IMO, Glicko rating incorrectly implemented for draws

[Roastasaurus]

This is not an argument, but you know why hockey and soccer are so incredibly boring?  All those dumb ties.  When the 2002 all-star game for Major League Baseball ended in a tie, people ranted and raved.  You know why?  Because ties are boring.     You just don't get the sense that you accomplished anything!

LOL, perhaps in soccer. But as someone who's fought in martial arts matches and tournaments, I can tell you that battles that end in ties are by far some of the most BRUTAL. 

[Roastasaurus]

I edited the title of my post, because it's been made clear that it's a matter of opinion rather than mathematics.

[Lucifer Rising]

It's not unpopular -- I think it was a good suggestion.  Also, Fossil, my argument hasn't been "That's the way it is, so deal with it."  The argument that I *did* make was that, if we were to change the current system, you would have to show me that the new system confered a significant advantage over the old one.  From what I've read here, I just haven't seen where that is the case.  

Like I said earlier, Roast, I do enjoy your posts.  

[masterfool]

Roast,
  I do see your point.  And I don't disagree with it.

However, I DO AGREE with LR.  At this point in the game, there is no such THING as a "tie" . . . and I don't much see the advantage to recoding the game to make one.

[Roastasaurus]

Roast,
  I do see your point.  And I don't disagree with it.

However, I DO AGREE with LR.  At this point in the game, there is no such THING as a "tie" . . . and I don't much see the advantage to recoding the game to make one.

Nod, I will start a post in "Ooh Ooh Pick Me" suggesting a double-elimination Swiss style Duel Tournament (for each weight class), and creating a new board (divided by weight class) that shows the ranks of winners. But it'll probably be a few days before I write this. Thanks all for comments.

[masterfool]

I wouldn't object to such a tournament.  In fact, I'm all for it.  I LOVED the "duel tournament" idea when it was first raised.

[Fossilized]

Also, Fossil, my argument hasn't been "That's the way it is, so deal with it."  The argument that I *did* make was that, if we were to change the current system, you would have to show me that the new system confered a significant advantage over the old one.  

The advantage would be that the "greatest fighter" rating would more accurately reflect fighting ability, as opposed to 'gold stealing' [there is a difference], for which two boards already exist.

[masterfool]

How is that an advantage?

It DOES reflect fighting ability, and NOT gold stealing ability.  Greatest Fighter does not take into account how much gold you sole, or even that you STOLE gold (you can win and steal 0).

I think there was a misinterpretation, or a misstatement, about how Greatest Fighter works.  You should think of it this way:

As an attacker, a "Win" is any battle in which you reduce your opponent to 0 HP or less in a 20 round period, AND you maintain at least 1 HP.  All else is a "Loss".  There is no "tie"

Therefore, the Greatest Fighter board accurately reflects fighting ability, as determined by the rules of the game.

Think of it like getting an 'out' in baseball.  The batter hits the ball to the 3rd baseman, and runs to first base.  3rd baseman gets the ball, and throws to the 1st baseman.  One of two things occurs:  (a)  The runner is "out"  (b) The runner is "safe" . . . "out" ONLY occurs in the specific scenario where the ball reaches the glove/hand of a person who touches the base BEFORE the runner or touches the runner before he touches the base.  In all other scenarios, INCLUDING when the runner and the person with the ball touch the base at the same time (a "tie"), the runner is "safe."

[Fossilized]

As an attacker, a "Win" is any battle in which you reduce your opponent to 0 HP or less in a 20 round period, AND you maintain at least 1 HP.  All else is a "Loss".  There is no "tie"

That is how it is currently defined. My point is that I think this is wrong. WHY should the attacker have to retain 1 HP?  If you and your opponent end on 0 HP, as far as the FIGHT is concerned, it's a tie - a double knockout. You both FOUGHT equally well and are both equally KO. I don't see why this is such a controversial issue - after all, this is the accepted and currently implemented system for duels.  Beyond that, it strikes me as being the more intuitively correct way of implementing things (and I'm a big fan of games being intuitive). The current system is like saying that a stalemate in chess is a loss for white because he/she took the first move. Of course, you COULD define the rules that way and chess players would have to put up with it. But is that the way it SHOULD be?  As far as I'm concerned, absolutely not.

