.
14 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Eric Phillips
United States
Greensboro
North Carolina
flag msg tools
Avatar
mbmbmbmbmb
russ wrote:
It seems analogous to the old solar system models which presupposed that planetary motions "should" be based on circles, and so made ever-more complex circles within circles to model the observed planetary motions, intead of simply jettisoning the idea that it "should" use only circles, and changing to a more elegant simple model with ellipses.


I hardly know what you guys are talking about, but this is a nice analogy.
5 
 Thumb up
0.01
 tip
 Hide
  • [+] Dice rolls
Martin Jackson
United Kingdom
Bath
flag msg tools
badge
Avatar
mbmbmbmbmb
MarkSteere wrote:

As far as I can tell, the Ing system was intended to be a way of preventing a cycle of moves but which was not superko.


What's wrong with superko?
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Mark Steere
United States
Palo Alto
California
flag msg tools
designer
publisher
Avatar
mb
moik wrote:
What's wrong with superko?

Nothing, as long as you have a photographic memory. A position may recur unexpectedly after dozens of intervening positions. Simple is not always elegant. Superko is brutish.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Russ Williams
Poland
Wrocław
Dolny Śląsk
flag msg tools
designer
badge
Avatar
mbmbmb
MarkSteere wrote:
moik wrote:
What's wrong with superko?

Nothing, as long as you have a photographic memory. A position may recur unexpectedly after dozens of intervening positions. Simple is not always elegant. Superko is brutish.

This is a somewhat common theoretical objection by people who don't have much experience with go. In reality you certainly don't need a photographic memory for superko, any more than you need a photographic memory to detect simple ko. In real life, go positions simply don't "recur unexpectedly after dozens of intervening positions".
5 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Mark Steere
United States
Palo Alto
California
flag msg tools
designer
publisher
Avatar
mb
russ wrote:
MarkSteere wrote:
moik wrote:
What's wrong with superko?

Nothing, as long as you have a photographic memory. A position may recur unexpectedly after dozens of intervening positions. Simple is not always elegant. Superko is brutish.

This is a somewhat common theoretical objection by people who don't have much experience with go. In reality you certainly don't need a photographic memory for superko, any more than you need a photographic memory to detect simple ko. In real life, go positions simply don't "recur unexpectedly after dozens of intervening positions".

Of course, and I didn't imagine that they did. But they could. Yes it's a theoretical argument from someone with no Go experience, but we're halfway into the theoretical realm already since any problem not handled by ko and triple ko is rare.

Superko, as applied to Chess, Checkers, and all manner of infinite games in addition to Go, is a brutish concept. It's a gimmick for converting infinite games like Go into finite games.

In the quest for Go perfection, superko isn't very satisfying, at least not for me. The "usually not a problem/beyond the scope of this text" standard Go hand waving strikes me as anything but elegant.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Russ Williams
Poland
Wrocław
Dolny Śląsk
flag msg tools
designer
badge
Avatar
mbmbmb
MarkSteere wrote:
Superko, as applied to Chess, Checkers, and all manner of infinite games in addition to Go, is a brutish concept. It's a gimmick for converting infinite games like Go into finite games.

In the quest for Go perfection, superko isn't very satisfying, at least not for me. The "usually not a problem/beyond the scope of this text" standard Go hand waving strikes me as anything but elegant.

Ultimately it seems a matter of taste - in this case, what seems "brutish" to you seems quite "elegant" to me.

As far as I understand your design approach, you (implicitly or explicitly) prefer that the graph of all possible game states and moves connecting them is a directed acyclic graph, thus guaranteeing that every possible game is one step closer to some state with no out-edges to successor moves. I could easily imagine someone considering such an explicit "forced progress every turn" approach to be a "brutish" or overly "blunt" way to force a game to have guaranteed termination. (I don't personally consider it brutish, for what it's worth. I think both your DAG approach and a superko approach seem natural and elegant.)

I agree that there is theoretically a potential problem with remembering past states with a superko approach, and I could imagine it being a real-life problem in some games, but it's simply not a problem in go, so the theoretical objection doesn't bother me like it bothers you. If the game works and is strategically deep and interesting, that suffices for me.

The reasons I really don't see it as a problem, at least in the case of go:

1: Players would need to be intentionally trying to play in a perverse way to set up such hypothetically confusing repeating situations, instead of trying to play to win, and I'm not very bothered by theoretical problems that happen if neither player is trying to win. (The theoretical mathematical logical programmer tester geek inside of me cares, but the actual pragmatic gamer doesn't care.) Note that if it could be a problem if one player was trying to win and the other was playing crazily, then I would care more.

2: Decent players will not have trouble recognizing a much older prior position even if one does reappear. Plenty of players (who are not superhumans with photographic memory) remember the sequence of moves in a game. Granted, total beginners may have trouble with this. Given that long repetitions don't happen in practice, I'm not bothered by the possibility that a beginner might not notice them.


