Arimaa Forum - Print Page


    
      
        Arimaa Forum
        (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi)
      

        Arimaa >> General Discussion >> Using tools for improving a match game
        
(Message started by: clojure on Oct 6^th, 2010, 7:15am)

Title: Using tools for improving a match game
Post by clojure on Oct 6^th, 2010, 7:15am

I have got the impression that Arimaa community considers trying out moves before playing them out a neutral, or even recommended thing, since there is the expert mode, a.k.a. planning window.

This felt quite weird when I got to know Arimaa. I have a feeling that in go and chess the calculate-moves-in-one's-head is crucial part of the play, and it would be considered rude if it was found out that opponent used another viewer for trying out moves in an online game.

If the above is correct, it puts forward a question where is the borderline for permitted external help. Is the player allowed to use bots in helping his performance? If so, the human vs. bot challenge gets a bit problematic nature in my opinion, since the human side can use the opponent against itself.

Does all this have any relevance at all?

Personally I don't use the planning window since playing without it is more exciting and helps improving reading skills.

Title: Re: Using tools for improving a match game
Post by rbarreira on Oct 6^th, 2010, 7:53am

I'm not sure that people actually use the planning window in non-postal games. Just the time to start it up and move pieces around might be better spent thinking instead. Maybe in some very specific situations?

As for using bots and how much aid is accepted, the challenge rules state this:

Quote:

The human player may use an aid such as a physical game board or a computer program so that pieces can be moved to assist in planning a move. No aid may be used which suggests a move or provides any information about the move aside from the position score defined in the Arimaa match rules.

Title: Re: Using tools for improving a match game
Post by Nombril on Oct 6^th, 2010, 11:56am

I agree with rbarreira: I use the planning tool for postal games, and when studying the mistakes I've made. I am sometimes tempted to use expert mode to try out a line during a 'live' game, but so far I just visualize it in my head.

It seems the number of pieces that can move leads to a lot more 'details' that need to be remembered about a position in only a few turns as compared to Chess and Go.

Title: Re: Using tools for improving a match game
Post by novacat on Oct 6^th, 2010, 5:10pm

I sometimes use expert mode in interactive games when I consider unclear or experimental moves. I tend to think slowly, so it mostly helps me see obvious errors quickly. I also tend to run low on time if I use it too much.

Title: Re: Using tools for improving a match game
Post by Fritzlein on Oct 6^th, 2010, 7:28pm

In games faster than 2 minutes per move, the planning window actually hurts my play. Games that are exactly two minutes per move are borderline for whether it helps or hurts me. Only in postal games am I positive that the plan window helps me.

Of course getting move suggestions or evaluations from bots is against the rules. There is a small gray area in that being told what moves are legal and being shown the position could be considered "machine assistance", but to me that seems very different.

Title: Re: Using tools for improving a match game
Post by clojure on Oct 6^th, 2010, 7:59pm

on 10/06/10 at 19:28:12, Fritzlein wrote:

There is a small gray area in that being told what moves are legal and being shown the position could be considered "machine assistance", but to me that seems very different.

I have an example that relates to shogi. There is a famous rule that allows one to put a captured piece back to board. But there's exception: it's illegal to put two pawns on same column. If one does that, it's a lost game.

A japanese amateur dan that is having a youtube channel on shogi seems to consider that this is an important part of shogi culture and thinks that in even in online games one should be able to make this illegal move.

There is also other losing moves, for example the western chess concept of infinite check is not advantegous for the checker. The idea being that to accept a draw one must have a opportunity not to accept it.

Title: Re: Using tools for improving a match game
Post by Fritzlein on Oct 8^th, 2010, 12:04am

on 10/06/10 at 19:59:12, clojure wrote:

That is a great example of how some people love tradition for the sake of tradition. What does forfeiting in case of an illegal move have to do with the greatness of shogi, or of any game? Strategy games should be about who is best at strategy, not about who is best at following rules.

In chess tournaments there is some reasonable penalty for making an illegal move, such as five minutes added to the opponent's clock. This is enough to prevent anyone from getting an advantage due to an intentional illegal move without turning chess into a contest in rule-following. For online games, players can be prevented from getting an advantage due to an illegal move simply by having the interface disallow illegal moves. Anyone who tries to make an illegal move simply loses the time it took them to try it. Why would you want anything more?

Title: Re: Using tools for improving a match game
Post by clojure on Oct 8^th, 2010, 1:24am

on 10/08/10 at 00:04:58, Fritzlein wrote:

It's hard to argue for, and I do find it odd. Still, I somehow can also understand how he feels. I cannot describe properly how I feel about this but the combination of the history of shogi, the psychological aspect, Japanese culture, etc might fit together. It can be a sthingy ful of salt that brings forth emotional aspects.

