It is currently Thu Mar 28, 2024 1:44 am

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 7 posts ] 
Author Message
Offline
 Post subject: Rules to goals. The different flavours of Go and Chess
Post #1 Posted: Fri May 13, 2022 2:11 pm 
Lives in gote

Posts: 470
Liked others: 62
Was liked: 278
Rank: UK 2d Dec15
KGS: mathmo 4d
IGS: mathmo 4d
Broad differences

In Go, every point is worth the same amount at the end. It takes a few moves (say 6) to complete control an area of your choosing even if your opponent plays closer, and it takes less (say 3) to secure it (as long as you defend it correctly). So Go has lots of little subgoals of securing larger areas, and negotiating over boundaries. The basic theory is that every move mostly affects the four neighbouring positions around it, attacking or defending them. Attacks on weak points are most threatening, and allow profit at the other three positions if the opponent must respond. However, only if an attack works, does it gain anything (from that neighbour).
Like Chess, friendly pieces almost always help, and the strength of your position is close to proportional to the number of them, throughout the game. However the position is more dynamic since pieces and move, and "commitment" mostly applies only to pawn structure, or other things you can't get back, such as the colour of your bishop, or the time it takes to move a knight to the other side of the board.
This is quite different to Othello, which I have little experience in, but which seems more random. But since it is apparently PSPACE-complete, perhaps it just needs more insight to allow preparation. Only the corners can possibly be pass-alive like in Go, so almost nothing is secure. But yet, for the game to finish at the end with secure points for each side, security must move from "capturable" to "uncapturable" over the course of the game. In Go, this tends to occur locally very quickly, whereas my impression is that it only occurs at the very end in Othello. Even the strongest players can lose by large margins to each other. You are only allow to play a move when you have a "clear line of sight" to it (with only the opponents pieces in the way), so the theoretical bases are about maximising the mobility or these lines of sight.

However Chess only has one key goal which is the capture of the opponent's king before your own (the etymology of chess is related to shah which translates to king). All positions are equal except for the initial configuration and the queening squares. Fortresses akin to Go are possible where if the opponent has few enough fighting pieces, then as long as you don't deviate, you can always defend your king. Though you won't necessarily be able to attack the opponent. So draws (by repetition, stalemate) may well be the best result. A king is able to slowly run away, but there is only finite space.

However, we can think of all such games as occurring in an abstract realm of space and time. This becomes a natural way to differentiate aspects of the game. For example referring to notes on complexity of Go problems, in the simplest positions, groups are just one move away from being captured, whereas in more difficult positions, there are multiple vital points and entrance points that either side can attack from, with more space.

Flows of value
In Go, control over secure areas is mostly fixed, and play is about balancing the development potential of each area and preemptively preventing the opponent from threatening your areas. If they have a valuable threat, you would like to have a valuable defense in return. This occurs because pieces can't move. However in practice "influence" can shift as mentioned above. If an area dominated by Black is attacked by White, they don't have to defend it as they can't turn it into territory with one move, but an attack may compensate by gaining influence on the other side of it. In top play, the accounting will balance so that the value of the gain is close to what was lost (plus or minus the time cost of playing a move). We can think of the influence as being able to move, but only if the opponent comes close. And certain weak points have very standard ways of moving with proper shape or counterattacks/defence points if the opponent attacks. Also, consider haengma. All value is local, and only masters need to consider long term global interactions.

In Chess, all pieces can move, but only in certain ways. So when defending, you only need to cope with fixed sorts of attacks, with a complex dance of manoeuvring with counterthreats. It seems there are many cascading critical points. There is a lack of time to make the situation as perfect as in Go since the situation keeps changing. If the opponent has no bishop, then you can leave diagonal weaknesses in your position for example. They may still be attacked by the queen, but the queen might be able to do less, since you are unwilling to sacrifice it to take out the opponents defence. In a way everything is position and manoeuvers with poking at the boundaries. There are subgoals of queening pawns, but early on, they are often distractions compared to king security (at least for weaker players).

Pawns are the weakest pieces, but they are at the boundary, and attacks must be made in coordination with their structure since they are so "dumb" that they can't move backwards, so they partially fix the shape from changing. Depending on the position, some pieces might be more valuable in your defence, and if they are necessary enough that they are irreplacable the opponent may suddenly (a la Tal) sacrifice one of their more valuable pieces since it allows sufficient attack on your king (perhaps enough that you will have to bring in a more valuable piece to the defence which can then be captured or else get checkmated, or simply that you lack time and resources to prevent a perpetual).

