Early Game

Has anyone figured out a good early game strategy yet?  I have been playing on and off for a while, and I am still stumped about how to play early on.

I currently play fairly randomly for a while, and use calculations and tactics to win (or lose) later on, but sometimes the game is already lost before I even start calculating.

Do any of the better players on this site have any tips on early game play?  This is what I have so far:

• Try not to move your back row guys until you need to, especially b,c,f,g files.
  
• Move your a- and h- file guys towards the center - that gives them more flexibility.
  
• Try to avoid 'losing moves' by trading your advanced guys for the opponents farther back guys.
  
• Don't cross the central line unless you can maintain your position, and/or cause disruption in the opponent's formation (i.e. force their back row guys to move).
  
• Advancing only on the flanks typically won't win, because your flank attacks won't be able to work together - you usually need some sort of threat in the center to break through (usually combined with a flank attack).  I learned this one the hard way.
  

Honestly, I don't have much more to go with until the fighting actually starts.  Do you guys have any other ideas (especially Ed!)?

I suspect that most other players, even the very top players, also thrive on calculation and are similarly clueless early on.  Is that true?

This is the reason why I am not a great fan of this game, I also don't understand the strategical points, I have basically the same ideas than you.

Edouard got something more than that for sure, but I know that he is winning a lot of his games by starting very soon in the game the calculation part.

Maybe we could learn more if we starting playing on a smaller board. It seems very big.

I think a pawn has more value if he is able to go everywhere on the board. So it loses some value when it crosses the main diagonals. Maybe this is a bit redundant with last point.

Also don't be the one to have to play when you reach the zugzwang point ;)

Just my 2c.

For small boards, see http://www.littlegolem.net/jsp/forum/topic2.jsp?forum=120&topic=19

I once extended TasmanianDevil's java program for 3x8 but java could not handle the memory needed. Someone should do it in C...

Can anyone solve 8x3 ?

A related problem is Dawson's chess which can be solved by combinatorial games: http://www.plambeck.org/archives/dawsonsemigroup.pdf

There is also work from Abdallah Saffidine, Nicolas Jouandeau, and Tristan Cazenave who have solved breakthrough with PN or PN² search up to 6x5 see https://chessprogramming.wikispaces.com/Breakthrough+(Game) but I can't find the solution in their paper... Maybe Richard J. Lorentz has more information?

If tltr :

The PPN² algorithm associated to race patterns has enabled to solve 6×5 breakthrough:

the game is a second-player win. We have found that on the smaller 4 × 5 board

the speedup due to parallelization is important until at least 64 clients.

In future work, we will try to solve breakthrough for larger sizes. The race

patterns used in this work had been devised by hand, but it is impractical if

we need many more patterns to solve statically positions earlier. We therefore

need to devise an algorithm to generate the race patterns and check them for

correctness automatically.

Zugzwang positions are still difficult to solve. Indeed, no winning race pattern

will be found in a zugzwang position, so an extension of the concept of race

patterns to be compatible with zugzwang positions or an orthogonal technique

would be desirable.

<br/>Hopping formatting will hold...<br/> +---+---+---+---+---+<br/>5| x | x | x | x | x |<br/> +---+---+---+---+---+<br/>4| x | x | x | x | x |<br/> +---+---+---+---+---+<br/>3|   |   |   |   |   |<br/> +---+---+---+---+---+<br/>2| o | o | o | o | o |<br/> +---+---+---+---+---+<br/>1| o | o | o | o | o |<br/> +---+---+---+---+---+<br/>   A   B   C   D   E<br/><br/>1. A2-A3: Loss in 26 moves<br/>2. E2-E3: Loss in 26 moves<br/>3. B2-B3: Loss in 24 moves<br/>4. C2-C3: Loss in 24 moves<br/>5. D2-D3: Loss in 24 moves<br/>6. A2-B3: Loss in 20 moves<br/>7. B2-A3: Loss in 20 moves<br/>8. B2-C3: Loss in 20 moves<br/>9. C2-B3: Loss in 20 moves<br/>10. C2-D3: Loss in 20 moves<br/>11. D2-C3: Loss in 20 moves<br/>12. D2-E3: Loss in 20 moves<br/>13. E2-D3: Loss in 20 moves<br/>

From Neutreeko:

java -Xmx400m Breakth55

For the pleasure:

<br/> +---+---+---+---+---+<br/>5| x | x | x | x | x |<br/> +---+---+---+---+---+<br/>4| x | x | x | x | x |<br/> +---+---+---+---+---+<br/>3|   |   |   |   | o |<br/> +---+---+---+---+---+<br/>2| o |   | o | o | o |<br/> +---+---+---+---+---+<br/>1| o | o | o | o | o |<br/> +---+---+---+---+---+<br/>   A   B   C   D   E<br/><br/>1. ?????: Win in 29 moves<br/>2. A1-B2: Loss in 22 moves<br/>

Can you believe it?