Rules like these don't really matter on high level since everybody has played so many times, and subconsciously avoid it easily. Though, there still occurs rare occasional losses due to this, there's even a video'ed pro game of a double pawn dropping loss.

It may relate to other traditional aspects of the game. There's lots of small details how you should put your stones onto the board etc. It makes the playing feel special and help the brain adjust to "shogi atmosphere".

Sure, considered purely as intellectual game, the rule doesn't make sense. :)

Title: Re: Using tools for improving a match game
Post by omar on Oct 8^th, 2010, 10:09am

I think in a human vs human game the most important thing is that both players have the same tools available to them whether it be a client that knows nothing about the game rules and acts like a physical board or a client that goes to the extent of suggesting moves or playing as an opponent before submitting the move. Beyond that's it hard to say how much we should hang on to tradition and how much we should embrace new technology. I don't think there is any right answer and the choice will probably change over time. Currently we tend to prefer minimum assistance, but maybe in the future people will prefer cyborg style games.

In human vs machine games I think how much technology a human can use to improve performance should be limited, otherwise it would just becomes a machine vs machine game. Using technology to prepare the move, checking if the move is legal and manually manipulating the state of the game to plan a move I think is about the extent of assistance we should allow. Using technology to assist beyond that (such as highlighting threatened pieces or potential blunders) would cause the game to not reflect the natural ability of the human player. I suppose a bot developer could argue that even this level of assistance has enhanced the natural ability of the human player compared to playing with a physical board. I would have to agree, but to keep things fair do we also want go to the extent of requiring the bot to use a camera to look at the board and physically move the pieces and press the clock :-)

Title: Re: Using tools for improving a match game
Post by Nombril on Oct 16^th, 2010, 4:27pm

I was wondering - is using material evaluator formulas (by hand or using the page Janzert has made available online) considered considered an allowable tool for use while playing, or something that is off limits? This is probably only applicable to a postal game.

Title: Re: Using tools for improving a match game
Post by Hirocon on Oct 16^th, 2010, 5:01pm

I frequently use the planning window in interactive games. Of course, I frequently find myself under time pressure, so maybe the planning window isn't actually helping me.

Another thing I'm guilty of is using the material evaluators (http://arimaa.janzert.com/eval.html) to decide if I should accept and/or initiate a trade sequence.

Title: Re: Using tools for improving a match game
Post by Sconibulus on Oct 16^th, 2010, 5:14pm

I've never looked at material evaluators to decide things, but I have used expert mode in the middle of the game. The plan window seems to take more time to load and such than it's worth.

Title: Re: Using tools for improving a match game
Post by rbarreira on Oct 16^th, 2010, 5:52pm

The World Championship rules state:

Quote:

Players are forbidden from using advice or suggestions from others or programs while playing the games.

I'd say a material evaluator definitely counts as using advice from a program, and if this rule is good enough for the world championship it's probably also good enough for other games...

The above might not make much sense for postal games of course, since on those you would have the time to calculate the evaluation yourself anyway. I wouldn't be surprised if some people disagree though, since it's a sort of a slippery slope...

Title: Re: Using tools for improving a match game
Post by chessandgo on Oct 17^th, 2010, 3:53am

I sometimes use the expert mode in interactive games too, and it probably hurts my play too :) I used to use it a lot at some point, but it felt like I missed even more first move answers than usual.

And I strongly agree with this

on 10/08/10 at 10:09:05, omar wrote:

I think in a human vs human game the most important thing is that both players have the same tools available to them

so it's cool that the standard game client does as much as is technically possible and allowed, making sure that everyone plays with the same tools.

Title: Re: Using tools for improving a match game
Post by Nombril on Oct 17^th, 2010, 9:53am

on 10/08/10 at 10:09:05, omar wrote:

In human vs machine games I think how much technology a human can use to improve performance should be limited, otherwise it would just becomes a machine vs machine game. Using technology to prepare the move, checking if the move is legal ...

Does the game engine warn a bot that it has submitted an illegal move, and allow them to put in a new one? With the illegal moves sometimes submitted by bots, I was wondering if they just didn't 'know' how to find and submit a second move, or weren't given that "tool" that humans are given? (Of course, this doesn't help if the clock is running out anyway...)

Title: Re: Using tools for improving a match game
Post by omar on Oct 18^th, 2010, 9:24am

on 10/16/10 at 16:27:22, Nombril wrote:

I would say it is allowed for a non-event game and your opponent has agreed to allow it. For event games like the postal mixer, we would want the game to reflect your natural ability so the rules says: "A player must not use a computer program to suggest or evaluate moves."

