Live data from Hacker News

Correcting for the first-player advantage in Risk

stat.columbia.edu

11–20 of 23 posts

Re: Correcting for the first-player advantage in Risk

#11
post #10

That is remarkably content-free. Where's the maths?

Or really any real content. The author didn't even try out his own example to see if it would work. And honestly how horrible the old Risk rules is really old. There are both new rules and much more enjoyable games these days.

Re: Correcting for the first-player advantage in Risk

#12
post #8

Remember when you were a new coder, and sometimes you'd see an off-by-one error that for the life of you, you couldn't track down? And after a long enough time searching, you just gave up and added or subtracted one from the variable? I don't mean to badmouth these suggestions, because I have none that are better, but they all feel like that to me. Voo-doo fixes. Just a thought.

And after a long enough time searching, you just gave up and added or subtracted one from the variable?

Don't pull that trick in my shop. Isolate your bug and fix it. Whatever is causing it is probably causing other bugs that will affect other people, including the poor sucker who has to clean up your mess.

Re: Correcting for the first-player advantage in Risk

#13
post #12
post #8

Remember when you were a new coder, and sometimes you'd see an off-by-one error that for the life of you, you couldn't track down? And after a long enough time searching, you just gave up and added or subtracted one from the variable? I don't mean to badmouth these suggestions, because I have none that are better, but they all feel like that to me. Voo-doo fixes. Just a thought.

And after a long enough time searching, you just gave up and added or subtracted one from the variable? Don't pull that trick in my shop. Isolate your bug and fix it. Whatever is causing it is probably causing other bugs that will affect other people, including the poor sucker who has to clean up your mess.

He refers to the known floating point rounding error. You should not expect a 'new coder', as he put it, to trully understand floating point rounding errors because to understand it, you need to already be a coder with some experience.

Re: Correcting for the first-player advantage in Risk

#14
post #13
post #12

Earlier quoted context omitted.

And after a long enough time searching, you just gave up and added or subtracted one from the variable? Don't pull that trick in my shop. Isolate your bug and fix it. Whatever is causing it is probably causing other bugs that will affect other people, including the poor sucker who has to clean up your mess.

He refers to the known floating point rounding error. You should not expect a 'new coder', as he put it, to trully understand floating point rounding errors because to understand it, you need to already be a coder with some experience.

"You should not expect a 'new coder', as he put it, to trully understand..."

Oh, but I do. If you don't understand it, figure it out. If you still don't understand it, ask. If you still don't understand it, sound the alarm, and we'll both sit down and figure it out together. And you will learn it.

But please don't use "newness" as an excuse to leave shit behind. Excellence comes from doing excellently, whatever it takes.

Re: Correcting for the first-player advantage in Risk

#15
post #7

Not having played Risk in ages, I have a few questions: * Is there a statistically significant first-mover advantage? * How big is it? * What is its mechanism?

You might be able to find the answers this over on board game geek http://www.boardgamegeek.com/boardgame/181

Re: Correcting for the first-player advantage in Risk

#17
post #12
post #8

Remember when you were a new coder, and sometimes you'd see an off-by-one error that for the life of you, you couldn't track down? And after a long enough time searching, you just gave up and added or subtracted one from the variable? I don't mean to badmouth these suggestions, because I have none that are better, but they all feel like that to me. Voo-doo fixes. Just a thought.

And after a long enough time searching, you just gave up and added or subtracted one from the variable? Don't pull that trick in my shop. Isolate your bug and fix it. Whatever is causing it is probably causing other bugs that will affect other people, including the poor sucker who has to clean up your mess.

Sorry--I should have been more specific. When I said "new coder," I really meant "new coder." I started programming when I was 9- or 10 years old, and thus worked without the onus of keeping my code clean for its future maintainers. As a victim of my own folly, however, I learned to avoid voodoo fixes by the time I was 14. :-)

Re: Correcting for the first-player advantage in Risk

#19
post #7

Not having played Risk in ages, I have a few questions: * Is there a statistically significant first-mover advantage? * How big is it? * What is its mechanism?

Well you should check out the stats of the national risk league or any of the multiple professional risk league sin Europe.

Re: Correcting for the first-player advantage in Risk

#20
I played risk online at warfish.net for a while until I felt like I had maxed out the strategy to within the noise level. It isn’t obvious to me that there is a first mover advantage. In our games the initial territories were randomly distributed, and I always felt that whoever got the “most closely clustered” territories had a tremendous advantage.

The best advice I’ve ever heard for competitive Risk is “play the other players, not the continents”. You want to try to create situations such that you are the least desirable player to attack. In general, this means turtling. Occupy a few territories that are close to each other with about 5 armies on each, and put as many armies as you can on one territory that is adjacent to enemies. Don’t take a whole continent because people will try to crack it so you don’t get the bonus armies. Try to get your armies off of continents that it looks like people will fight for. Try to get a card every turn, and hold them as long as you can.

You stop turtling when and only when you can eliminate a player in one move and get a bunch of cards from them. You will usually do this by trading in a set of cards, putting all of the armies on your big territory, and marching it all over the place. You will almost always succeed if you have about 0.75n + 1.0t armies to push with, where n is the number of armies you will need to kill, and t is the number of new territories you will need to occupy.

(Corollary, if you’re weak then try to use all your cards so that you’re not a valuable target.)

Eliminating a player who has 5 cards almost assures that you will win the game.

After a player or two has been eliminated the game will transition to one where the players try to hold continents to collect bonus armies. If the players are fairly competent then the winner will mostly be decided by how lucky people are with collecting cards. The yield from trading in a set of cards starts to dwarf the armies collected from holding territories. Almost as important, the new armies can be place freely.

I have never seen Asia successfully held for even a single turn.

(Regarding the first alleged first-mover advantage: obviously it can’t be disadvantageous to move first. If it were, then the first player would take no action so as to avoid the disadvantage. By induction, if it was disadvantageous to go first then the game would never advance past the initial state.)

Post reply on HN