Author |
Topic: Bot Bashing Records Discussion (Read 10183 times) |
|
Arimabuff
Forum Guru
Arimaa player #2764
Gender: 
Posts: 589
|
 |
Re: Bot Bashing Records Discussion
« Reply #45 on: Apr 8th, 2008, 11:13pm » |
 Quote  Modify
|
on Apr 8th, 2008, 6:33pm, Fritzlein wrote:| I agree with mistre's game comment that the record for largest handicap should go to whoever does it first, not to whoever does it in the fewest moves. For material handicap the question is whether it can be done at all, and the proof of concept is more important than refinement. Just my two cents; other opinions may vary. |
|
I can live with that. I played quite a few games to reduce my moves to a minimum so far but I decided to stop that and move on to the next handicap.
That means that Arimaascore1 is a case closed since EMHHDDCRRRR and EMHHDDCC are both impossible. The former leaves too much ground uncovered and the latter lacks the possibility to kill rabbits which is also vital.
|
| « Last Edit: Apr 8th, 2008, 11:31pm by Arimabuff » |
 IP Logged |
|
|
|
Arimabuff
Forum Guru
Arimaa player #2764
Gender:
Posts: 589
|
|
Re: Bot Bashing Records Discussion
« Reply #46 on: Apr 9th, 2008, 4:53am » |
Quote Modify
|
I was able to beat Arimaascore2 with a EMHD handicap, three times (the last two times in a row!); but apparently as soon as I up the ante even a little bit that bot becomes a raging (as well as raving) lunatic! And won't let me catch a break. What is it about this bot's code that says "If your opponent has lost more than EMHD then pounce on him!!!" ?
Anyway, you're welcome to try.
|
|
IP Logged |
|
|
|
mistre
Forum Guru
Gender:
Posts: 553
|
|
Re: Bot Bashing Records Discussion
« Reply #47 on: Apr 9th, 2008, 8:36am » |
Quote Modify
|
Just to clarify my earlier comments - If you match your OWN handicap with fewer moves, you are welcome to have the later game linked to on the bot bashers page. However, if you match another player's handicap with fewer moves - it does not beat his record.
|
|
IP Logged |
|
|
|
Arimabuff
Forum Guru
Arimaa player #2764
Gender:
Posts: 589
|
|
Re: Bot Bashing Records Discussion
« Reply #48 on: Apr 9th, 2008, 8:57am » |
Quote Modify
|
on Apr 9th, 2008, 8:36am, mistre wrote:Just to clarify my earlier comments - If you match your OWN handicap with fewer moves, you are welcome to have the later game linked to on the bot bashers page. However, if you match another player's handicap with fewer moves - it does not beat his record.
|
|
That's what I got from your comment, but "matching your own handicap with fewer moves" is also a way of showing that the first one wasn't just a fluke, as sometimes happens in this business. It's even more of a confirmation when you renew the exploit the very next game.
|
|
IP Logged |
|
|
|
Arimabuff
Forum Guru
Arimaa player #2764
Gender:
Posts: 589
|
|
Re: Bot Bashing Records Discussion
« Reply #49 on: Apr 9th, 2008, 2:56pm » |
Quote Modify
|
The more I think about it the more it becomes obvious to me that 99of9's game with shallowblue should be taken off the handicap list.
First, the bot kills off four of his pieces that he didn't have to sacrifice himself (the camel plus three rabbits).
Had he killed them himself as handicap procedure normally requires, it would have taken him with an average of three steps by piece THREE MOVES to kill them all.
In addition, the bot spends FOUR MOVES to kill the pieces that he didn't use to maneuver his OWN pieces toward a goal.
Therefore, when you think RATIONALLY about it, 99of9 got an unjustifiable bonus OF SEVEN MOVES in comparison to someone who would have killed his pieces himself and we can all see that in spite of that SEVEN MOVE delay he can barely make it to goal before the bot does.