In Chess you have to make a move, and sometimes that means breaking your defence, especially if you only have pawns and vital defenders remaining. The AI has demonstrated more subtle concepts of using pushing pawns to control more space and break the opponent with zugzwang.

In Go, you have defensible positions fixed in shape, but all Chess positions are different, and thinking in terms of completing a defense is probably not for the strongest players, since chess is more dynamic with lots of tension and constant renewal of computation is required to defend as the shape changes. Perhaps it is just my inexperience, but I think that Go language and Go mathematics is more relevant to strength than anything similar in Chess. The concepts required for Chess are often tactical and hence unique to its ruleset. They can be memorised. Perhaps there is a reason that Go is played on a larger board. Still, "optimisation of positional advantages" remains interesting in any game, and maths can provide many helpful insights with proof.

Brilliancy

Sacrifice of more valuable pieces is often what is called a brilliancy in Chess (at least for weaker material based players), since this is so surprising to both sides.

In Go, there are such things as "tesuji" which settle a position by attacking/defending the weak points, and AI games often have a flurry of them that are very complicated. They are surprising because they attack from "inside" the opponent's region (normally you profit from an attack by playing on the outside and forcing the opponent to retreat inside), but when the opponent is weak enough and you are strong enough, sometimes this allows you to capture more stones.

Temperature
In Go, temperature is the basis of much theory, since optimal play is about accounting for fixed sources of value. Physicists shouldn't read much more than this from the concept's name. In Chess, you can give numbers to the values of pieces, and that was mostly sufficient for Deep Blue. However, as the value of pieces depends on the positions, perhaps non-equilibrium thermodynamics becomes more relevant.

Boundaries
As in any two player game that is settled by "control" over some intersection, at the end of the game, key intersections are either controlled by black or white. However, how does this get settled before the end of the game. Most often, we talk of boundaries (mostly in space), between what is most likely Black control and what is most likely White control.

In Chess we have more defence oriented structures or more attacking oriented. In Go, we have thick and thin. In Go, if we have large territories (such as in the endgame), the fighting is about what profit can be made at the boundary, and profit is largest if we can threaten to penetrate the large territories (which is only possible against thin groups). In a sense, this is like the defence of the king in chess, except that each boundary tends to be independent. In Go, thin coverings are most easily punctured or lived in by making threats to cut (and divide and conquer), and yet that tends to mean they likely weren't worth much anyway (no man's land, not territory for either side), or that the opponent has their own problems to worry about such that they don't have time to both attack and defend. Yet this requires strength to manage, and this is the sort of Go intensity and reading out tactics that top players worry about.

In both games, you fight for control of vital points and breaking through. However in Chess, both sides get lamer with the exchange of pieces, and we can say that "empty space" wins. Whereas in Go, the capturer tends to have very strong local control, so areas tend to get increasingly solidified, and one of the players wins. Perhaps there is a reason we avoid draws in Go?

Case study: Go shimari
I present my updated thinking about weak points in Go. AlphaGo shocked us with new variations of attaching to shimaris. Why?

Against a single corner stone, the corner is the main weak point and moves should threaten to chip away the corner (perhaps by capturing the first player's move if you have enough support), Moves are more valuable since they have a larger threat.

Against a shimari which is only two stones in the corner, they control the corner better regardless of if the opponent approaches (6-3 was the traditional approach to the 4-3 3-6 enclosure), so threats to cut through become more valuable locally if they also cut off the main development of the shimari. This is less valuable than getting the corner, but still one of the top opening moves (say move 8-20). However, since you will have less access to the corner, you have less eyespace, and therefore you are more vulnerable to attack. Hence, control of the side (say the middle star point) becomes more important for securing the development value of shimaris. Note that control isn't strong without eyespace, so normally, neither side plays around the middle star point directly, unless they have support (normally by approaching the opponent's corner). Also, moves from the centre maintain threats against such weak points, so there is still plenty of value to fight over even if one side or the other adds a move.

Silly koan

The board positions changes the value of moves. What is the change of a choice?

NB: A larger group nearer to the edge of death increases the value of moves, but caps it due to the possibility of a response (sente gains nothing), but if this is a double attack, or saves your own group, then it helps. Even getting strength near to your own strength can be very valuable if it sets up a chain reaction making the opponent weaker nearby. Consider Hua Yigang's repeated mentioning that in the modern era, we consider low-low harmony good as well as low-high harmony.

