Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 9:49 AM
Ok, but only since you asked
Simply stated: If you wanna play rated, you play a set. If you don't wanna play a set, don't play rated.
This solves every one of your issues.
'I've got winner' CAN be played with sets, but the one who has been sitting longer would rotate out if there's a tie. Or whatever you guys come up with at the table. If it's this informal you can always play unrated.
Working on a new line is primarily done unrated.
In terms of handicapping yourself against beginners -- there's nothing about this that would be WORSE by playing out a set where there's just one rating adjustment after the set is complete. It's not like you just beat him as player 2 where he got whacked for points, then asking to switch so you can whack him even worse from the player 1 seat. If you play the set, there is just one whacking for winning the match, as there should be. Heck, he even gets the chance (however small) to redeem an initial P1 loss by trying to return the favor. Plus, you don't have to continuously risk tons of points by putting yourself in the player 2 seat often. Seems like a win/win all around. Again, you can always play unrated.
If you only have time for one game play unrated. This is like currently saying "I only have time for 6 moves". It's an unfinished match, and if you know you will not finish, either don't start or play unrated.
If you have to abruptly disconnect, the options would be the same as they currently are. You can always request to cancel the game (match). You could just kill the window and let the opponent decide if he wants to force you to resign. ETC. This is exactly no different than the current system. If you sit down and start a rated game (match) you are committing a certain amount of time. If you do not have the time play unrated. If something comes up you may have to face the consequences of breaking your time committment, just like it is now.
Other issues that you did not bring up are those under the "What happens if...?" syndrome, such as what to do if a player bolts after losing as player 1, etc. With some logical thought these issues can be easily solved. In fact, (and maybe this is my bias showing through) I have yet to hear ANY argument against sets being the standard for rated games that was not simply logistics or trivial concerns -- in other words, no argument as to why the SYSTEM PROPOSED is more flawed or does a worse job with respect to ratings and with respect to logically how the game really should be played (fairly).
Since I have this opportunity, I'll add another point about playing pente in sets that I forgot to mention in previous posts: People that create, organize, and run pente tournaments have already come around and realize, quite logically, that in order to have a tournament where there is a clear winner determined fairly, the participants in the tournament must play every match is set based fashion. Especially in elimination rounds, but the same also applies (more subtly and indirectly) in swiss/round robin rounds. Ask yourself for a minute or two why this is so? Why do people play sets in tournaments? Why not just flip a coin to determine if you play as player 1 or as player 2 in the finals of a big tournament?
Well, in case anyone has missed it, let me tell you:
PLAYER 1 HAS A SIGNIFICANT ADVANTAGE IN PENTE
So why do you play sets in tournaments? For the same reasons that all matches should be played in sets. It just makes sense. It's fair. It creates an even playing field. There is exactly no advantage. There is no external forces to affect whether the ratings become skewed. It makes the most sense given the nature of the game.
Let me make one very last attempt to make it clear why I'm connecting the rating system flaws to the player 1 advantage problem (and the obvious set based solution to the player 1 advantage problem). This is a point that I think many of you are still missing:
THE EFFECT OF THE PLAYER 1 ADVANTAGE IS NONLINEAR WITH RESPECT TO THE RATING OF EACH PLAYER.
This is important, and even if you don't like math I hope you all will take some time to consider this fact. Let me explain with an example:
Let's say that a 1000 player plays against an 1100 player. Now, any time there's a difference of exactly 100 points in rating, the formula says that the higher rated player can gain 11.51 points and stands to lose 20.49 points. This means that the EXPECTATION is that the 1100 player must win 64% of the time (nearly 2 out of 3) to maintain the same rating. This expectation is a function of the differences in skill which are advertized by each player's rating.
However, at the SAME TIME, there is another force at work here. If a 1000 player plays a 1000 player, he might expect to win 51% of the time from the player 1 seat and 49% of the time from the player 2 seat (these numbers are estimations). Thus, even though they have the EXACT same skill, the player 1 advantage creates an EXPECTATION of winning more than half of the games. Likewise, if the 1100 player plays another 1100 player, his expectation might be to win 52% of the time as player 1. This doesn't appear to make much difference at all compared with the 64% expectation dictated by the ratings formula that the 1100 player must win against the 1000 player.