<br/> +---+---+---+---+---+<br/>5|   | x | x | x |   |<br/> +---+---+---+---+---+<br/>4| x | x | x | x | x |<br/> +---+---+---+---+---+<br/>3|   |   |   |   |   |<br/> +---+---+---+---+---+<br/>2| o | o | o | o | o |<br/> +---+---+---+---+---+<br/>1| o | o | o |   | o |<br/> +---+---+---+---+---+<br/>   A   B   C   D   E<br/><br/>1. ?????: Loss in 20 moves<br/>

Courtesy JK Haugland

Returning to David's original question, in November I asked edbonnet about Wanderer's play. I'm sure he doesn't mind me sharing his response as he has been quite open about his ideas for Breakthrough strategy, and I think his comments are apropos to this discussion.

Judging from the two games that wanderer lost in the championship, there might be something to improve in the control of the center.

In the opening, with White you want to establish pawns in c4-d4-e4-f4 (with Black in c5-d5-e5-f5). Why is that? Well, with Black your milestone is the Hadrian’s wall (zugzwang wall) and with White you want to reach the winning race wall. In both cases, you need a good grasp on the center.

Against Stop Sign and me, wanderer did a poor job controlling those central squares.

I'm not familiar with the terms Hadrian’s wall, zugzwang wall,  and winning race wall... if you don't mind explaining. Thanks!

http://www.littlegolem.net/jsp/forum/topic2.jsp?forum=120&topic=56

the above link discusses Hadrian's wall.  I had several games with this formation before it had a name.  It seems to be a true test of the game.

As far as opening strategy, David Scott has very good and correct concepts.   I used all of those concepts when I was playing regularly.

when the game was introduced at littlegolem in 2005, I had no concept of the strategy, but did a little research and found one useful idea  that served me well in my many years at the top.  Columns work well, they are hard to attack , so I sometimes used openings with the idea to build quick columns (for example, early pieces on b1, b2, b3 and b4).    This is just one opening idea.

Another very important concept in the opening is balance.  For example, if one player has 8 pieces on the a b and c files, then you might also want 8 pieces on those three files as well.  If you are outnumbered on a flank or in the center, you might be vulnerable, so a good defensive opening strategy is to maintain balance across the columns.  (If both players play this way, it can lead to Hadrian's Wall)

Of course it all comes down to calculation.  I found I could often calculate forced wins early (sometimes I would realize that I was lost by force, so then I would try to find monkey wrench moves).  But in truth, I would most often find myself in complicated positions and found myself calculating like crazy using my magnetic analysis board, only to determine the game was unclear.   Eventually I accepted that I was spending way too much time on analyzing each game, which is the main reason I decided to stop playing.

Interesting point about columns, Ray.  I also have found them decent - I used to experiment with creating columns on the c- and f- files, leaving gaping holes down the a- and h- files with only a minimal defence at the back.  The idea is to lure the opponent into advancing a wing piece deep down the side of the board (freezing two pieces with one - usually a good strategy for them), but the strong column would prevent them from coordinating with any other forces, and I would use my tempo advantage from the hasty advance to dominate the center.  I don't really know whether that actually worked - I won some games, but usually difference in calculation skill is what determines the winner of a game.  I also experimented with an almost opposite strategy of advancing on the a- and h- files, trying to wedge into the enemy camp, because whenever you are challenged, you just advance forward - you can't be exchanged off without wrecking your opponents back row formation.  Unfortunately, that strategy failed against Ray and wanderer, who would strengthen the center and prevent me from coordinating my advanced wing piece with anything else.  Of course, their superior calculation skills might have helped them win as well.  That's why it is so hard to figure out what works.

Now the present championship is peacefully ending, I can maybe add something to the discussion.

Let me briefly sum up the current situation.

As shown in many games, if White (resp. Black) does not put pawns in c4-d4-e4-f4 (resp. c5-d5-e5-f5),

she will lose mainly because the inherent lack of space will put her in zugzwang.

We already know that Hadrian's wall wins for Black. So, our concern is how to avoid Hadrian's wall with White.

You addressed that concern convincingly in our last game, and as you probably noticed, you were winning:

http://www.littlegolem.net/jsp/game/game.jsp?gid=1705158&nmove=35

At move 36, I played c7-b6 which is surprising since after your correct 37.f2-g3, Black cannot get Hadrian's wall;

if 38.g7-h6 then 39.g3-h4 or even more embarrassingly 39.g4xf5 followed by 41.g3-g4.

Fair enough. So, why didn't I play 36.g7-h6 instead? To answer that, one has to analyse the position arising after the

forced sequence 37.f2-g3 38.h6-g5 39.b3-b4 40.c7-b6 41.b4-b5.

This position (let's call it for the moment the hybrid wall) is the third (and last) crucial position without captures to analyse (the two others are the race wall and Hadrian's wall).

Considering this game or my game against kingofthebes, I think Black cannot prevent both the hybrid wall and the race wall, which is in favour of the initial position to be a win for White.

Wanderer does not seem to understand this hybrid wall. To satisfy our curiosity, what happens after 42. b6-a5? Can you play a few moves with Wanderer until he sees what you see?

Okay, I analyzed the position again, and apparently I missed something.

I thought that 43.b5-b6 was winning since 44.c8-c7 and 44.a5-b4 are both losing but I underestimated 44.a5-a4

which is in fact winning for Black.

By the way, there are "two" hybrid walls: one with a pawn in d1 and one with a pawn in a1.

Now, I think that both are winning for Black.

Even with a pawn in a1, 45.a1-b2 46.a4-b3 is a zugzwang.