Top
 Profile  
 
Offline
 Post subject: Re: Rules to goals. The different flavours of Go and Chess
Post #2 Posted: Fri May 13, 2022 4:52 pm 
Lives in sente

Posts: 905
Liked others: 22
Was liked: 168
Rank: panda 5 dan
IGS: kvasir
Chess play is full of other goals besides the singular goal of capturing a king. Maybe because the pieces do move it is natural think in terms of goals and plans. It is typical that the position will evolve from one type to another, changing the players approach and sometimes the advantage. Maybe the concepts that many people have of chess are not applicable to the kind of chess played by 99.9% of chess players, not sure. What usually counts in human play seems to be just improving the position.

I'd say Go has a lot less variety of goals and plans but it depend on perspectives and definitions, what exactly is a goal? At least playing Go with the idea that you are going to achieve specific outcomes on the board is detrimental in my experience, but maybe it is the only way to play chess? Maybe only my own perspective.

Top
 Profile  
 
Offline
 Post subject: Re: Rules to goals. The different flavours of Go and Chess
Post #3 Posted: Wed Jun 29, 2022 11:03 pm 
Lives in gote

Posts: 470
Liked others: 62
Was liked: 278
Rank: UK 2d Dec15
KGS: mathmo 4d
IGS: mathmo 4d
To me, this sounds like in Chess things are treated more like discrete combinatorics, whereas in Go, the combinatorics is so symmetric that it becomes possible to find general "numbers" but chopping up the discrete value components, dividing by powers of 2. So you shouldn't think that one move achieves anything discretely specific to any particular intersection. Instead, a bit more abstract imagination is required, combining things in different ways than the lines of the board.

Top
 Profile  
 
Offline
 Post subject: Re: Rules to goals. The different flavours of Go and Chess
Post #4 Posted: Tue Jul 26, 2022 5:26 am 
Lives in gote

Posts: 470
Liked others: 62
Was liked: 278
Rank: UK 2d Dec15
KGS: mathmo 4d
IGS: mathmo 4d
mobility, space, control, attack/defence

Trying to think about parallels.
So many concepts are shared or borrowed? But are they really optimal?

In chess, there is the time cost of moving pieces. In Go, although pieces can't move, we still talk of mobility of groups, access etc. This is because we identify the connected components as the emergent building blocks of the game. It is their weak points and potential territory that create big moves. If weak points of the same group are near each other (e.g. hane at the head of 2 stones), then the opponent has powerful moves there as one move has access to many weak points. If that group can cut its weak points apart, then the opponent has less access and mobility in attacking in weak point and may have to split if they want to attack both. In similar ways we can think of space advantage in go as a situation where even if your weak point can be attacked, the opponent doesn't have much space to do so.

Go is based on control, whereas Chess isn't. But it obviously still matters. However, the relation between heavy pieces and control is a bit different. The queen in chess isn't valuable because it can occupy and hold a square. If anything the opposite may be true as it is so valuable it is more easily kicked, even by a supported pawn. However, its mobility and hence influence spans the whole board. It is fast to switch attacking points. There isn't much similar in Go that I can think of as connection is so important, so moves tend to be slower. Perhaps a large group can be thought of as a queen since it concerns many squares, but the group doesn't move as such. Instead it is the weak points (and key moves, vital points) that can move around the group. Over time, the weak points tend to clump together in concentration until the climax where the group is near the boundary of life and death and the weak points are the most valuable (affecting the whole group, not just the local area) before it settles one way or the other (normally alive).

In Chess, attack is also often not necessarily to kill but for more subtle profit, such as space, other positional gains, exchanges. We can think of pieces as more valuable to control than normal squares because (like in Go), they influence more than one square, so taking them will not only help your control of that square but also remove your opponent's influence via that piece. However, what is the cost? If your opponent has support on that square, then you lose your attacking piece. In Go, it isn't quite as direct, but similar principles hold. There is a scale (order) on support/control/value of attacking groups. And if you attack with a weak group, your opponent's counter-attack will be more powerful if they have more support and more valuable if your weak group is larger. However, if you do succeed in killing with a weak group, there can't be a bigger profit than that since the weak group gets strong from it, getting something from nothing. There isn't quite the same possibility in chess as all pieces remain weak with no such thing as pass-alive. The closest thing is probably queening, since a queen is much more resilient to being able to escape attacks due to its mobility. Mobility helps escape in Go, but eventually you run out of space as more pieces are placed on the board, so local life is required. Otherwise, it isn't that dissimilar from getting into zugzwang. Intuitively making life isn't that far from needing to kill the opponent's king locally for each living group, except it is much more simple in Go.

Openings (weak points) are only valuable to attack if the attack "works" as a combination. I find it hard to understand this concept in Chess as simply as in Go, because Go rules are simpler and more symmetric, so a combination tends to just mean that the opponent has n weak points, so you may get n moves in a row by threatening them (if the opponent responds). Like Go, multiple moves tend to be required to accomplish any particular goal, but there are many possible goals and so one move can threaten many goals. This makes the game difficult.