But, alas, what if this was a 1300 player against a 1400 player? What if this was a 1600 player against a 1700 player? Finally, what if *gasp* it's a 1900 player against a 2000 player? Well, let me tell you something. The linear ratings formula dictates that the 2000 player must beat the 1900 player 64% of the time. However, if a 1900 player plays another 1900 player of the exact same skill level, the EXPECTATION is that he will win his game as player 1 MORE THAN 90% OF THE TIME! You see, as ratings climb, the EFFECT of the player 1 advantage GROWS SUBSTANTIALLY. This growth of this effect is NONLINEAR with respect to how the ratings adjustment formula dictates that a player should have a certain level of success against another player based solely on skill. In fact, at the highest level, the effect of the player 1 advantage is so great that it is insurmountable with respect to the ratings adjustment formula! Thus, the top players stop playing to the great lamentation of the pente community. And ratings at the top of the ladder are skewed an inaccurate.
Playing pente based on sets solves this crisis by creating an absolutely even playing field. There will be no need to ever change the tournament rule again. It may be changed just for the shear enjoyment of playing under a new variant of the game, not to reduce any inherant advantage. In set play there is no advantage. This external force is removed from the equation.
And we can get back to the business of rating and adjusting ratings based solely on skill.
Posts:
54
Registered:
Feb 21, 2003
From:
Hawaii Home page
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 10:36 AM
up2ng,
You make some good points. I have only worked on a new line here twice. Both times I did it rated, but I can see where most players would do it unrated. Makes sense. However, in general, your 'solution' seems to be, if you want to do whatever, and whatever is not a set, play unrated. That does not seem like a very good solution at all.
The fact the ratings are difficult when a full set is not played does not justify making so many games unrated as you suggest.
I would even argue that doing so would make the ratings LESS acurate because so many fewer games would be counted in players' ratings.
Pente is a great game, but, unfortunately, not a perfect game. The P1 advantage is clearly a flaw. Playing sets negates the advantage, but it is unrealistic to expect everyone to play two games of opposite color every time.
No one has put forth an arguement as to why 'the SYSTEM PROPOSED is more flawed or does a worse job with respect to ratings' because there isn't one. Clearly, ratings would be better if everyone always played sets instead of just games, but Pente is a game, not a set. There are too many times when players will play and uneven number of games, for whatever reason, and forcing us to change that playing is not a good thing. A player's unbalanced history of games, much like the P1 advantage, is just something we have to deal with. Personally, I still think calcualting a weighted average of a player's games played as P1 and games played as P2 is the best solution.
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 12:50 PM
This post is a response to the previous post by thad and dmitriking's first post in this topic.
I agree with that. The reasons that dmitri cites for keeping the current behavior where your rating can go down when you win are still valid issues that must be considered. However, instead of being evidence supporting the backwards rating behavior, they are really evidence of the INADEQUACIES OF THE PROVISIONAL/ESTABLISHED SYSTEM.
The provisional/established system says that after a certain number of games a person's rating becomes established. This is simply not true. Different games reveal different amounts of information about the player's skill. So the rating system should take this into account. The glicko and glicko-2 systems do this because they don't dump everyone into one of two rating categories.
dmitri: "nonsense. the provisional system is useful and its advantages have been discussed. to say it is a non continuous function is injecting meaningless math jargon as a means of confusing people. first, your statement is vague at best; the rating is still continuous it just changes at a different rate. second, if it weer discontinuous, who cares?"
The idea of how established a rating is (let's call this "rating confidence") does not fit naturally into the categories of established and provisional. The more games you play, the more confident the system can be about your rating. This is why it is important to have a continuous function that determines rating confidence.
A discontinuous measure of rating quality cannot simply and adequately represent the information that we are trying to represent because it only has two possible values. There will always be special cases like the situation dmitri is talking about that the current system will not handly correctly.
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 1:07 PM
My statement "A discontinuous measure of rating qualty..." should be "A discontinuous measure of rating confidence...". I inadvertantly switched terminology.