Therefore, the Greatest Fighter board accurately reflects fighting ability, as determined by the rules of the game.

Obviously, the boards reflect the best fighters according to the current rules - again, my whole point is that those rules are wrong, or at least, that the board is misnamed. The current definition of win/loss presupposes that the only reason anybody launches a raid is to steal gold. As far as I'm concerned, that's just not the case. As I stated above, I often launch raids to test my strength and win glicko.  In the example I cited, I selected my opponent based on their combat record, not on gold (they had hardly any) in the hopes of climbing the rankings. It's a different way of playing the game. One that isn't accommodated by the rules as they currently stand (except in duels, where it is). Sure, you can write the rules so that a tie counts as a loss for the attacker (as has been done) - but that doesn't make it right

If you're not interested in climbing the GF board then this whole topic is probably an uninteresting and insignificant problem, but if your ambition is aimed at gaining (or regaining) the top spot, then being told you've lost the fight, when the other dragon is in worse shape than yours is really, really annoying 

[Comethenastrainer]

I completely agree with the points that Roast made, and I think Fossilized makes some good ones here too.

It's very frustrating to knock out one's opponent to zero, albeit suffering a knockout oneself in the process, and lose tons of glicko over this even though the fact that you were fighting two dragons actually means you were the better fighter on the day.

I understand and am not arguing aginst the practice of recording such a drawn result as a loss to the raider, since I have indeed failed to raid any treasure. However, I don't believe that a battle-strength measuring mechanism as glicko should reward either the attacker or the defender with either increased or reduced glicko, in the case of a draw. Such a neutral glicko outcome would better represent the fighting outcome of the battle (as opposed to the stealing outcome, which as we know won't always be the same due to factors such as elemental attunement, starting treasure size, etc).

Does this remove the defender's 'natural' advantage in defending?
No, because
(a) They still record a win, and
(b) Even if they are knocked out with an assistant (i.e. obviously perform the worst), they still lose no glicko from the fight.
So draws still leave them with clear advantages.

But this change in the system would result in the very necessary, IMO, change in helping glicko better reflect who are the better fighters.

WHY should the attacker have to retain 1 HP?  If you and your opponent end on 0 HP, as far as the FIGHT is concerned, it's a tie - a double knockout. You both FOUGHT equally well and are both equally KO. I don't see why this is such a controversial issue - after all, this is the accepted and currently implemented system for duels.  Beyond that, it strikes me as being the more intuitively correct way of implementing things (and I'm a big fan of games being intuitive). The current system is like saying that a stalemate in chess is a loss for white because he/she took the first move. Of course, you COULD define the rules that way and chess players would have to put up with it. But is that the way it SHOULD be?  As far as I'm concerned, absolutely not.
Obviously, the boards reflect the best fighters according to the current rules - again, my whole point is that those rules are wrong, or at least, that the board is misnamed. The current definition of win/loss presupposes that the only reason anybody launches a raid is to steal gold. As far as I'm concerned, that's just not the case. As I stated above, I often launch raids to test my strength and win glicko.  In the example I cited, I selected my opponent based on their combat record, not on gold (they had hardly any) in the hopes of climbing the rankings. It's a different way of playing the game. One that isn't accommodated by the rules as they currently stand (except in duels, where it is). Sure, you can write the rules so that a tie counts as a loss for the attacker (as has been done) - but that doesn't make it right

[sparkle]

+1

I haven't read anything in this thread but I agree with the title. It's simply more accurate. I cannot understand why draws weren't considered into the system. Perhaps the glicko calculation was just too difficult? However, I would support them if they wanted to go ahead with the idea.