In Go, direct attack is unlikely to be helpful or powerful until a group is very weak or large and weak. Instead every attack must balance the value of the attack with the territory it builds.

Beyond this, I probably don't understand Chess enough. Chess seems much more human, and designed and hence more complicated to me with subtler tactics than in Go.

And yet if heuristics worked to win, perhaps the rules complexity makes strategy and evaluation simpler with more obvious imbalances and critical points? I suspect it is more the smaller board rather than the rules complexity though. Go to me is more like physics than war.


Last edited by dhu163 on Tue Jul 26, 2022 5:32 am, edited 1 time in total.
Top
 Profile  
 
Offline
 Post subject: Re: Rules to goals. The different flavours of Go and Chess
Post #5 Posted: Tue Jul 26, 2022 5:32 am 
Gosei

Posts: 1494
Liked others: 111
Was liked: 315
How can you assert that Chess is not based on control, I mean isn't control over the squares fundamental?

_________________
North Lecale

Top
 Profile  
 
Offline
 Post subject: Re: Rules to goals. The different flavours of Go and Chess
Post #6 Posted: Tue Jul 26, 2022 8:37 am 
Lives in gote

Posts: 470
Liked others: 62
Was liked: 278
Rank: UK 2d Dec15
KGS: mathmo 4d
IGS: mathmo 4d
I mean relatively. Winning in Go is sum of control. Control in chess is more of a subgoal than main goal.

Top
 Profile  
 
Offline
 Post subject: Re: Rules to goals. The different flavours of Go and Chess
Post #7 Posted: Tue Aug 09, 2022 6:59 pm 
Lives in gote

Posts: 470
Liked others: 62
Was liked: 278
Rank: UK 2d Dec15
KGS: mathmo 4d
IGS: mathmo 4d
I was once a weaker than average club level player.

Perhaps I can point out some parallels, having spent a few minutes studying chess for the first time in many years.

Piece coordination in Chess seems akin to efficiency in Go, though the ability of the pieces to move adds many dimensions. We can think of virtual control if a piece can move without being harassed from its current square to a place where it could be of use.

If information is value in wordle (max information leads to max score), value is value in Go and attack/defense is value in chess.

Note more info means more casework, more complicated strategies. In a game like go with perfect information and huge numbers of possible strategies, game design requires some simplification so that optimal strategy isn't too complicated. Sentestrat for example, which also applies in opening.


Just as groups are hard to kill in Go, in chess it is expected that the kings are hard to kill. so small advantages for the endgame are important.

Just as Tian Ji's strategy is relevant in Go with diminishing returns to control, W, the side with weaker stones is interested in making sacrifices to reinforce the positions they have influence over to become strength. Likewise, if the opponent is attacking in Chess, you would like to exchanging off pieces, or at least control their vital development squares so that they can't develop. Likewise the value of weak points or vital points spreads so that if the pieces you use to control them can be vulnerable, the value of attacking them can be worth up to the value of your control piece and the square being controlled. This can sometimes spiral upwards, but this normally only happens if one side has missed a tactic.

There are balance equations in chess of tension. If you attack with a bishop, perhaps I can defend with a knight etc. All akin to the boundary of life and death in Go, but not quite since chess can involve much manoeuvring without anything happening. In Go (almost) every move does something irreversible (under optimal play). Chess seems more violent to me than Go, but I'm not sure that is really rational, as in Go it is the empty spaces that tend to "die" silently, turned into territory.

There is a sort of aji that arises from having more space. If you push the pawns forward then you can manoeuvre and develop with more options to optimise over and even more speed to change from one formation to another. Meanwhile the side with less space is choked and sometimes must sacrifice to gain options.

There is also infiltration. Development is important to make threats and counter them. But the shape of threats depends on if each side is threatening to infiltration, attacking pawns, pieces, or the king. Moving the queen early is unhelpful as it can always move. We can say the marginal advantage is small.

"Experience" is probably a guide to estimating the relative value of different types of advantages, or at least what to look for (what functions of the board), much like in Go. Strength is the confidence about what the real maximum threats of each side are and hence precisely what to defend against.

We still have weak points based on if there is a soft exposed part of the defence. Perhaps my Go theorising topics shows my chess history.

Both have leftover light groups. Once fighting is over and both have avoided major losses relative to par, each may have annoying small advantages left. Not yet a threat, but can become dangerous if its potential is ignored. For example advanced unsupported pawns.

Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 7 posts ] 

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group