Essentially, a provisional/established system is similar to saying that everyone can have one of only two possible ages: old and young. We measure our age in years because there are more than just two possibilities. The same situation exists in the measure of rating confidence.
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 2:27 PM
up2ng: You present a very good argument for playing pente in sets. However, I think there is one issue that you are neglecting. Playing sets is a social solution to the mathematical problem of creating a good rating system.
I also agree with your KISS principal except for one thing. Complicated systems *require* complicated models to describe them. You can't describe the behavior of subatomic particles using arithmetic. (Well, you might be able to, but it would be arguably more complex because of the absence of abstraction that the higher math gives.)
As thad said, there isn't an argument that your proposed system is more flawed mathematically. The flaw is a social flaw and that is the reason that many people do not agree with you. A good general principle to adhere to is to give people as much freedom as possible. The reality of the situation is that people consider a game to be the atomic unit of pente. No matter what your argument is, you're not going to change the fact that novice players think in terms of a game.
Set play is a special case of this more general idea of game play. If you think pente should be played in sets, then you can accomplish that without any problems in a game-oriented environment. (If the rating system prevents this, then it should be changed.) A novice player who cannot see the significance of the player 1 advantage should not be required to use a system that does not make sense to him--provided that the good players do not suffer from the result.
I think that your argument to the last sentence will be that our current rating does make the good players suffer. That is why I support a rating system that has a player 1 rating, a player 2 rating, and a combined rating. Another possibility would be to have a set-based rating and a game-based rating.
In short, I believe that a game-based rating system can be devised that acknowledges your points about set-based play but does not force all players to use set-based play. Just telling people to play unrated for game-based situations and rated for set-based situations doesn't make much sense to novice users.
I do support your idea that set-based play be strongly emphasized in the printed rules of the new pente that is coming out. The reason I like this idea is that it doesn't force people to play sets. It communicates your point without restricting freedom.
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 2:29 PM
mightybyte and thad: if you do NOT allow a plauer's rating to go down after a win when provisional, then he can do exactly what i said -- beat a 1600 provisional player, then 19 in a row against a 600 player, for an established rating of 1800! this is bad. but i dont think a glicko system is needed.
1) make a player provisional until he plays 20 DIFFERENT OPPONENTS!
2) as for sets: if a player plays only one game of a set, the ratings arew unfairly affected. to solve thad's poroblem (a legitinate one) of people having to play lots of unrated games, the system could "save" a sinlg game as part of an unfinished rated set, ready to be completed the next time they are both at the table.
If I do not accept a game invite right away, it means I will once I have fewer games in progress.
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 2:55 PM
I agree that it is possible to fix the current system in some way. Your suggestion #1 is definitely a good one that I had already thought of. A glicko system isn't required to solve our problem, but I think it is still superior to fixing the current system. Why not adopt something that can represent the situation better? Even though you say the glicko system is not *needed*, no argument has been presented as to why we shouldn't adopt the system. The only argument that I can see is the amount of work required, but that is kind of intrinsic to any suggested improvement.
Your suggestion #2 about sets is also a good one that could make the idea of sets transparent. However, players might still wonder why their rating didn't go up after a game. I think it would nice to have one rating that goes up after each game and one that would do set-based. It would also be interesting to compare those two ratings. The other problem with saving the single game is the disk space required. I don't think this problem is big enough to eliminate the idea, but it is also a factor.
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 7:08 PM
Wow, this has been quite a series of threads. Amazingly, it seems like some agreements/conclusions are actually being reached, which frankly I never really expected.
Let me just clean up and recap what's going on here. There are two separate issues being debated in this same thread:
1) The provisional system -- pros/cons/solutions
2) The established system -- pros/cons/solutions
I have mostly been focusing on issue #2. The gliko posts seem to mainly be focusing on issue #1, which is also in need of help so I'm glad it's being addressed.
