Brandubh 7x7

Tafl rules
User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Mon Nov 29, 2021 10:04 pm

Alex Hnefatafl sent this post:

Introduction

Despite the lack of computer analysis, I can confidently say that brandubh has a much smaller tree of possible moves than any checkers on 64 boards. At the same time, the percentage of useful combinations is very small. As a result, this option can be considered only as training and is not an indicator of the player's skill. All combinations can be learned with due diligence.

There is only one option for starting the game and is symmetrical for all sides. This is d2-e2. Since d2-g2 as black leads to an instant loss without compensation, and on the move d2-f2 there is a winning combination as in the example game. (1) d2-f2; d3-f3 2) f4-f5; e4-e1 3) ×f2-f1; f3-a3 4) b4-b5;c4-c2 5)...) And then black cannot stop the escape.

The most popular response from White to the first move d2-e2 is c4-c2. Beginners often make the wrong move 2) b4-b2, which immediately leads to a defeat after the reply 2) ...; d3-a3. This is followed by: 3) × b2-a2; c2-c7 4) d6-e6; d5-b5. And it turns out the same position as in the case of an erroneous first move d2-f2.

The most popular continuation is 1) d2-e2; c4-c2 from the side of black is 2) d6-e6; ... Then, as a rule, follows: 2) ...; d3-a3 3) c6-c5. The resulting position is roughly equal. White cannot immediately start to run away to the corner because of 3) ...; d4-d3 4) × d7-d6; d3-b3 5) d1-a1; ... Therefore, it is necessary to look for another way out after the move 3) c6 -c5 Move a piece to d5 or e4.

It is also worth considering the second continuation on moves 1) d2-e2; c4-c2. Namely move 2) b4-b3. The rest of the answers are more passive and have a dubious chance of winning. For a more detailed analysis this format is not enough, therefore, I will limit myself to analyzing a typical black error. 1) d2-e2; c4-c2 2) b4-b3; c2-b2. The move 3) × d1-b1 is a mistake, since what follows is 3) ...; d3-g3 4) f4-f5; d4-d1. After this move White is guaranteed to win.

Also, we have not considered White's second answer. Namely 1) d2-e2; d3-d2. It might be worth discussing. In any case, building AI according to the rules of the brandubh would be a big step forward. I advise you to pay more attention to the rule of repetition of moves (White always loses on repetition). It acts not only with respect to 2-4 cells, between which there is movement, but also with respect to the entire orthogonal.

User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Sun Jan 16, 2022 9:46 pm

Damian Walker describes a version of Brandubh rules in a leaflet:
http://www.cyningstan.com/data-download ... ub-leaflet
Throne friendly to all. Corners hostile to all.
King captured by 2 everywhere, even on and next to the throne.
It is a simple small-board-effect that details of the normal rules may break down on the smallest board 7x7.
The 7x7 king has but 4 men, all placed right next to the throne, and as soon as the king leaves the throne, they could all immediately be captured, were the throne hostile.
So to avoid that, the throne must be friendly.

Were the corners friendly too, however, then attackers could easily block the way to the corners by standing next to them.
So the corners must still be hostile.

A normal throne protects the king well. Too well on the small board 7x7.
So in the rules described by Walker, the king is less protected and can be captured by two attackers everywhere also in the center of the board.

These changes seem a natural consequence of the small board 7x7.
But the game balance is always a surprice and one can never know before a thorough test.

Update Feb. 9th 2022
Game balance +1.27 from 69 games. It works.
Last edited by Hagbard on Wed Feb 09, 2022 8:02 am, edited 1 time in total.

Irish Raven
Posts: 5
Joined: Sun Oct 17, 2021 7:53 pm

Re: Brandubh 7x7

Post by Irish Raven » Tue Jan 25, 2022 11:41 am

For the record ,the version Brandubh 7x7 (Walker) implemented on the site looks to be a modified version of Brandubh 1 - not Brandubh 2. This means that the king is also captured if surrounded by 3 pieces next to the throne. This is not stated explicitly in the rules leaflet. See end of game Irish Raven v. Thermoglider in Brandubh 7x7 Walker tournament No. 281.

Irish Raven
Posts: 5
Joined: Sun Oct 17, 2021 7:53 pm

Re: Brandubh 7x7

Post by Irish Raven » Tue Jan 25, 2022 12:27 pm

Hagbard, Thanks for the modification and the reversal of the last move. It is now clear that Brandubh 7x7 Walker version is a development of the Brandubh 2 rules.

User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Tue Jan 25, 2022 12:56 pm

- And the Walker would be the oldest version then, from before Dec. 2014.