Title: Re: Using tools for improving a match game
Post by omar on Oct 18^th, 2010, 9:38am

on 10/17/10 at 09:53:59, Nombril wrote:

I am not sure if AEI warns the bot about an illegal move. The arimaa.com gameserver does not. If a bot submits an illegal move then most likely it does not have all the rules programmed. In some cases the bot may have crashed. Giving the bot a second chance to submit another move would not help unless the bot is programmed to look for this. For the human player it would mean having to wait around until the bot times out to claim the win. So I decided it was better to immediately end the game if a bot submits an illegal move.

Title: Re: Using tools for improving a match game
Post by rbarreira on Oct 18^th, 2010, 10:01am

on 10/18/10 at 09:38:53, omar wrote:

The AEI gameroom script doesn't warn of illegal moves, but if the bot crashes before finding a move it does a clean restart of the bot, giving it a second chance to submit a move (this has actually saved my bot from losing some games when it had a crashing bug).

I agree that giving bots a second chance to submit a legal move would be next to useless. In the time a developer makes the code to exclude a rejected move, the same developer could instead program the missing rules and end up with a better bot (such as the 3rd repetition rule).

Title: Re: Using tools for improving a match game
Post by Nombril on Oct 18^th, 2010, 2:20pm

on 10/18/10 at 09:24:28, omar wrote:

Hmm. I guess this makes sense. I had originally thought 'program' = 'bot'. But I guess there is a 'program' as part of the webpage that is calculating the values.

I must admit that I have used the evaluators between moves (after the pace slowed down, anyway) to see if I was 'ahead' or 'behind' in material. I didn't use them to decide on any moves, but not because I understood it to be against the rules, but because I don't believe the evaluations.

Title: Re: Using tools for improving a match game
Post by Fritzlein on Oct 18^th, 2010, 6:04pm

I didn't know that it would be against the rules to consult the material evaluators in a postal game. My hazy memory is that I may even have included computer evaluations in analysis of the Mob game when I was the One, and had promised not to use computer assistance. Doh!

Title: Re: Using tools for improving a match game
Post by Hippo on Oct 19^th, 2010, 7:38am

I am not using it, but I don't think there is something wrong when computer answers you questions where you don't need to specify position of a piece on the board.

Material evaluators seems to be OK for me.
If someone needs the material evaluation for 3 cases ... it's just matter of time to compute the value using pencil and paper ... making the calculation in excell would be illegal?
... could I write in "paintbrush instead on paper?" ...

I fully vote for all kinds of positional evaluators to be illegal.

Title: Re: Using tools for improving a match game
Post by omar on Oct 21^st, 2010, 1:03pm

Yes, there is a definite distinction between a material analysis and a positional analysis. At one point when the Arimaa scoring function to determine the winner of a game that has run out of time was complex the game client would display the score automatically. The score actually used a bit of positional information as well (rank of the farthest rabbit).

If a player used a material evaluator to make a decision in a game than I don't think that moves reflects the players ability at the time of the game. It would be great if the player went over the game after it was finished and used the material evaluator to learn or submitted the move and then see what the evaluation says. But doing it while the decision is still pending has the potential to effect the decision.

Allowing material evaluations also puts us on a slippery slope of defining exactly what kind of external tools can be used for deciding a move. I think it would be better to just stay clear of that.

Title: Re: Using tools for improving a match game
Post by pago on Oct 21^st, 2010, 1:46pm

Quote:

Yes, there is a definite distinction between a material analysis and a positional analysis.

I think that the distinction between material evaluator and positional evaluator is sometimes not so clear.

For example if you can locally apply a material evaluator in different area of the board and concatene the results to get a global result : is it a positional evaluator and is it still a material evaluator ?

Title: Re: Using tools for improving a match game
Post by Nombril on Oct 21^st, 2010, 1:50pm

Omar, what you are saying does make sense, and it sounds like it is a final decision. Can Janzert add a note under the heading to clarify that this tool should be used by spectators and to analyze completed games, and not by players during a game?

Since I (and a few others) had not realized this, it would help prevent unintentional cheating.

Title: Re: Using tools for improving a match game
Post by omar on Oct 24^th, 2010, 6:36am

OK, I've sent a message to Janzert.

Title: Re: Using tools for improving a match game
Post by Fritzlein on Oct 24^th, 2010, 10:21pm

on 10/21/10 at 13:03:58, omar wrote:

Allowing material evaluations also puts us on a slippery slope of defining exactly what kind of external tools can be used for deciding a move. I think it would be better to just stay clear of that.