I did propose one VASTLY simpler solution to issue #1 by brainstorming that we could scrap the provisional formula and replace it with the established formula, BUT: Set K = 320 / ((n / 2) + 1) --> n/2 is integer division n ranges from 0 to 19 over the first 20 games like it does currently. In this case, K will start out at 320 and glide smoothly down to 32 at which point the rating becomes established and the normal established formula would be used, where K = 32, thereafter. Hand in hand with this change, I would propose a "provisional cap" where you CANNOT become rated higher than the cap, let's say 1900 or 2000, while provisional. This solves the "flaw" of gaining when you lose and losing when you win while preserving some protections against ratings fraud attempts to hijack the #1 position on the board.
It also addresses the main problem with the gliko system. The problem with gliko is that it is just too complex. People "trust" a rating system where they can see the formula, it is perceived as fair. I forsee trouble when a player starts noticing that his rating is changing "seemingly randomly" after he completes a game (match) and this would cause lots of grumbling -- probably to the point where another big thread like this will be filled with folks calling for a simpler system. Just as an extra argument against gliko (and I must say here that gliko is certainly a decent idea by the way), you say you haven't seen any mathematical reasons why gliko is a worse system. I'd say that this "appears" to be true after trying to wade through some of those gliko websites. But I cannot say for certain because I cannot see a hard and fast formula for adjusting the ratings. If these adjustments are made based on more and more "parameters" that must be tweaked to the right values in order to see the correct resulting behaviors -- then I am firmly AGAINST this. This type of math is more subjective and less inherantly accurate than using an actual formula. But like I said, I'm not sure because I cannot understand the gliko math -- and that's a bad thing.
Ok, moving back to issue #2 and the ingenious set-based solution.
It appears that thad has agreed with and mightybyte has clarified, that playing in sets is the correct solution:
Thad: "The P1 advantage is clearly a flaw. Playing sets negates the advantage..." "because there isn't [a valid mathematical argument against sets]. Clearly, ratings would be better if everyone always played sets..."
Mightybyte: "Playing sets is a social solution to the mathematical problem of creating a good rating system."
Oooh, well, in this first attempt, he got it backwards. Interchange the words "social" and "mathematical" and you've got it right.
"there isn't an argument that your proposed system is more flawed mathematically. The flaw is a social flaw and that is the reason that many people do not agree with you."
WOOOOOOOOHOOOOOOOOOOO! Thank you SOOOOOO much for getting it right the second time! I fully realize that social flaws are tough to overcome when trying to do the right thing. Racial discrimination was not solved over night. It is WRONG, but it has been a long road to make good progress.
Now if ONLY you had both stopped talking after making the correct statement! But alas, I must respond to all the bad stuff that followed.
Thad: "if you want to do whatever, and whatever is not a set, play unrated. That does not seem like a very good solution at all."
Eliminate "not" and "at all" from the second sentence and this becomes a true statement.
"I still think calcualting a weighted average of a player's games played as P1 and games played as P2 is the best solution."
This is atrocious. No such calculation exists that is fair and accurate. End of story.
Mightybyte: "...novice players think in terms of a game."
I must admit that the "freedom" idea is important when designing software. But if the game itself is supposed to be played a certain way (as a set in this case), the freedom is not restricted at all. You have the freedom to play one "game" in tennis too but you will never be rated based on the result of such a contest (NOTE: In tennis, a "game" is the unit of points played while one player controls the serve. Hmmmmmm.......) Playing one game of a set is simply an unfinished match. If you wanna play this way, play unrated. Furthermore, if NEW players join the site when ALREADY converted to set-based play, I ASSURE you it will make perfect sense to them. They will not need to "unlearn" a bad habit. They will simply play the game correctly from the start.
"Set play is a special case..." um, no. It's the only case that matters.
"If you think pente should be played in sets, then you can accomplish that without any problems in a game-oriented environment. (If the rating system prevents this, then it should be changed.)"
I have already pointed out that one catastrophic side effect of playing a set in a "game-based" system leads to the FACT that it is better to play as player 2 first, then player 1 rather than the other way around. This is mathematically indisputable and is so important that I would argue ratings become almost irrelavent and meaningless when calculated under such a system.
"I think that your argument to the last sentence will be that our current rating does make the good players suffer."