Irish Raven
Posts: 5
Joined: Sun Oct 17, 2021 7:53 pm

Re: Brandubh 7x7

Post by Irish Raven » Wed Jan 26, 2022 1:09 pm

@Hagbard - Can I ask for clarification as to what rule to handle move repetitions has been implemented in Brandubh 7x7 (Walker)? The repetition rule in Walker's rule leaflet - item 13 , seems to indicate repetitions can lead to a draw although I know on this site we have been moving to a 'White wins' in the event of repetitions.

User avatar
Hagbard
Posts: 691
Joined: Sat Mar 14, 2015 6:07 pm
Location: Copenhagen
Contact:

Re: Brandubh 7x7

Post by Hagbard » Wed Jan 26, 2022 3:23 pm

The standard here is used: Black wins.

Irish Raven
Posts: 5
Joined: Sun Oct 17, 2021 7:53 pm

Re: Brandubh 7x7

Post by Irish Raven » Wed Jan 26, 2022 7:12 pm

@Hagbard - Thanks for clarification.

fishcakes
Posts: 3
Joined: Mon Oct 18, 2021 9:35 am

Re: Brandubh 7x7

Post by fishcakes » Mon Jan 31, 2022 5:02 pm

Hagbard wrote:
Mon Nov 29, 2021 10:04 pm
Alex Hnefatafl sent this post:

Introduction

Despite the lack of computer analysis, I can confidently say that brandubh has a much smaller tree of possible moves than any checkers on 64 boards. At the same time, the percentage of useful combinations is very small. As a result, this option can be considered only as training and is not an indicator of the player's skill. All combinations can be learned with due diligence.

There is only one option for starting the game and is symmetrical for all sides. This is d2-e2. Since d2-g2 as black leads to an instant loss without compensation, and on the move d2-f2 there is a winning combination as in the example game. (1) d2-f2; d3-f3 2) f4-f5; e4-e1 3) ×f2-f1; f3-a3 4) b4-b5;c4-c2 5)...) And then black cannot stop the escape.

The most popular response from White to the first move d2-e2 is c4-c2. Beginners often make the wrong move 2) b4-b2, which immediately leads to a defeat after the reply 2) ...; d3-a3. This is followed by: 3) × b2-a2; c2-c7 4) d6-e6; d5-b5. And it turns out the same position as in the case of an erroneous first move d2-f2.

The most popular continuation is 1) d2-e2; c4-c2 from the side of black is 2) d6-e6; ... Then, as a rule, follows: 2) ...; d3-a3 3) c6-c5. The resulting position is roughly equal. White cannot immediately start to run away to the corner because of 3) ...; d4-d3 4) × d7-d6; d3-b3 5) d1-a1; ... Therefore, it is necessary to look for another way out after the move 3) c6 -c5 Move a piece to d5 or e4.

It is also worth considering the second continuation on moves 1) d2-e2; c4-c2. Namely move 2) b4-b3. The rest of the answers are more passive and have a dubious chance of winning. For a more detailed analysis this format is not enough, therefore, I will limit myself to analyzing a typical black error. 1) d2-e2; c4-c2 2) b4-b3; c2-b2. The move 3) × d1-b1 is a mistake, since what follows is 3) ...; d3-g3 4) f4-f5; d4-d1. After this move White is guaranteed to win.

Also, we have not considered White's second answer. Namely 1) d2-e2; d3-d2. It might be worth discussing. In any case, building AI according to the rules of the brandubh would be a big step forward. I advise you to pay more attention to the rule of repetition of moves (White always loses on repetition). It acts not only with respect to 2-4 cells, between which there is movement, but also with respect to the entire orthogonal.
Hi, this analysis is very interesting. Thank you! I was wondering if there is a source for the all Brandubh games played that I could download to perform data analysis on? To check these claims about specific continuations and such.

Colophonius
Posts: 2
Joined: Fri Feb 04, 2022 1:38 pm

Re: Brandubh 7x7

Post by Colophonius » Mon Aug 22, 2022 2:37 pm

After having played a match of ten games against Draganov, where Black won 9 times out of 10, we began to speculate, if Walker's rules could favor Black too strongly, given, that the player is experienced enough. That leaves the question, whether the rules for Brandubh 2 were actually more balanced than we thought with the majority of players just not being experienced enough, or whether the Brandubh 2-rules should be changed a little more in Walker's direction - not too much of course - for example by capturing the king on the throne with 2 pieces instead of 4.

What do you think about re-introducing Brandubh 2 in general (to give the more experienced players the opportunity to further test the balance) and capturing the king on his throne with just two attackers also in this variant in particular?

It could also be helpful, if games consisting of the exact same moves would not be counted each for the measurement of the game balance, because they all would have the exact same move pattern; therefor they should be counted only as one possible game. The archive is where all the games are registered.

Post Reply