Avoiding a slippery slope is not so easy. When I am playing a postal game, I must obviously be allowed to calculate FAME values in my head, because that is purely my own mental power at work, but would it be cheating to use pencil and paper to calculate FAME values? That could be construed as a mechanical aid for my insufficient memory. If pencil and paper is OK, how about an abacus? If an abacus is OK, how about a calculator?

Or maybe you want to ban using a pencil and paper as calculating aids during postal games so that we don't get on a slippery slope. In that case, how can we justify allowing a board to be set up, and variations to be played out on that board? Isn't the board itself a mechanical aid to memory and visualization? Why can I use a machine to help me with the mental function of visualization but not the mental function of arithmetic? Should we insist, for the sake of purity, that the only thing you can do when you are playing a game is close your eyes and think?

And by the way, for making sure your thinking is pure during the course of a postal game, you definitely must not read my book or volunteer to edit chessandgo's book. The ideas therein might help you out. Also if there is an elephant-horse attack going on in your postal game and some people are discussing elephant-horse attacks in the chat room, you had better log out of the chat room and not read the archive. Furthermore, as long as your postal game lasts, you had better not listen to any live commentary on AWL games, in case the games are similar to game you are playing and the commentary gives you some ideas. By the same token, the Arimaa Festival's think-out-loud match, strategy workshop, and team games are definitely off limits for anyone who is currently involved in a postal game.

I'm joking about all the restrictions. Furthermore, I am not advocating the opposite course that we remove all restrictions and allow all forms of assistance, mechanical or otherwise. My point is that the existence of a gray area shouldn't force us to extremes to avoid ambiguity. I am in favor of letting people use Janzert's material evaluation page in postal games and even live games if they want. I don't think that puts us on a slippery slope any more than it does to allow a plan window or a board for moving pieces around to test variations.

To me it is still a judgment call who wins from a trade, even after you have seen what the material evaluation formulas say, just like which line to play is still a judgement call after you have seen it displayed in the plan window. Neither activity subtracts significantly from the strategy of Arimaa as far as I am concerned. I would be less comfortable about assistance that informed players about hanging pieces, although there would still be strategy in the face of universal blunder protection. I would be entirely uncomfortable with players being able to see Marwin's top five move suggestions after a one-minute search and their relative evaluations, although chessandgo would probably still beat me even if I had this assistance.

I think we need to keep an open mind to each other's preferences. Omar is so comfortable with being able to move pieces around to help himself think that he wrote it directly into the rules, but to someone else that might seem like cheating or undermining the character of the game. On the flip side, Omar sees formulaic material evaluation as so much like cheating that it needs to be explicitly written out of the rules. I personally don't quite get the distinction. I understand that we have to draw a line somewhere, but I would like to keep the discussion open. It is difficult enough to make clear-cut and obvious rules; it is rather more difficult to justify such rules.

Title: Re: Using tools for improving a match game
Post by rbarreira on Oct 25^th, 2010, 6:00am

on 10/24/10 at 22:21:06, Fritzlein wrote:

I am in favor of letting people use Janzert's material evaluation page in postal games and even live games if they want. I don't think that puts us on a slippery slope any more than it does to allow a plan window or a board for moving pieces around to test variations.

I agree with you regarding postal games (as your post illustrated, it's quite impractical to restrict what people do during postal games). But I don't agree that they should use evaluators during live games.

If someone makes a positional evaluators webpage, should that be allowed during live games? I think most people would say no. If we agree on that, what's the sense of allowing material evaluators, which are the biggest component in a positional evaluator?

I think that for live games, omar has the right restrictions - any external entity (computer or person) that gives you advice on the game you're playing should be off limits.

Title: Re: Using tools for improving a match game
Post by clojure on Oct 25^th, 2010, 7:04am

The most important thing with regarding fairness is that everyone that is participating in the tournament, knows clearly what the rules are.

It would help if all the allowed help is within the client. i.e. if external material evaluator is allowed, then the client should provide one or have a link to one.

About pen&paper, books, computer-aid. I personally think that there is clear distinction between calculating with pen&paper or with computer-aided tool. The former is a passive tool that you actively use, the latter has active behavior independent of you. Thus, also using external board as a external memory would be different from automatic material evaluator since a board is a passive entity.

Now, I don't say that even passive tools should be allowed. Just that there is distinction that might make sense when evaluating whether it should be allowed.

There is another distinction. Namely, books are passive. But they can have intimate knowledge about particular situation. As an example, let's say that chessandgo's book is known only to the other player in a game. Now if the game reaches a position where there is clear pattern, or strategic elements, the other player can look up all the important aspects that should affect his decision in short time. Definitely this is an advantage.