PS: It occurs to me that if the players don't notice that a repetition occurred, nothing bad happens anyway! Either they will do something different the second time through, and the game progresses, or they will keep repeating it and at some point even a total idiot would notice that they were repeating.
5 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Mark Steere
United States
Palo Alto
California
flag msg tools
designer
publisher
Avatar
mb
russ wrote:
PS: It occurs to me that if the players don't notice that a repetition occurred, nothing bad happens anyway! Either they will do something different the second time through, and the game progresses, or they will keep repeating it and at some point even a total idiot would notice that they were repeating.

This is nothing bad? Cycles tend to be self-maintaining. So if you went through a cycle once you'll probably go through it or a similar cycle again, and it may be some time before you even realize you're cycling. A game is supposed to entertain you, not make you feel like an idiot.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Eric Phillips
United States
Greensboro
North Carolina
flag msg tools
Avatar
mbmbmbmbmb
MarkSteere wrote:
A game is supposed to entertain you, not make you feel like an idiot.


In the right circumstances, that can be entertaining.
5 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Michael Howe
United States
Cromwell
Connecticut
flag msg tools
I'm with Mark on the superko question. I don't like a rule that requires players to remember previous board positions or keep a record of the game and refer to it to resolve rules questions. I think a ruleset in which the current board position completely determines the situation is preferable. The superko rule, or the ko rule for that matter, or the draw by repetition rule of chess, for example, are necessary to keep a game from going on forever, but they are necessary evils, if you ask me. In fact, eliminating the need for the repetition draw was one of the motivations for me to invent a finite, drawless chess variant (in which no one but me seems to have much interest, but such is the fate of chess variant inventors).
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
David Bush
United States
Radiant
Virginia
flag msg tools
designer
badge
Avatar
mbmbmbmbmb
MarkSteere wrote:
moik wrote:
What's wrong with superko?

Nothing, as long as you have a photographic memory.

A "photographic memory," whatever that may be, is not necessary for superko. If a position has repeated, that implies that every stone placed on the board since the first instance of the position has subsequently been captured. This is something go players with a modicum of ability would notice. If you're not strong enough to notice that, it is highly unlikely you would arrive at a superko position in the first place.
4 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Jonathan "Spartan Spawn, Sworn, Raised for Warring!"
United States
Sellersburg
Indiana
flag msg tools
designer
ΜΟΛΩΝ ΛΑΒΕ
badge
"By the power of truth, I, while living, have conquered the universe."
Avatar
mbmbmbmbmb
I feel Ive taken a wrong turn and ended up in the Twilight Zone. goo
3 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Mark Steere
United States
Palo Alto
California
flag msg tools
designer
publisher
Avatar
mb
twixter wrote:
A "photographic memory," whatever that may be, is not necessary for superko. If a position has repeated, that implies that every stone placed on the board since the first instance of the position has subsequently been captured. This is something go players with a modicum of ability would notice. If you're not strong enough to notice that, it is highly unlikely you would arrive at a superko position in the first place.

Admittedly it's a fine point but one that seems to provoke a defensive posture of highly unlikely this and highly unusual that. Go is an infinite game. The so-called superko rule is a hijacked, generic concept - one that makes the pie rule seem dignified in comparison.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Billy McBoatface
United States
Lexington
Massachusetts
flag msg tools
KGS is the #1 web site for playing go over the internet. Visit now!
badge
Yes, I really am that awesome.
Avatar
mbmbmbmbmb
I just a couple weeks ago had a discussion about Ing's attempts to avoid both superko and unending games. I'm on the side of superko. My point was that yes, in some cases superko makes groups that "feel" dead or alive actually be the other way around, but that's just a surprising position in go. Beginners are often surprised by a snapback; "huh? my group is dead? But I just captured something!" And even experienced players can be surprised by an under the stones tesuji. Living (or dying) thanks to superko is just one more weird thing to watch out for when you play with superko.

And I agree with Russ. I've played probably around 5,000 games of go, and I'm pretty sure that I've seen exactly ONE superko-requiring game, and it was a standard triple ko that is easy to spot - no photographic memory is required to apply it.

Edit: Superko makes pie rule seem dignified? Oh my no! Superko is simple: "You cannot create a loop that brings the game back to where it was before." Pie rule is crazy: "After your first move, AND ONLY YOUR FIRST MOVE, if your opponent really likes your first move, they can take it away from you and change colors." I give superko the elegance win here.
6 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Canada
Victoria
British Columbia
flag msg tools
mbmbmb
russ wrote:
Well, yeah; I've never understood why some people seem to have an aversion to superko (simply don't repeat a previous board position) which seems quite natural and elegant, and instead want to produce a huge network of convoluted complex rules full of bizarre special cases. (Japanese and Ing rules being the prime offenders...)


