Last Man Forward Einstein forum

12 replies. Last post: 2016-02-28

Reply to this topic Return to forum

Last Man Forward
  • Ingo Althofer at 2013-10-19

    already back in 2006 I had designed a deterministic
    counterpart to the EinStein game: Last Man Forward.
    Pieces can move forward only in single steps. Always
    a piece with the farthest distance to the goal has
    to be moved. So, it makes sense to self-capture to
    make the remaining pieces faster.

    Now I realized a version for “Zillion of Games”. Those
    who have that software can download “Last Man Forward”
    and play against the bot.
    Last Man Forward

    The basic version is with 5 pieces on 5x5-board. But there
    are also variants, with up to 7 pieces on 8x8-board.


  • dandolo at 2013-11-03

    Nice game.

  • Ingo Althofer at 2015-12-08

    Hello again,

    end of June/start of July 2016 the next Computer Olympiad will take place in Leiden (NL). There is interest in having the game “Last Man Forward” on 8x8 board with 7 pieces for each player in the  event. Feel free to contact me when you also have interest.


  • William Fraser at 2015-12-09

    I am interested in trying to write a program.

    Do the pieces start on D1, C1, C2, B2, B3, A3, and A4?

  • zwack at 2015-12-10

    For what it's worth, the Zillions rule file defines the starting board as:


            (Red   (reddy2 a2 b1) (reddy3 a3 b2 c1) (reddy4 a4 d1))

            (Black (blacky2 g8 h7) (blacky3 f8 g7 h6) (blacky4 e8 h5))


    Since this is a little hard to visualize, the configuration looks like,a2i8b2h8b1h9c1g9d1i7a3i6a4f9

  • William Fraser at 2015-12-11

    Thanks, zwack.

  • Ingo Althofer at 2015-12-12

    Hi William,

    you are the guy behind the strong dots-and-boxes bots, right?!

    With you we would have already three parties for participation (one would be Andrew Lin, the second one myself). Would it be possible for you to come to Leiden in person? Andrew and me will be there - however we have already lots of obligations (I will operate my bots in Ewn and in Last Man Forward - and will operate David Fotland's bot in “Frisbee Go simulation” - and I have a request to operate a Go bot of Detlef Schmicker for some of the days…)

    The week in Leiden will definitely be fun. Likely also Mark Lefler (one of the two fathers of Zillions) will be there - and many others, for instance Richard Lorentz from the Breakthrough “Wanderer”.

    On the starting position: yes, 7 pieces in the corner, leaving a1, b1, a2 empty.

    Cheers, Ingo.

    PS. Thanks to Zwack for his explanation.

  • William Fraser at 2016-01-01

    I've got a problem.

    I think I've solved the game.

    The 8x8 board, 7 piece per side, database is only 32G and takes roughly 2 hours to compute.

    And it looks like the correct moves at the start are self-capture (specifics could be forthcoming, once I have a better UI), so one could probably optimize by computing a smaller database if one needed to.

    In terms of a deterministic EWN, I would suggest the following:

    Place the pieces as in traditional EWN.

    Roll an ordered collection of 35 dice for each player – like the bag in Golem Word Game.  (I believe that is the longest possible game….).

    One player chooses whether to:

    A.  play the Red stones

    B.  play the Blue stones

    C.  go first

    D.  go second

    The other player choose the remaining variable.

    It has been shown that exactly one of those 4 options is winning.

  • Ingo Althofer at 2016-01-01

    Hi William, congrats!

    So now at least two people have solved 7-piece LMF on 8x8 board (the faster one is Andrew Lin and did it three weeks ago).

    As a consequence, at the Olympiad LMF will have to be played on bigger boards (10x10 or 12x12) and with more pieces (for instance 9 per side).

    Please, contact me with mail at

    of course, you have to substitute ballaballa by the famous email symbol.

    Cheers, Ingo.

  • William Fraser at 2016-01-01

    Just to clarify, when I said “for each player” I meant “for each color” not “for each person”.

  • Ingo Althofer at 2016-01-02

    Hi William,

    > Just to clarify, when I said “for each player” I meant “for each color” not “for each person”.

    you are of course right.

    Perhaps you know that I am an enthusiastic member of the ICGA, in particular also supporting the ICGA Journal. I want to encourage you and Andrew Lin to write a paper on the solution of LMF. The nice thing is that we have two independent solutions, so they can be counterchecked.


  • Ingo Althofer at 2016-02-28

    Hello everybody:

    some news on “Last Man Forward”. Andrew Lin has solved the game on 10x10-board with 7 pieces for each side (player 1 starts on a4,a3,b3,b2,c2,c1,d1). So, also this variant is no longer available for the Computer Olympiad.

    A way out of the “solving-dilemma” might be the version with 4 types of stones: “a” and “A” belong to player 1, “b” and “B” to player 2. Turn order is cyclic with abABabAB… Starting position on 8x8-board with 7 pieces for each type is

    “a” on a4,a3,b3,b2,c2,c1,d1. Move directions are north, east, north-east.

    “b” on a5,a6,b6,b7,c7,c8,d8. Move directions are south, east, south-east.

    “A” on h5,h6,g6,g7,f7,f8,e8. Move directions are south, west, south-west.

    “B” on h4,h3,g3,g2,f2,f1,e1. Move directions are north, west, north-west.

    Moves may be to free cdells. Captures and “self”-captures are also allowed.

    Team aA loses when either all a-pieces are captured or all A-pieces are captured.

    Team bB loses when either all b-pieces are captured or all B-pieces are captured.

    Team aA wins if an a-piece reaches h8 or an A-piece reaches a1.

    Team bB wins if a b-piece reaches h1 or a B-piece reaches a8.

    This game would be too complicated to be solved with today's hardware and software.

    Feel free to ask questions when things are unclear. Ingo.

Return to forum

Reply to this topic