YOU people refused my handicap game because of a lousy four steps that I slipped in there; so what do you say about a cheat of SEVEN MOVES Vis a Vis the NORMAL procedure of our newfound scrupulousness?
I say we remove that game from the handicap list otherwise all our previous conversation about the legitimacy of my game is nothing but a TRAVESTY.
|
|
IP Logged |
|
|
|
mistre
Forum Guru
Gender:
Posts: 553
|
|
Re: Bot Bashing Records Discussion
« Reply #50 on: Apr 9th, 2008, 4:38pm » |
Quote Modify
|
As great an achievement that 99of9's games against Shallowblue are, I am going to agree with Arimabuff.
At the very least, the pieces that Shallowblue takes should not count for the handicap. Also, in one of the games, 99of9 only takes 3 steps instead of 4 twice during the sacrificing moves.
|
|
IP Logged |
|
|
|
Arimabuff
Forum Guru
Arimaa player #2764
Gender:
Posts: 589
|
|
Re: Bot Bashing Records Discussion
« Reply #51 on: Apr 9th, 2008, 9:16pm » |
Quote Modify
|
on Apr 9th, 2008, 4:38pm, mistre wrote:| ...At the very least, the pieces that Shallowblue takes should not count for the handicap... |
|
I think it goes farther than that. In the game 9391 at move 4b, 99of9 uses the dog as a tactical device, making it an active piece of the board. Indeed the dog is pushed toward a trap that isn't available in order to prevent the elephant from pushing the camel there and forcing it to use the OTHER trap. I say that that's a breach of procedure right there. Therefore, it also ends the legitimate sacrificing sequence.
However, I don't think that 99of9 would appreciate to have his game listed as a HDCCR handicap. So, I think it'd be preferable to simply remove it and put it under a new category, I'd propose the category "believe it or not". It would be composed of all the games that are both exceptional and non classable in the other categories.
There are quite a few of these games, for instance a game where the player (yours truly actually  ) manages to kill ALL of Bomb's pieces except for the elephant with only HIS ELEPHANT passing the middle board. And only HIS ELEPHANT doing the killing and WITHOUT himself losing any pieces. In a way it’s the reverse of IDAHO’s Botbash where the elephant was forbidden to pass the middle line.
I believe Karl has this one on record but I can retrieve it if need be.
Anyway, my point is that these two games don't belong in the handicap category and that they may give newcomers a false sense of carelessness about our definition of a handicap game.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: Bot Bashing Records Discussion
« Reply #52 on: Apr 9th, 2008, 10:00pm » |
Quote Modify
|
I support the inclusion of a section of amazing exploits that don't fall into the main botbashing categories.
|
|
IP Logged |
|
|
|
Arimabuff
Forum Guru
Arimaa player #2764
Gender:
Posts: 589
|
|
Re: Bot Bashing Records Discussion
« Reply #53 on: Apr 10th, 2008, 3:07am » |
Quote Modify
|
on Apr 9th, 2008, 10:00pm, Fritzlein wrote:| I support the inclusion of a section of amazing exploits that don't fall into the main botbashing categories. |
|
That's a far better title than the one I proposed.
Let's set some ground rules:
To enter there a game must either represent some unique exploit never to be seen in another game OR be the very first of a select few that do a similar thing.
Now of course the word exploit is open to interpretation.
I for one don't believe that using a bot's inability to play after you lost your last rabbit or a bot's randomness of moves after he is faced with an unavoidable goal to be an exploit.
I think we can all agree on that.
If nobody answers negatively to this post for the next 4 days I will consider that it is therefore unanimously adopted and make the changes myself if nobody else does.
|
|
IP Logged |
|
|
|
99of9
Forum Guru
Gnobby's creator (player #314)
 
Gender:
Posts: 1413
|
|
Re: Announce Bot Bashing Record - New!
