Hnefatafl Internet Championship

[Hagbard]

The timing idea of Cacreal in combination with the method above was tested

All games for the latest two years 2021 and 2022 were reviewed regarding the timing.

Had the timing method been the here mentioned, then
out of 12047 games only 23 games would have got into trouble with end game timeout limits < 8 hours, and only 7 players were involved.

[Hagbard]

This is my bid for an evaluation of the Championship Tournament 2022:

The tournament used the same format as last year, described here:
http://aagenielsen.dk/diagonalformat.php

Changes from last year:

• This year also a tournament line of Sea Battle Tafl.
• For each of the Hist. Hnef., Sea Battle and Race Games lines were used two game variants.

What worked well:

• That each strong player has only two opponents at a time gives a spectacular quality to the games. (Until 2020 strong players could have up to 7-8 opponents at a time.)
• To wait with every next round until the current round of all lines was finished.
• To run all game variants again as common tournaments, after their use in the World Tourn. was finished.
• A Beat The Winners round for all players after the World Tourn.

What could be improved next year:

• The setup should follow the description above.
• The minimum time out of 12 hours can be tightened to 6 hours.

The winner was found after 108 days (last year 86 days), and the whole tournament was finished after 108 days (last year 89 days).
33 players joined the tournament.
In total 348 games, 16309 moves and 5349 killed pieces.

Tournament page:
http://aagenielsen.dk/wtfturnering2022.php

Congratulations to World Tafl Federation Champion 2022, Mario Aluizo ("casshern"), Los Angeles, USA, winner of both Copenhagen, Hist. Hnef. and Sea Battle!
Also congratulations to Plamen Draganov ("Draganov"), Sofia, Bulgaria, winner of Race Games!
Many thanks to the four umpires Mike Coveny, Hamish Lawson, Adam Bartley and Mikkel Berg-Nordlie!!
And to the 33 players who took part!

Also a big thank you to Luk Martens ("Masterluke") who created the game pieces for all the variants!! (except Tablut)
And to Mikkel Berg-Nordlie ("mihkkal") who created the game pieces for the Tablut!

[Hagbard]

The Diagonal Tournament Format was tested again:
http://aagenielsen.dk/wtfdoubleelimtest4.php

[Hagbard]

Suggestion for the format of World Championship tournament 2023.

Start: Sep. 15th.

Format as tested May 2023:
http://aagenielsen.dk/wtfdoubleelimtest4.php

Four lines as last year:
Copenhagen Hnefatafl 11x11.
Historical Hnefatafl: Saami Tablut-w 9x9 & Welsh Tawlbwrdd-w 11x11.
Sea Battle Tafl: Sea battle-2526 9x9 & Sea battle 11x11.
Race games: Daldøs 3x14 & Saami Sáhkku 3x15.

Timing:
x4 (time buffer 5.5 days, incr. per move 8 hours).
When a game lasted a long time, the increment per move is gradually reduced.
Protected miminum time per move 2 hours (!).

[Hagbard]

Test of WTF-tourn. timing 2023.

The x4 timing, which is planned for the 2023 tourn., was tested using the Copenhagen variant:
9 players played 72 games. 6 of the games timed out = 8%. The longest game lasted 19 days.

Compare this with the WTF-tourn. 2022, also the Copenhagen variant:
25 players played 176 games. 30 of the games timed out = 17%. The longest game lasted 43 days.

So, the x4 timing now used resulted in fewer timeouts and shorter games, both cut by half.

Update 20.7.2023
Another test, using the Tablut variant:
6 players played 30 games. 1 of the games timed out = 3%. The longest game lasted 18 days.
Compare with WTF-tourn. 2022, also the Tablut variant:
17 players played 64 games. 12 of the games timed out = 19%. The longest game lasted 22 days.

Update 30.7.2023
And another test, using the Tawlbwrdd variant:
3 players played 6 games. 0 of the games timed out = 0%. The longest game lasted 18 days.
Compare with WTF-tourn. 2022, also the Tawlbwrdd variant:
7 players played 16 games. 1 of the games timed out = 6%. The longest game lasted 19 days.

[Hagbard]

World Tourn. 2023, background for the choice of variants.

Copenhagen.
We're used to the "Copenhagen Hnefatafl 11x11".

Hist.Hnef.
9x9:
The "Historical Hnefatafl 9x9 (Saami Tablut-w)" is well tested.

11x11:
Possibilities are
"Historical Hnefatafl 11x11 (Welsh Tawlbwrdd)", balance -1.21 (Tawlbwrdd Bell 11x11)
"Historical Hnefatafl 11x11 (Welsh Tawlbwrdd-w)", balance +1.13 (Tawlbwrdd Bell-w 11x11)
"Tawlbwrdd Lewis cross 11x11", balance -1.06
"Tawlbwrdd Lewis cross-w 11x11", balance -1.16

Normally "Historical Hnefatafl 11x11 (Welsh Tawlbwrdd-w)" would be the choice,
but the Moscow players who eagerly study the Hist.Hnef. and are also in top of the rating list, are very fond of the
"Tawlbwrdd Lewis cross 11x11". Could be with white begins, if we have enough experience with this before the tournament.