While this is a true statement, it is not the main argument. The system itself is dramatically flawed for everybody, not just for the good players. It's just that you often have to become good to realize how screwed up it is. The root of the flaw for established players has NOTHING to do with the formula. Leave the damned formula alone. Apply the formula to a contest that has a level playing field so that it becomes meaningful.
"...support a rating system that has a player 1 rating, a player 2 rating, and a combined rating. Another possibility would be to have a set-based rating and a game-based rating."
These are terrible suggestions in every way. Adding more flawed ratings and flawed data and flawed parameters to a flawed system is not the correct approach to solve the problem. But you can go ahead and give it a try -- while having all these separate ratings certainly would not HURT anything that is already broken (except to cause more complexity, confusion and general discontent), it CLEARLY does not solve the problem.
AND PLEASE FOR THE LOVE OF GOD DWEEBO I'M BEGGING YOU PLEASE PLEASE PLEASE DO NOT ALLOW SET-BASED PLAY TO BE OPTIONAL! Do not bastardize the solution and inject VAST inconsistency into an already broken system. Do it right or don't do it. If it's done half-@ssed I will quit the game. Or at the very least I will boycott the site. I would support aboloshing ratings and leader boards completely before I support making set-based play "optional". I do not ever want to hear the phrases "set-based" and "optional" in the same paragraph ever again -- I will start throwing sh!t.
"...a game-based rating system can be devised that acknowledges your points about set-based play but does not force all players to use set-based play."
No. Sorry, it is impossible. I have proven so countless times and I'm tired of it. Play unrated. I do not necessarily expect to ever see this problem fixed, but i can GUARENTEE that without this solution, the system will remain broken. This is fact, and is mathematically indisputable.
Some quick miscellaneous responses:
"1) make a player provisional until he plays 20 DIFFERENT OPPONENTS!
2) as for sets: if a player plays only one game of a set, the ratings arew unfairly affected. to solve thad's poroblem (a legitinate one) of people having to play lots of unrated games, the system could "save" a sinlg game as part of an unfinished rated set, ready to be completed the next time they are both at the table."
I think these are more like "patches" than full solutions. They help make the problem less of a problem but they do not solve the problem completely. I do not support "saving" a half-match. This policy would be abused.
Well, I have broken my promise The reason I didn't want to make any more comments on this topic is because I'm starting to have a hard time controlling my tone. I hope I did not offend anyone.
The reason I am so fired up about sets is because I know how to agree to disagree on differences of opinion. But when I'm arguing FACT, I have a hard time letting it go in the face of resistance.
Posts:
1,032
Registered:
Dec 16, 2001
From:
Powell, OH
Age:
37 Home page
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 8:38 PM
up2ng: I have to agree with KISS, it is something software developers have to follow due to the complexity of creating "good" software, and it also makes sense for ratings.
When I mentioned set-pente as "optional" I guess I was thinking more short term, as a way to try it out and let people get a feel for it and experiment. Like the other experimental pente variations we could have "Set-Pente" that must be played in sets. I know I said optional and set in the same sentence, don't kill me In line with what you said about social factors, people aren't going to buy into Set-Pente right away and you're not going to convince the world by writing posts at DSG, people are going to have to see it in action I think.
Also, something else that I don't think has been really covered yet that is related. I know some top players who stop playing the game because of the P1 advantage, and I don't think Set-Pente will make them more interested. If I knew how to always win as P1, why would I want to play, even WITH Set-Pente. I'd always win as P1 and expect to lose as P2 against an equal opponent. Obviously no one is that good, but I know some players who see it as inevitable, P2 can't keep coming up with new defenses forever. These players want changes to the rules of Pente (pente variations), I know there have been discussions of this before here at the site.
Now, variations to Pente is another "hot" topic and one that we can all argue back and forth about so I'm not going to discuss specific variations, we can do that in a separate thread. I just wanted to say that I believe Pente will eventually be solved and at that point Set-Pente will not help, only changes to the rules will.