« Reply #54 on: Apr 10th, 2008, 5:40am » |
Quote Modify
|
The overarching reason for my principle is not to suicide as fast as possible, but simply to make sure I have not "played in advance" before the "start" of the handicap game, all I have done was playing toward suicide. In my opinion what the bot has done in that time is not relevant to whether I beat it with the handicap. (It is also nontrivial to regulate what the bot is allowed to do in that time, because the player does not have control over what it is doing - remember we may be able to mostly predict shallowblue, but whatever rule we use should be able to apply to even human handicap games.)
on Apr 7th, 2008, 4:46pm, 99of9 wrote:"All the first steps should be in the direction of a trap to suicide in (or to allow another piece to do so), before any other moves are made." Seems like a good general principle to me.
I'd be interested to hear other definitions of principles. |
|
Both shallowblue games conform to this definition. If nobody else likes this definition (which I have tried to go by for a long time), and wants to make it stricter, please can you at least state and agree on the definition we will be going by.
By the way, if the "instant handicap" becomes technically possible, achieving such handicaps will be easier than in any of these games, because the bot will get 0 moves before starting.
Ultimately I don't really care whether or where these games are listed, but I don't agree with moving them until someone can state an elegant and simple principle that people agree on.
|
|
IP Logged |
|
|
|
99of9
Forum Guru
Gnobby's creator (player #314)
Gender:
Posts: 1413
|
|
Re: Bot Bashing Records Discussion
« Reply #55 on: Apr 10th, 2008, 7:40am » |
Quote Modify
|
on Apr 8th, 2008, 11:13pm, Arimabuff wrote:| That means that Arimaascore1 is a case closed since EMHHDDCRRRR and EMHHDDCC are both impossible. The former leaves too much ground uncovered and the latter lacks the possibility to kill rabbits which is also vital. |
|
EMHHDDCRRRR against ArimaaScoreP1 with gold in game 73905.
|
|
IP Logged |
|
|
|
Arimabuff
Forum Guru
Arimaa player #2764
Gender:
Posts: 589
|
|
Re: Announce Bot Bashing Record - New!
« Reply #57 on: Apr 10th, 2008, 9:06am » |
Quote Modify
|
on Apr 10th, 2008, 5:40am, 99of9 wrote:The overarching reason for my principle is not to suicide as fast as possible, but simply to make sure I have not "played in advance" before the "start" of the handicap game, all I have done was playing toward suicide. In my opinion what the bot has done in that time is not relevant to whether I beat it with the handicap. (It is also nontrivial to regulate what the bot is allowed to do in that time, because the player does not have control over what it is doing - remember we may be able to mostly predict shallowblue, but whatever rule we use should be able to apply to even human handicap games.)
|
|
First of all what the bot does during your sacrificing period can't be taken for your own benefit. The sacrifices must be made by you NOT THE BOT, if the bot kills off one of your pieces he doesn't do that to do you a service. Your version of the fact would imply that since IN YOUR MIND you are sacrificing your pieces then anything the bot does for ITS OWN agenda is part of that sacrificing. That's not how it works.
Imagine that instead of a bot, you'd have a human player; do you think that he would have agreed that anything he'll kill during YOUR SACRIFICING PERIOD would be counted AGAINST HIM? And if you got him to agree to that rather crooked rule don’t you think that he would spend this time NOT KILLING any of your pieces and maneuvering instead? The rules of Arimaa do not change because of what's IN YOUR MIND at the time you play. You have to prove MATERIALLY that you are giving up your pieces, not simply say "hey, since I am sacrificing my pieces, now the bot is working for me!!!". That's ludicrous.
Same thing in Chess if you voluntarily sacrifice a piece during the game by letting your adversary take it, that never has and never will be taken as a handicap.
Following that, the three rabbits AND the camel are definitely off the table.
Quote:
Both shallowblue games conform to this definition. If nobody else likes this definition (which I have tried to go by for a long time), and wants to make it stricter, please can you at least state and agree on the definition we will be going by.
