Does anyone have expirience with the misere version of dots and boxes (ie: where you are trying to lose)? Intuitively I would say that the games would rarely be close but I have no clue on strategy.

Is there something similar to the chain rule? Would anyone like to play? (we can just play dots and boxes unrated here and agree that we try to lose).

Yes I would like to play.
Usually misere forms are more complex than normal form even for simple combinatorial games, see Dawson Chess for example (which looks a bit like Breakthrough).
I can see why it should not be close: if closing is square is not mandatory, all the squares will have three sides up to the end, where last player will close them all, see for example 2x1 where 1st player wins 2-0 by playing central line…
Or do you want to force a player to complete a 3-sided square?

There can be close games and parity change battles…

This paper (http://erikdemaine.org/papers/DotsBoxes_GONC4/paper.pdf) shows how first player can win by approximatively n/6 on a 1xn board and conjectures a win for second player on 2xn boards with n big enough.

Yeah, I would think the last player would just be forced to take everything. But maybe a player could be interested in taking boxes to change parity.

For example if a player is faced with a half heated handout might chose to take the box (incuring one penalty point) to prevent the double cross (which I guess changes parity).

It actually seems like it could be really interesting but like dots and boxes it might have a steep learning curve.

It would be fun if we could gather a couple of at least semi-expirienced dots and boxes players to explore it.

Carroll, I will invite you to a game. Thanks for the paper.

Sounds interesting, I'd be up for that.

Great. More people?

I made the tournament, sign up if you are game :)

http://www.littlegolem.net/Ctrn.action?view=&dto.id=752

On the description of the UT, you say “the champion is the one who finishes last”, usually for misère games it is “last move loses”.
Here it should be the player with less completed squares wins, even if I agree that there are chances for the last moves to complete most squares.

I think the description means that the tournament winner is the one who finishes last in the tournament.

yeah that was what I meant.

Btw, the 1xN game seems very easy to win for p1 just by making lots of vertical lines and never allow for any double crosses (sacraficing a box where necessary).

Yes the Swedish version where all the edges are drawn is much more interesting, for example can you figure out how 1st player can win by 1 point in the 1x17 Swedish game?

I'm pretty good at getting fewer boxes than my opponent, so I should do well here :o)

I have heard that one many times Martyn, when I challenge people at misere Othello, but they often have to eat their words. :-D

Well if I do have to eat my words, at least I'll be doing it from the top of the tournament table. Don't mind either way to be honest :o)

Happy you are participating martin :)

And just to be clear, noone is forced to take boxes if 3 sides are completed.

Thanks for clarifying, Christian. That's what I'd have guessed, but as you can probably tell I've never played this before

I think none of us have :)

Appearently there is some confuson, as to weather you should resign or play the games out. I am for playing the games out, but I agree that it is sort of weird that you can “win” by resigning all your games.

Does anyone have some thoughts? It is just important that we follow the same standard in the tournament so we can see who won and who lost.

To resign makes more sense in my opinion. It feels kind of wrong to call the one loser who gets less boxes, since that is the actual aim.

If you review old games, nothing will make it appear as a misere game, so you could have a hard time to understand how it was played if you get to the end and it seems you won this game… which was a lost misere game.

So I think resigning makes more sense.
But then the winner of the UT will be the one at the top.

Ok, it is decided. We will want people losing (getting more boxes) to resign before the game is over. This means that the champion will be the one on top.

If someone doesn't want to do there, there is of course nothing I can do about it (nor do they gain anything from it except making it more confusing to see who won the tournament).

So please remind your opponent to resign in the end of the game.

I will update the tournament description to reflect the change.

Yes, this is sane decision. If there are many players and second round is needed, it would be a problem to have the winners at the bottom.

I just hope the players will respect it, but I guess the old system had the same problem.

It would be fun if the site would implement it as a variant. It could even have its own championship :)

Link to the Misere Dots and Boxes tournament.

10 players have registered so far.

Starting today?

I am interested by anyone thoughts about the strategy…

I think it is a game simpler than normal D&B, but I will not elaborate why just now…

My only strategy is not to get these games mixed up with my current D&B games.

Looking at the ratings of all you guys I have a simple stategy: make myself and my opponent believe it's a normal game! :-)

Dryad is already off to a flying start, I think I need a better strategy!

I have no idea whatsoever what I'm doing…

So you have the right strategy as it is exactly what Dryad told me!

Eh… I didn't tell you anything new o.o

Can we have a post-mortem of what we learn from the games in the tournament?

If it's a genuinely interesting variant, perhaps we could petition Richard to introduce it as an official variant?

Misere Reversi is also interesting, but I am not sure about the other games here. Perhaps DVONN would be funny.

It would be a lot of fun if richard would implement it.

The strategy i have found is that it depends on the number of double crosses. Player 1 wants an equal number of double crosses and player two and odd one. So the battle is how much to sacrafice to win the chain war, pretty similar to the normal one.

I don't like it, but maybe that's because I suck ;)

I knew it was going to be a parity battle of double crosses, it follows almost directly from the proof of the chain rule I once studied. I have a hard time evaluating regions though. I never studied nim theory, but I'm not sure if that would've helped me here.

Does anyone think a 4x4 board would've been more appropriate for this tournament. I feel like there's a lot of “meaningless” moves at the start. I know, I know, we all thought the same about regular DnB when we started :p

No, I think 4x4 is too small. What makes it interesting in 5x5 is that when it comes to sacrifices to win the parity battle, the interior 9 squares are quite heavy compared with the about two thirds of the 16 border squares.

So to answer the OP, we can have quite close scores in this version.

Hehe I am pretty bad at this. Were all games lost by the player who took the final boxes? This seems similar to new players playing the normal game where the winner of the chain battle always wins.