Sea Tafl.
9x9:
The "Sea battle-2526 9x9" is well tested.

11x11:
Possibilities are
"Sea battle tafl 11x11", balance -1.06 (Dardell setup)
"Sea battle-204577-w 11x11", balance -1.31
"Sea battle T circle 11x11", balance +1.30
"Sea battle tafl 11x11" has the best balance.

Race.
Daldøs:
Possibilities are
Daldøs 12, measured haphazardness 83%
Daldøs 14, measured haphazardness 57%
For some reason a high haphazardness is measured for the board size 12,
therefore Daldøs 14 is chosen.

Sahkku:
Sahkku is well tested.

[Hagbard]

Improvement of Round 1 setup.

At the moment 20 players would be set up for Round 1 this way:

Code: Select all

   00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19
00     x                 x
01  x     x
02     x     x
03        x     x
04           x     x
05              x     x
06                 x     x
07  x                 x     x
08                       x     x                       x
09                          x     x                       x
10                             x     x                       x
11                                x     x  x
12                                   x     x  x
13                                   x  x     x  x
14                                      x  x     x  x
15                                         x  x     x  x
16                                            x  x        x  x
17                          x                    x        x  x
18                             x                    x  x     x
19                                x                 x  x  x

A few details about this setup is a bit unfortunate:
- Player 00 has a much easier round than player 01.
- Player 01 (second player) is in considerable risk to be eliminated already in Round 1 by the two neighbouring equally strong players.
- Player 07 has a much harder time than player 08.

The following setup would be better:

Code: Select all

   00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19
00        x
01           x
02  x           x
03     x           x
04        x           x
05           x           x
06              x           x
07                 x           x
08                    x           x
09                       x           x  x
10                          x           x  x
11                             x           x  x              x
12                             x  x           x  x 
13                                x  x           x  x
14                                   x  x           x  x
15                                      x  x           x  x
16                                         x  x           x  x
17                                            x  x        x  x
18                                               x  x  x     x
19                                   x              x  x  x

- Players 00 and 01 have only one opponent in Round 1 and are thus not eliminated.

[Hagbard]

Simulator runs show, that to compensate for the one less opponent in Round 1, players 00 and 01 should have an opponent more in the next round.
So the Round 2 setup would for ten players be:

Code: Select all

   00 01 02 03 04 05 06 07 08 09
00              x  x           x
01                 x  x  x
02                       x  x
03                          x  x
04  x                 x
05  x  x
06     x        x
07     x  x
08        x  x
09  x        x

[Hagbard]

Code: Select all

   00 01 02 03 04 05 etc.
00        x
01           x
02  x           x
03     x           x
04        x
05           x
etc.

The Round 1 setup described above (Aug. 29) still has a flaw in the top.
The round is much easier for players 00 and 01 than for players 02, 03 & seq., even with a compensating extra opponent in the next round.

The top setup should be

Code: Select all

   00 01 02 03 04 05 etc.
00     x  x
01  x        x
02  x           x
03     x           x
04        x
05           x
etc.

If player 00 is considered to have the strength 6, player 01 strength 5 etc.
then player 00 meets the opponents' resistance 5 + 4 = 9
player 01 meets the opponents' resistance 6 + 3 = 9
player 02 meets the opponents' resistance 6 + 2 = 8
etc.

Dividing the resistance with the player's own strength gives a relative resistance = how tough the round is for the player:
player 00 meets the rel. resist. 9/6 = 1.5
player 01 meets the rel. resist. 9/5 = 1.8
player 00 meets the rel. resist. 8/4 = 2
etc.

Code: Select all

   resistance  rel.resistance
00          9             1.5
01          9             1.8
02          8             2
03          6             2
etc.

So, with this change the Round 1 is more equally tough to the top players, and it is still an advantage and slightly easier to be the topmost players 00 and 01.

[Hagbard]

Code: Select all

   00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
...
08                    x           x  x
09                       x           x  x        x
10                       x  x           x  x
11                          x  x           x  x
12                             x  x           x  x
13                                x  x        x  x
14                                   x  x  x     x
15                             x        x  x  x

There's a flaw in the low end also.
Round 1 is much harder for player 15 than for player 14.

The low setup should be

Code: Select all

   00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15
...
08                    x           x     x
09                       x           x     x
10                    x     x           x     x
11                       x     x           x     x
12                          x     x           x  x
13                             x     x        x  x
14                                x     x  x     x
15                                   x  x  x  x