I don't really know why Ing objected to superko, but I will hazard a guess. A superko rule requires a copy of the entire game state to be retained somewhere - your memory, a mechanical reproduction, or some other information system. I don't fully understand (!) the Ing rules but they seem an attempt to resolve ko questions using only the existing board state and knowledge of one or maybe two previous individual moves. If the board is viewed as a world state (it figuratively spans ground to heaven, after all) with players who are finite state machines and no metamachines to record the entire board, superko can't be recognised or enforced. So Ing is perhaps an attempt to work only from the available information on the board without "cheating" by recalling the possibly distant past.

I personally have no problem with games that can end in draws, and rather like it when that happens occasionally, so a position where both players' best interest produces a continual cycle and ends play is fine with me. A lot of folks really want one winner and don't like that. The last half-point of komi is another interesting inelegance in that vein (speaking only of the idea of a "half-stone", not the issue of first player advantage and how much it might be worth).

The fun and beauty of Go is in how so much game emerges from so few rules - as you say, I'd rather live with an occasional idiosyncratic outcome than complicate the process in pursuit of some assumed goal.

I do have to wonder if those New Ko Rules are a machine translation from Cantonese - they're damn near impenetrable.

3 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Mark Steere
United States
Palo Alto
California
flag msg tools
designer
publisher
Avatar
mb
wmshub wrote:
Superko makes pie rule seem dignified? Oh my no! Superko is simple: "You cannot create a loop that brings the game back to where it was before." Pie rule is crazy: "After your first move, AND ONLY YOUR FIRST MOVE, if your opponent really likes your first move, they can take it away from you and change colors." I give superko the elegance win here.

Semantics. Player 2 can switch places with Player 1 as his first move. The pie rule isn't rocket science. Determining if the next move will complete a cycle might be, captured stone tallies notwithstanding. It's certainly not guaranteed to be obvious.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Billy McBoatface
United States
Lexington
Massachusetts
flag msg tools
KGS is the #1 web site for playing go over the internet. Visit now!
badge
Yes, I really am that awesome.
Avatar
mbmbmbmbmb
toh! wrote:
I don't really know why Ing objected to superko, but I will hazard a guess. A superko rule requires a copy of the entire game state to be retained somewhere - your memory, a mechanical reproduction, or some other information system.
I've heard several people who would have more direct information than me say that Ing's reason for discarding superko was because he felt that it some positions it gave the "wrong" result. I've never heard anybody say that Ing discarded it because it was too hard to use.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
David Bush
United States
Radiant
Virginia
flag msg tools
designer
badge
Avatar
mbmbmbmbmb
MarkSteere wrote:
twixter wrote:
A "photographic memory," whatever that may be, is not necessary for superko. If a position has repeated, that implies that every stone placed on the board since the first instance of the position has subsequently been captured. This is something go players with a modicum of ability would notice. If you're not strong enough to notice that, it is highly unlikely you would arrive at a superko position in the first place.

Admittedly it's a fine point but one that seems to provoke a defensive posture of highly unlikely this and highly unusual that. Go is an infinite game. The so-called superko rule is a hijacked, generic concept - one that makes the pie rule seem dignified in comparison.

Heh, as long as we're talking about fine points, it's debatable whether Go is infinite or not. From a Mathematical point of view, 19x19 Go is most emphatically not infinite. An upper bound on the state space is 3^361, large but finite. If by "infinite" you mean "beyond human measure," Well that remains to be seen does it not?

At any rate, what does the infiniteness or lack thereof of Go have to do with superko?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Mark Steere
United States
Palo Alto
California
flag msg tools
designer
publisher
Avatar
mb
Infinite is another way of saying cyclic.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Rodrigo Tjäder
Brazil
Rio de Janeiro
Rio de Janeiro
flag msg tools
Avatar
mbmbmb
twixter wrote:
Heh, as long as we're talking about fine points, it's debatable whether Go is infinite or not. From a Mathematical point of view, 19x19 Go is most emphatically not infinite. An upper bound on the state space is 3^361, large but finite. If by "infinite" you mean "beyond human measure," Well that remains to be seen does it not?

At any rate, what does the infiniteness or lack thereof of Go have to do with superko?

What he means by infinite is that the game can get stuck in a Cycle (http://senseis.xmp.net/?Cycles) and go on forever, never ending. The superko rule prevents that from happening.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
1 , 2  Next »   | 
Front Page | Welcome | Contact | Privacy Policy | Terms of Service | Advertise | Support BGG | Feeds RSS
Geekdo, BoardGameGeek, the Geekdo logo, and the BoardGameGeek logo are trademarks of BoardGameGeek, LLC.