Hnefatafl Internet Championship

[Hagbard]

Improvement of Round 2 setup.

At the moment players would be set up for Round 2 this way:

Code: Select all

   01 02 03 04 05 06 07
01  -        x  x              
02     -        x  x
03        -        x  x
04  x        -        x
05  x  x        -
06     x  x        -
07        x  x        -

Player 05 has a much harder round than all other players.

The following setup would be better:

Code: Select all

   00 01 02 03 04 05 06 07 08 09 10 11
00              x           x
01                 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
10        x           x
11           x           x

[Hagbard]

Improvement of Race Games Round 1 setup.

At the moment 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
00                    x  x  x  x
01                       x  x  x  x
02                          x  x  x  x
03                             x  x  x  x
04                                x  x  x  x
05                                   x  x  x  x
06  x                                   x  x  x
07  x  x                                   x  x
08  x  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

Player 09 has a much harder round than all other players.

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

[Hagbard]

National Champions 2023.

2013-2017 was in UK organised every year a UK Hnefatafl Championship tournament.
2017 and 2020 were national championship tournaments run on this website.

On occasion of the 25 years anniversary I'd propose that we have again national champions 2023.
Counting the games only of the latest year and grouping the results in countries and larger regions, this picture emerges:

USA.
1975 Casshern (151) USA
1503 Jrton80 (385) USA
1382 Staticmatic (19) USA
1368 Pasta (28) USA

Canada.
1764 Cacreal (123) Canada
1617 Holidayinnes (98) Canada
1582 Garun19 (20) Canada
1452 Mrcrow (20) Canada

UK.
1577 Chaolaibhne (20) UK
1541 Herjan (30) UK
1481 Xerxes (16) UK
1470 Ailill (46) UK
1447 Rideroflight (39) New Zealand
1391 Platypus (38) UK
1328 Tonythebook (381) UK
1199 Damianwong12 (37) UK
1135 Docbullen (26) UK

Scandinavia.
1528 Mihkkal (40) Norway
1349 Hagbard (41) Denmark
1321 Ekve2 (11) Finland
1236 Lalli the hunter (63) Finland

Benelux.
1564 Jilles (58) Netherlands
1560 Pc.a59xy (36) Netherlands
1435 Skye (139) Netherlands
1330 Masterluke (320) Belgium
1296 Fraech (94) Belgium

Germany.
1543 Colophonius (248) Germany
1434 Cromwell (28) Germany
1347 Luizz (26) Germany
1334 Themightyglider (197) Germany

Latin and Basque countries.
1671 Irish raven (35) Spain
1545 Floki (129) France
1529 Wesley alves (8) Brazil
1398 Archibald k. (18) France
1366 Branan (74) France
1352 Sybil (8) Italy
984 Mika (73) France

Central Europe.
1986 Draganov (1006) Bulgaria
1787 Odinhimself (48) Bulgaria
1621 Boucher228 (16) Moldova
1509 Potapych (318) Moldova
1434 Arcrue (55) Slovakia
1350 Brave-frog (676) Ukraine
1178 Konrad7890 (56) Poland

Russia.
1858 Alex hnefatafl (257) Russia
1772 Dimetr (245) Russia
1522 Paul7 (134) Russia
1446 Fiji270 (13) Russia
1332 Rosk (100) Russia


We'll soon find the Hnefatafl World Champion 2023,
and from the result of 2994 games among 49 players the latest year I propose that
national champions 2023 are:

USA
1975 Casshern (151) USA
#2 1503 Jrton80 (385) USA

Canada
1764 Cacreal (123) Canada
#2 1617 Holidayinnes (98) Canada

UK
1577 Chaolaibhne (20) UK
#2 1541 Herjan (30) UK

Scandinavia
1528 Mihkkal (40) Norway
#2 1349 Hagbard (41) Denmark

Benelux
1564 Jilles (58) Netherlands
#2 1560 Pc.a59xy (36) Netherlands

Germany
1543 Colophonius (248) Germany
#2 1434 Cromwell (28) Germany

Latin and Basque countries
1671 Irish raven (35) Spain
#2 1545 Floki (129) France

Central Europe
1986 Draganov (1006) Bulgaria
#2 1787 Odinhimself (48) Bulgaria

Russia
1858 Alex hnefatafl (257) Russia
#2 1772 Dimetr (245) Russia

[Hagbard]

WTF tournament duration and number of opponents.