Ok, I said I wouldn't mention specific variations, but I will say I've been thinking that D-Pente sounds like fun to me, coming up with the 1st 4 moves that are balanced would be interesting. I am thinking more and more about implementing it here at DSG, I know a bunch of you would like that
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 9:06 PM
i agree with up2ng -- sets mandatory or not at all.
variants help, but one side will always have the edge, so variants will not eliminate the need for sets.
i realy like two ideas:
1) provisional until 20 different opponents and
2) a cap on a provisional rating. the actual can be kept secretly, like the robots portions of the human ratings, and then if over 1800 after provisional period, it goes to 1800.
If I do not accept a game invite right away, it means I will once I have fewer games in progress.
Posts:
542
Registered:
May 9, 2002
From:
Northeast USA
Re: DSG Rating Suggestion
Posted:
Feb 5, 2004, 11:03 PM
Dweebo:
It seems that you are pretty closely in line with me about many of these issues. In terms of experimenting with set-based pente in some optional way, I guess my main concern is that it was suggested previously that people would literally have an option in the game room to set "set-based" ratings adjustments or not before the match -- and whether you used set-based ratings or didn't, they would both end up factoring into the SAME rating! This is clearly inconsistancy at its finest and simply cannot be done. Now, if you consider "set-based" to be some sort of "experimental variant", sort of like G-pente, poof-pente, or D-pente (you're right, D-pente rocks and makes the "best" attempt at reducing the player 1 advantage in terms of rule changes -- would love to see it show up at DSG soon ) with it's own separate rating, then this wouldn't be the end of the world to try things out -- I just fear that it would hardly ever be used, such as how G-pente isn't really used much now. But you are right, there's no way just talking about it will convince everyone.
Let me briefly touch on your other point about pente eventually being solved because I have a philosophy about that. First, let me mention that when I originally came up with and started pushing set-based pente, it was because I was originally attempting to come up with a rule change variant! I was examining why there was a need for a rule change. I finally conceded that there really is an inherant player 1 advantage (I was stubbornly against this fact for a long time -- as I developed as a player it just became undenyable). Next, I thought hard about WHY there is a player 1 advantage. While doing so, I made many attempts at creating rule changes that would completely eliminate the advantage. Then, it hit me like a ton of bricks: No rule change within a single pente game exists that can eliminate one player or the other from having at least some advantage. This fact is undenyable. But I still wanted a solution. The game could not be ruined because it is unfair. So I started thinking outside the box, took a different approach, and said -- hey, it seems to be convention that people play sets to make it "more fair". But is it exactly fair? This is when I made the calculation and discovered that playing from Seat 2 first is advantageous. I was frankly shocked about this. Eventually, I saw the light, and understood the fact that ratings should be adjusted ONLY ONCE -- at the conclusion of a set. This makes an even playing field where the expectation is to tie. But the expectation of one person to win a set rather than the other is solely based on skill as determined by your rating.
Thus, it is the changing over to sets that is the REQUIRED part, and it is the creation of rule changes that is the OPTIONAL part, not the other way around.
Now, on to the philosophy. First, if the game were on an even playing field, I think you would find less disinterest amongst the top players. The reason is, as you said, "no one is that good". So, although the expectation to tie is overwhelming, there is at least some expectation that one player will win the set.
Let's compare this to tic-tac-toe for a second. In tic-tac-toe, white has the advantage -- this is undenyable. White goes first and will have 5 Xs at the end compared to the 4 Os. There are more ways to choose 5 out of a 3x3 grid to create 3 in a row than there are to choose 4 out of a 3x3 grid to create a 3 in a row. However, an INDIVIDUAL game of tic-tac-toe does not EXPECT to end in a win. In pente, for all practical purposes, someone ALWAYS wins. So, the player 1 advantage becomes a real factor and becomes so great that it becomes THE overwhelming factor in the outcome of an individual game. Now, in tic-tac-toe, the individual game itself is unwinnable. In other words, the EXPECTATION to TIE is orders of magnitude greater than the expectation to win due to the player 1 advantage. Now, I am going to make a leap here -- I am trying to say that IF you play pente in SETS, the "general behavior" becomes similar to a tic-tac-toe game. The expectation becomes a tied result. In fact, playing pente in sets yields 0 advantage for one player, where as tic-tac-toe yields "close to 0" advantage for player 1 (because the TIE is such a greater factor).