If this happens, the records should also show the draws:

Attack/defense/assist/duel record: X-Y-Z instead of just X-Y

X= wins, Y= losses, Z = draws

Of course, these are big upgrades, and we could do without them now. For now, enjoy the game

[Lucifer Rising]

Of course, these are big upgrades,

No, not really -- as I pointed out in my earlier posts, the "draw" people simply want to redefine the conditions under which a battle can be considered a win or loss.  The rational is perfectly valid, which is why this isn't a BAD idea.

Is this a good idea, though?  Well, not necessarily.  The reasons is that the conditions which currently define what constitutes a win or a loss are ALSO perfectly valid.  What do you do when you have two perfectly, but mutally exclusive sets of conditions in such a scenario?  Just go ahead, pick one, and consider it a day.  That's basically the same point I made to Roastasaurus earlier, and why he eventually gave up the argument. 

Think about chess.  You may win, lose, or draw.  Why are there draws, though?  Why not just have either winning or losing?  There *is* no reason.  That's just how the tradition developed.  We could have just of easily had a game where draws were not a possibility.  It changes the game, of course, but there are no grounds saying that one is "better" than another -- certainly not in the way that chess is a "better" game than checkers.

I hear what you're saying, guys, but I think the discussion is a dead end.   

[Scythero]

I believe that the reason there are draws in chess is because due to the rules of the game, players can have situations where the do not have enough material to mate, or one player is unable to move while they have not lost. Draws aren't arbitrarily decided, they are most often forced as a strategy. However, chess has to deal with multiple pieces that are either there or they are not there (no gray areas), while here we just have two (or three) dragons with a HP stat that gets depleted.

[Comethenastrainer]

What do you do when you have two perfectly, but mutally exclusive sets of conditions in such a scenario?  Just go ahead, pick one, and consider it a day.  That's basically the same point I made to Roastasaurus earlier, and why he eventually gave up the argument. 

Think about chess.  You may win, lose, or draw.  Why are there draws, though?  Why not just have either winning or losing?  There *is* no reason.  That's just how the tradition developed.  We could have just of easily had a game where draws were not a possibility.  It changes the game, of course, but there are no grounds saying that one is "better" than another -- certainly not in the way that chess is a "better" game than checkers.

I hear what you're saying, guys, but I think the discussion is a dead end.   

I'm not sure that the fact that games can hypothetically have two or three outcomes means that this discussion is logically at an end, and as I'm not proposing that draws should henceforth be recorded in the battle scores as draws, I won't enter into the chess discussion (as it's irrelevant to my points about glicko).

The reality is that in a double-knockout, where the attacker has fought both an assistant and defender, it has clearly performed better on the day, and should not lose glicko although it has lost the raid.

The system currently effectively rates glicko based on whether or not gold has been stolen, which is a misleading system considering the amount of gold stolen often bears little relation to how effectively one's target has been beaten.

[SaintG]

The reality is that in a double-knockout, where the attacker has fought both an assistant and defender, it has clearly performed better on the day, and should not lose glicko although it has lost the raid.

I dispute this reality 

The reality of this game is that there is no difference what so ever between the attacker having all his health depleted in round 1, having a double (or triple) knock-out in round 12 or having all involved dragons at full health at the end of round 20. All conditions equal a failure from the attacking dragon to achieve his goal. AKA "a loss".

Glicko in no way accounts for how close you were to win or any other conditions than the score of both participants (and the stability of that score if you want to nitpick  ). It measures if you won or not

The system currently rates glicko based on whether or not the attacking dragons achieved the goal of an attack (winning) or not (losing). There simply is no draw in the rules, and as Lucy says, that's a design choice with no clear answer. Using a binary outcome probably simplifies calculations, though. What should a be counted as a draw? Only knock-out? Does it count if the Assistant is the last dragon standing? Outlasting round 20?

All in all I think the current model works well, and I have problems seeing the alternative work as well. It is after all a well know strategy in almost all ranking sports (boxing, other combat sports, chess, tennis)  that you can either fight a few tough matches (and win them) or a lot of easy ones to climb the rankings. Both strategies will get you very high.