The WTF tourn. 2023 lasted 81 days, and each strong player met at most 8 opponents totally, and only 2 at a time.

This is the shortest tourn. duration since 2013.
And the three years of Diagonal Tournament Format have the smallest number of opponents since 2011.

Code: Select all

year   no. of part.   duration     opponents

Round Robin:
2011        8            21             7
2012       13            42            12
2013       13            68            12
Groups:
2014       23           151            13
2015       19           110            13
2016       29            91            14
2017       29            87            11
2018       26           101            14
2019       25            87            13
2020       34           186            17
Diagonal Format:
2021       30            89             8
2022       33           108             8
2023       36            81             8

[Hagbard]

This is my bid for an evaluation of the Championship Tournament 2023:

The tournament used the same format as last year, described here (revised):
https://aagenielsen.dk/diagonalformat.php

Changes from last year:

• The Diagonal Tournament Format was revised.
• The timing was revised.

What worked well:

• The revised tournament format and timing x4 worked as it's supposed to.
• To give players extra warnings about timeouts if necessary, to minimize timeouts.
  This year effectively noone timed out unless he really insisted on doing so (except one player who was caught by surprice).
• To show the round setups beforehand to the players. Twice players commented some detail in a setup, resulting in an improvement.
• The final games were between the two strongest players, and the World Champion was not found until the last game finished.

What could be improved next year:
Perhaps timing could be tightened from 8 to 6 hours per move.


The winner was found after 81 days (last year 108 days), and the whole tournament was finished after 81 days (last year 108 days).
36 players joined the tournament.
In total 378 games, 16484 moves and 6287 killed pieces.

Tournament page:
https://aagenielsen.dk/wtfturnering2023.php

Congratulations to World Tafl Federation Champion 2023, Plamen Draganov ("Draganov"), Sofia, Bulgaria, winner of both Copenhagen and Sea Battle!
Also congratulations to Mario Aluizo ("Casshern"), Los Angeles, USA, winner of Historical Hnefatafl!
And congratulations to Mike Coveny ("Holidayinnes"), Ottawa, Canada, winner of Race Games!
Many thanks to the six umpires Steve Lonsdale, Daniel Kress, Adam Bartley, Mikkel Berg-Nordlie, Mike Coveny and Hamish Lawson!!
And to the 36 players who took part!

Again 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]

Quick timing is a big problem.
Compare fx. Los Angeles and Sofia.
In Los Angeles leisure time after work 17 o'clock until 22 o'clock when time to turn in
corresponds to Sofia 3 o'clock at night until 8 o'clock morning.
In theory the two players will never meet live and can only exchange one move each per day.

The WTF tourn. timing x4 was found by virtually running last years' games and see at which time parameter they stopped timing out. That was at 8 hours per move.

But that only works when all players move at least twice a day.
Which means for the most remote players, morning and evening.


The combination
- time per move 8 hours
- time buffer 5.5 days

has the effect that if players (two very remote players) are able to move morning and evening, that is 12 hours per move, then their time buffers are precisely used up after a long game of about 100 moves.


The remote player problem is special. Los Angeles is our most remote player location.
Playing over such distance means that the two players have many hours and lots of thinking time to make a move in the evening, but also have to make a move in the morning with no thinking time just moments before rushing to work.

A very remote player with response time 24 hours would work with time-per-move 19 hours.
Our only options are 2 hours (x1) or 4 hours (x2) per move.
That means that very remote players would have to choose the x10 timing when they want 24 hours per move and real thinking time.

[jilles]

Even if the players are relatively close, quick timing may still be rather unfair. For example, if a player in Sofia and a player in Berlin are playing, regularly making moves throughout the day, the player in Berlin will usually make the last move of the day, so that the night time will be taken from the Sofia player's time. In practice, this happens to work because making moves so often will refill the time buffer, but this still leaves the Berlin player with more possibility to delay.

A second surprising effect is with the lower time per move when the game has lasted for a long time. This goes so low that only the time buffer can defend against losing on time. Therefore, it seems a bad idea to use the time buffer to think longer about a move, since if the game lasts long enough, the player with the least time buffer left will lose.

I believe it must be accepted that "correspondence" format games last a long time. An alternative is playing "online", so games start and end within one or two hours.