So it's good to have another distinction of external help: namely, does it codify knowledge of Arimaa more than the rules.

Just for making an example, I think books are allowed in some tournament play in go.

So two orthogonal distinctions are important: passive/active and context-sensitive knowledge.

Title: Re: Using tools for improving a match game
Post by Nombril on Oct 25^th, 2010, 8:07am

I guess the distinction I see is who/what is doing the deciding. For the material evaluator page, Janzert has decided which formulas should be considered. So the player is accepting advice from someone else about what formulas to use. (Personally I don't think they are a big help, so that is why I just suggested we include a note with the tool to make sure we are all aware of the ruling by Omar.)

By researching books, previous games, or commentary: (a.) Both players have the information available, and (b.) The player must decide if it is applicable to the game. (I have certainly seen a lot of examples now where exchanging just one piece, or a piece being one square over, makes the difference between winning and loosing.)

With regards to (a:) That was the only hesitation I had in helping to proofreading chessandgo's forthcoming book. Do I now have an unfair advantage because I have this advice available? But maybe point (a.) isn't a necessary criteria. For example, people playing over a real board can learn a lot from each other, and this info isn't available to everyone.

At some point, we'll need to discuss the new game tree analysis tool clojure is developing. I could see this as being very helpful in a postal game for someone like me. I'm very stubborn about trying to evaluate lots of different variations in the plan window. Being able to save these variations would be a big help. I've considered writing down various options, but haven't yet. Is it OK to use pen and paper to record the options? Is it OK to use the new software tool to record the options, so I have them available in a few days after the next move has been made?

Title: Re: Using tools for improving a match game
Post by Fritzlein on Oct 25^th, 2010, 12:58pm

on 10/25/10 at 07:04:31, clojure wrote:

About pen&paper, books, computer-aid. I personally think that there is clear distinction between calculating with pen&paper or with computer-aided tool. The former is a passive tool that you actively use, the latter has active behavior independent of you. Thus, also using external board as a external memory would be different from automatic material evaluator since a board is a passive entity.

The plan window, however, is not a passive entity: it actively removes pieces from the game for you even if you didn't realize they would be captured, and it actively prevents you from moving frozen pieces even if you didn't realize they were frozen. So by the active/passive distinction, the plan window should be against the rules.

Again, I am not saying that we can't or shouldn't make distinctions, just saying that there is more than one way to do so.

Title: Re: Using tools for improving a match game
Post by clojure on Oct 25^th, 2010, 1:29pm

on 10/25/10 at 12:58:02, Fritzlein wrote:

The plan window, however, is not a passive entity

Yep. I was thinking at the time about real board (had forgot the plan window). Also, as it is only a plan window, maybe it should allow illegal moves, especially when it's an event game.

Title: Re: Using tools for improving a match game
Post by omar on Oct 25^th, 2010, 9:56pm

Actually I didn't mean this to be a ruling. It's just my opinion of how to interpret "A player must not use a computer program to suggest or evaluate moves.". In general for an event game follow the rules provided for it. For non-event games, it's between you and your opponent to decide what tools can be used.

Title: Re: Using tools for improving a match game
Post by Fritzlein on Oct 25^th, 2010, 10:41pm

on 10/25/10 at 21:56:56, omar wrote:

It's just my opinion of how to interpret "A player must not use a computer program to suggest or evaluate moves."

Oh, good. I'm glad that even for event games I will be free to use a calculator. :P

Title: Re: Using tools for improving a match game
Post by Janzert on Oct 26^th, 2010, 12:07pm

Have to say I agree more with Fritzlein on this. My main reason is that it seems to me the client forbidding illegal moves is a much greater help and saves way, way more games than the material evaluator ever can.

I also would hate to see the rule go the route of "only things built into the client". Because I would someday like to, and really hope others will, build third party clients. It would be quite restricting and just plain hard to do if these clients had to conform to feature parity with the official client.

Having said that though if Omar doesn't change his mind I will, reluctantly ;), put up a notice on the material eval page.

Janzert

Title: Re: Using tools for improving a match game
Post by omar on Oct 27^th, 2010, 10:54pm

To the extent that my decision was a ruling it applied only to the Postal Mixer. In general for an event game follow the rules provided for it. It is possible that we may have events in the future that do allow using computers for assistance; kind of like Advanced Chess. For non-event games, it's between you and your opponent to decide what tools can be used.

So having a note on the material evaluator page is not really necessary, it's just that people sometimes forget or don't realize and gentle reminder would help prevent this oversight.