Probably, but Carroll sacrificed 12 boxes in one of our games. ;-)

In my game against Aldiris, it turn out that my foreseen 12 boxes sacrifice was actually 13… so I tried something else he spotted easily.

I think it is better not to sacrifice too many boxes at in the same move to take advantage of each free move to decrease the boxes left around the border.

It doesn't seem as deep as normal rules. The LCB collapses onto something is even simpler than the SCB in normal rules (0:X or X:0, compared to -2/-1/=/+1/2), and there's nothing akin to the LCD in normal rules. So I concur with Christian that it's mostly a matter of counting the cost of sacrificing enough. That in turn is still interesting, but it is the feature that makes 6x6 normal different from 5x5 normal. (And most people I know consider 6x6 less elegant as a game.)

I kinda agree with aldiris - the initial 20+ moves do appear to be mostly random, pretty much anything can happen in almost any game after that number of moves.

I've won my first game, but clearly my opponent understands the game better than me as he resigned before I could prove to myself that I had a win. (There's still some sacrificeable not-quite-chains.)

@FatPhil: I don't understand a word of what you guys are discussing, it's all far over my head. In spite of that, prooving your position was won looks comparatively easiy to me. Makes me wonder if in fact I have even understood less than I thought! :-)

Richy, you're winning one of our games, if I'm not mistaken ;)

I agree now that 4x4 would be too small and 5x5 or even 6x6 would be better. Mainly because of a better ratio of edge boxes vs middle boxes. The middle boxes generally seem to be sacrificed (i.e. taken) to force your opponent to take the boxes on the edge.

After what i’ve played now, I feel that the game is more similar to the original that I initially would have thought.

I've learnt a lot - let's do this again!

This one was as close as it gets

http://www.littlegolem.net/jsp/game/game.jsp?gid=1554186

It was an interesting tournament ! Congrats to Dryad !

I have been puzzled by one of my games where I won despite having wrong parity, I hope that someone could clear it up for me.

In game 1554222, I have the impression it is the equivalent of misere nim *2+*2 which loses for next player, but:

A) it is not an impartial game, but a partisan game (color matters)
B) maybe equivalently, in other situation of this type I had the impression the player with right parity had the right number of waiting moves.

I have the right parity in the end, but it seems around move 25, second player has enough room to make another chain. This room was then converted to a shape with 0 or 2 chains.

quads are confusing…

Are they just always counted for two “chains”?

In misere dots, anything is a chain (long chains, double crosses, triple crosses, even quadruple crosses with right shape you cannot cut in two double crosses: T4, quintuple crosses: X) except quads that just don't count… In the end, the player without control will play in a quad and the controlling player will just turn that into two double crosses.

I'm prepared to bet that in misere dots, it is a first player win.
And as most misere games, far enough for the ending position behave like normal games, that could shed some light on the outcome of normal dots and boxes.

So the player in control would want many quads but not long chains since they make it expensive to stay in control.

Not really, because:
- for quads they can cost the player in control at least 3 squares if other player attacks into them before they are complete
- for long chains, you can usually trade at least the four last squares (3+1) against the start of the chain. So it costs you squares only if it is longer than eight squares.

The real problem for the player in control is to decide how many chains he must get before starting to sacrifice so that it will not cost him more than 12 squares to keep the right parity. And he must also be assured that the rest of the board will not turn into another chain. Or be sure he will get another one for a disguised control with wrong parity.

That is true for the 5x5 but in general long chains seem to favour the player not in control.

It is a fun thing about making a difference if quads are complete or not.

http://www.trmph.com/dnb/board#5,b9d9f9g10j11h9a8a6b5d5e6e8f5h5i6i8k10k8j5c4c2b1d1f1g2g4h1k4

First player to move and win (non-misere).

If it's non-misere, why do you post it in a thread on misere dots?

If you haven't solved the puzzle yet, why do you start bitching at me? I was referring to the last post by Christian K.

Oh OK :-P

The tournament is over. Well done dryad.

I'd play d11, personally. But I agree with Taz, I wouldn't play it in a misere thread.

WP Dryad.

AGAIN!

http://www.littlegolem.net/jsp/forum/topic2.jsp?forum=110&topic=142

Don't really get your point, but here you are. :P

How does one defeat the “rotational symmetry” strategy when playing misère dots and boxes? I can see how to do it for the case where the board has odd side lengths (one just sacrifices the middle box), but I don't see how to escape from it otherwise.

I think one answer is that when 1st player will do a double-cross threat (that is a double-cross minus the closing move), 2nd player will not want to play the symmetric move because then 1st player can chose the parity.

See for example :
http://www.trmph.com/dnb/board#5,f11f1e10g2e8g4f7f5d7h5d9h3b9j3i2c10k2a10h7d5i8c4i10c2h11d1b5j7b3j9b1j11b7
here p2 should not play 34.j5 because of :
http://www.trmph.com/dnb/board#5,f11f1e10g2e8g4f7f5d7h5d9h3b9j3i2c10k2a10h7d5i8c4i10c2h11d1b5j7b3j9b1j11b7j5c8a8k4

but should play http://www.trmph.com/dnb/board#5,f11f1e10g2e8g4f7f5d7h5d9h3b9j3i2c10k2a10h7d5i8c4i10c2h11d1b5j7b3j9b1j11b7c8a8i4j5 instead.

Here it is easy because it is the end of the game, but 1st player can force p2 to take a difficult decision beforehand.

An example on 4x4:
http://www.trmph.com/dnb/board#4,d9f1e8e2d7f3b7h3f7d3g8c2h7i8b3a2e4e6c6g4c4g6b9h1c8i2b5h5

Ah, yes. Thanks!