Hmm, this analogy is turning out to not work as well as I had hoped but I hope I am getting the idea accross. I'm saying that, tic-tac-toe has a (very small) advantage for player 1 since player 1 goes first. In a pente set, the more skilled player has a (small, but still much larger than the "very small" used above) advantage, but this advantage in this case is based solely on skill (ratings difference). As the game becomes "more and more solved", the advantage to win a set becomes smaller and smaller. But, as I will argue next, due to the game's inherant complexity and subsequent inevitability of human error, this advantage will never reach 0, even when the game becomes "completely solved".
This leads into the actual philosophy I keep promising. The philosophy is that at the very highest levels of pente, it becomes a 1 player game. It is a game of "solitaire" if you will. It's a vast and beautiful puzzle, to be struggled with by the player in order to excercise the mind. This puzzle can indeed be solved. Once the puzzle is solved, the fact that the mechanics of pente involve another player becomes completely meaningless and irrelavent. The hero becomes unbeatable. The game is conquered. It's very much like playing with a rubic's cube. Perhaps player 2s job is to muddle up the rubic's cube as furiously and as cleverly as possible. You can struggle and struggle with the rubic's cube for weeks at a time and still have trouble with it. You will make mis-steps. You will make steps backwards. After battling hard, you will often be able to "complete" the puzzle, putting the cube into it's final state. But you have still not "solved" HOW you got the cube to its final state. The next time someone comes along and muddles it up, you will have to struggle all over again.
BUT, eventually, after many many many repititions and much studying of the puzzle, you can actually SOLVE the rubic's cube. In other words, there is a precise, finite, algorithm that can be used in each and every permutation to complete the puzzle, and you simply have to follow the steps accurately and it will be solved. (If you don't believe me, look it up on the internet, you will be shocked and awed.) Once you master this algorithm, you can bring it into the office, tell your cubemate that he can muddle it up as much as he wants, and you bet him $10 that you will have it solved in less than 30 seconds. Guess what, you will win every time. (A co-worker of mine actually did this.)
However, I would argue that pente is even vastly more complex than the rubic's cube, and once you think you have the full solution, there may be special cases that do not fit the model. Furthermore, even when you solve the rubic's cube, when you are pressured to complete the permutation in less than 30 seconds from a starting point that you have never seen before -- well, you may make just one move where you twist to the right when you should have twisted to the left! BAM! You just lost! For even if you corrected your mistake right afterwards, you just lose two moves, and this mistake cost you the ability to complete the puzzle with optimum efficiency. In pente, one such mis-step is death -- and guess what, it happens all the time, even to the world's best.
For this is a game of solitaire which you play for weeks or months at a time. This period of time is "preparation" time. It's the time you spend studying for your final exam. But guess what, even a Harvard Law Rhode's Scholor has to show up and take the bar exam. You have to put that preparation to the test. Sure, maybe you solved the game based on knowing the correct move is "someplace" within the DSG database -- or within the MM_AI opening book. But can you apply your knowledge? THAT is the contest at the highest level. I give you the hardest test question I can come up with. You see if you can apply your knowledge under the pressure of time to adequately pass the test. Those that perform the best under these conditions are the best players, and should be rated the highest.
But, unless we are playing sets, this whole concept is meaningless with respect to ratings, and thus, the highest level of the game is ruined.
Posts:
39
Registered:
Dec 16, 2001
From:
~THE BACK WOODS~
Re: DSG Rating Suggestion
Posted:
Feb 6, 2004, 12:29 AM
Using the tourney rule should be accompanied by playing tourney style Pente (set-based). Tourneys must be played using set-based systems in order to be fair. Therefore, we should not be ranked or rated for anything less. If our tournament games here at DSG are going to be played using the tourney rule combined with set-based play, then why should our regular daily play using the tourney rule be any different? Why the double standard?
WHY? WHY?? WHY??? Set the standard! Keep the game consistent!! Pente should be played in sets!!!
