hist-games: Hnefatfl, Tablut, etc.

Mats Winther mwi9 at swipnet.se
Sat Aug 5 05:30:50 PDT 2006

I have continued my little research on the Hnefatafl family. I have
corrected a rather serious bug in my earlier implementation, improved
its strength, corrected setup of Brandubh, and added Large Hnefatafl,
and Alea Evangelii.

I have tested the following rules, proposed by someone on the
Historical Museum, Stockholm. In order to capture the king both
two-way and four-way interception are used. This could explain the
confusion between whether to use the one or the other. The solution is
that both were used. Four-way interception is used on the central
square, and on the four orthogonally adjacent squares. On the rest of
the board two-way interception is enough to capture the king. On the
four adjacent squares to the centre, four-way interception is formally
used, but as the centre square functions as capture square only three
pieces are needed to capture the king on these places. This rule
increases the movability of the king. Otherwise it would be very easy
to capture the king on these four squares as only one piece would be
necessary. Why would the king be less safe on his own courtyard than
anywhere else on the board? It would not be logical. This rule is very
logical and increases the chances greatly for the White side.

The four corner squares are the escape squares to be attained by the
king. They also function as capture squares. No other piece than the
king may stand on the centre or on the corner squares.

I have run the computer (1.6 MHz) against itself and have found that
Black has the upper hand, but only little. This is ideal, and is how a
hunt game should function. It is more demanding to play Black. It
demands greater understanding. White, on the other hand, must mostly
try to rely on tactical tricks. All the four variants that I've
implemented work perfectly. The Alea Evangelii is very demanding to
the computer, since the board is so big and there are so many pieces.
I ran it on a 1.6 Mhz with 2 minutes per move. In the 58th move Black
made an error which allowed white to escape with his king to a corner
square immediately thereafter. The program now plays strategically
very good as black. It avoids exchanging pieces for no gain.

It is not correct what they say everywhere on the Internet, that this
game is lopsided. On the contrary, I would say that it is perfectly
balanced. Otherwise it would not have been so immensely popular. These
rules are probably the correct ones. The escape to the corner squares,
and not merely the rim, should be the correct rule, because it carries
greater symbolical value. The four adjacent squares to the centre
should not be weaker than other squares. The use of capture squares in
the corners clearly improves the game.

I suspect that variants, unlike these, that seem clearly lopsided were
actually played with dice. It is probably necessary to make the game
favourable to black when using dice, because otherwise White could
make better use of his winning strategies. One such game is Ard-Ri. It
is evident that White will lose several pieces already in the opening
(if the opening setup is correctly understood). White doesn't stand a
chance. Perhaps it would be different with a die. We know, from fact,
that these games were sometimes played with dice, on the British
islands. I might implement the lopsided games as dice games to see
what happens.

One curious historical fact that Alea Evangelii was actually played on an
18x18 board, that is, a Go board proper. The pieces were placed on the
intersections, just like in Go. Pieces are captured by surrounding (only
more extensive in Go). Are these mere coincidences?

Mats Winther

More information about the hist-games mailing list