[Hagbard]

I believe it must be accepted that "correspondence" format games last a long time.

Thank you for the thoughts!


It's a dilemma. All players, with a few exceptions, want quick games, and in reality it's not possible because of time zones etc.

The x1 timing (2 hours per move, buffer 2.5 day) was found by over a long period gradually tightening the timing, and finding that at x1 players generally still don't time out.

But we lost our Los Angeles player in the process.
His rating curve shows that it's a real problem. For many years it climbs completely steadily, until at a certain point in the time tightening the rating suddenly drops like a stone because of timeouts.
- And that happened to several others, too.

We tested extreme quick timing with 15 minutes per move and then 30 minutes per move. But after some time the test players left this option because it's not convenient in real life.


To ease play across time zones I'd imagine to add two options:
x5 and
x10 with rating.

The options would then be

• x1 (default)
  Time per move 2 hours.
  Time buffer 2.5 day.
  Avg. time per move for a 100 moves game: 3 hours.
• x2
  Time per move 4 hours.
  Time buffer 3.5 day.
  Avg. time per move for a 100 moves game: 6 hours.
• x5
  Time per move 10 hours.
  Time buffer 6.5 day.
  Avg. time per move for a 100 moves game: 13 hours.
• x10
  Time per move 20 hours.
  Time buffer 11.5 day.
  Avg. time per move for a 100 moves game: 26 hours.

(Hat tip Xerxes and Draganov who mentioned the time zones.)

[Hagbard]

Suggestion for the format of World Championship tournament 2024.

Start: Sep. 15th.
Four lines as last year.
Timing x5.

Tournament format.
Triple Elimination.
Round 1 set up as last year, but with 2, 4 and 6 opponents.
Round 2 the same, but with 2 and 4 opponents.
Described here:
https://aagenielsen.dk/diagonalformat.php

Simulator runs show that this format takes 4 rounds, the same as last year.
Last year with timing x4, each round took 20 days and the tournament lasted 81 days till Dec. 4th.
With timing x5, the tournament would be expected to last a month more, till New Year.

Race Games format.
Half Round Robin.
Round 1 runs every second game in the tournament table.
Round 2 runs rest of the games.
Points are counted from the two rounds; not elimination.

[Draganov]

Here is my evaluation on the last World Championship and some proposals for improvements.

1) Additional criteria:
Last year I won because of the game length (I won against Alex Hnefatafl in less moves than he won against me). However, this criteria seems too thin in my opinion. Here is my proposal for a better additional criteria to be used to resolve ties: Sonneborn–Berger score and direct comparison between the victories against shared opponents.
Last year Alex and I split the points in our match. We both played against Casshern (finished 3rd) and Cacreal (finished 4th). Alex won against Cacreal both games, I won 1 game against Cacreal and lost the other.
However, I defeated Casshern 3:1 while Alex lost to Casshern three times and won only one game. Overall, I played 6 games in total against Casshern and Cacreal and won 4 games and lost 2 times. Alex Hnefatafl also played 6 games against Casshern and Cacreal winning 3 games and losing 3 games. Taking this additional criteria into consideration makes the winner clearer.

2) Regarding the timing, I think the problem is that players are losing time during the night hours. I also saw that the time increment is not enough to compensate night hours. In my opinion, the minimum time increment must be at least 6 hours per move.

3) Fast-paced tournament for blitz players. In my opinion, we need a kind of a blitz tournament. Cacreal and I played such a fast match and it worked well for us since we both are fast players. However, the problem with the night hours still persists. Maybe some changes in the software could help us to have such a blitz tournament and don't lose on time. I think it will be good if both players have some mandatory amount of moves to make per day. For example 10 moves per day for each player. If one of the player can't play 10 moves, he will pay from his time buffer. The easiest way to implement this is to setup time increment of 2 hour per move and time buffer of 2 days. So, playing 4-5 moves in the night just before bed will secure the night hours. The winner of the tournament could be declared as the blitz champion.

4) Regarding the race games, I think the players should not be eliminated. Maybe it will be better if we play round-robin format and have Daldos in R1 and Sahkku in R2 and then combine the points to find out the winner. In case of a tie, we could have a Sahkku match of 6, 8 or 10 games between the tied players. Sahkku games are fast and round-robin or a match will not take much time.