By the way, if the "instant handicap" becomes technically possible, achieving such handicaps will be easier than in any of these games, because the bot will get 0 moves before starting.
|
|
If "instant handicap" is made possible then the games we will play then will not be taken as a competition with the games played now, we will have to create a new category called "instant handicap" to cover these NEW games. Because difficulty is part of the game and that you don't cheat your way in a hall of fame; to replace a game by one of yours, yours must ACTUALLY be harder to achieve otherwise that would be making a mockery of the whole thing.
So we will have two categories then "Old fashioned handicap” and "Instant handicap". It would be absurd to act otherwise. As I said, there must be an ACTUAL step up in difficulty for your game to replace the game that was before.
Quote:
Ultimately I don't really care whether or where these games are listed, but I don't agree with moving them until someone can state an elegant and simple principle that people agree on. |
|
The principle is that since the bot IS NOT at your service but is supposed to work against you, to claim that a piece is part of a handicap you have to actually move it to the trap YOURSELF not let the bot help you. If it can't be done then it simply means that this particular handicap is IMPOSSIBLE with this particular bot. I don't see why we should have a problem with that.
In addition, a piece you intend to sacrifice for handicap is NOT YOURS to play with BY DEFINITION and you can't use it for obvious tactical ploy like you did with that dog to make the bot work harder to do HIS JOB of killing off your pieces.
Just the idea that YOU would claim a bot’s work as yours strikes me as so funny that I couldn't repress a chuckle while writing the preceding sentence.
I think we can tolerate a piece to be sacrificed blocking a rabbit by it's sole presence but it should be obvious that it isn't used ACTIVELY in a tactical ploy or it can no longer be claimed as part of the sacrificing sequence which we all agreed can't be sliced up in two or more... and must precede any effective tactical move that you have in mind.
Again, a bot is working FOR ITSELF, and is not at your service. Anything he takes during the sacrificing sequence is not added to your benefit. If you can't protect your OWN pieces during that period then as the say goes "Those are the breaks!"
That seems only fair to me.
You wanted a clear definition, I gave you one, and I believe that I am not the only one to think it reasonable.
Unless you find someone here to adhere to your strained principles that seem to be invented solely to justify your shallowblue games I will proceed to the changes I said.
I will wait however four days to do so as I said before.
|
|
IP Logged |
|
|
|
Arimabuff
Forum Guru
Arimaa player #2764
Gender:
Posts: 589
|
|
Re: Bot Bashing Records Discussion
« Reply #58 on: Apr 10th, 2008, 9:11am » |
Quote Modify
|
on Apr 10th, 2008, 8:40am, mistre wrote:
...It would be easy to add a sub-category for Amazing Exploits and list 99of9's games against Shallow Blue among others... |
|
Good idea!
|
|
IP Logged |
|
|
|
Janzert
Forum Guru
Arimaa player #247
Gender:
Posts: 1016
|
|
Re: Announce Bot Bashing Record - New!
« Reply #59 on: Apr 10th, 2008, 9:38am » |
Quote Modify
|
Here is what I personally would have used for the criteria to count for sacrifice games.
The player has to sacrifice the maximum number of pieces possible for every move at the beginning of the game (e.g in the second move the player has to sacrifice 3 pieces, 3rd move 2, 4th move 2, 5th move 1, 6th move 2, etc.). Once a move is made in which the maximum number of pieces is not sacrificed the player's pieces that have been removed from the board at that point are counted as the handicap.
I can see where the opponent could block the player from sacrificing pieces to meet the above criteria. But I think it would probably become a problem to sort out all possible situations and where to draw the line if it was expanded to try and cover those situations.
I haven't actually looked at many of the games discussed so have no idea how many would count or not by that criteria.
Janzert
|
|
IP Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Reply
Notify of replies
Send Topic
Print