• English
  • Deutsch
  • Français
  • Español

A new score system ?

46 replies [Last post]
baron rojo's picture
Last seen: 6 years 50 weeks ago
Tech bot
Posts: 2
Points: 2
I like yeku's proposal... should be tested

And lyo... I also don't mind if they reset all accounts... or if they only reset my accounts :) I would go up over and over again if I wanted ;)

Regards

Rote
east amloo's picture
Last seen: 2 years 22 weeks ago
Modo
Posts: 1258
Points: 1260

A periodic semi-reset of points (Each month would be good, as in the first 7-8 months of 2006)

It's not true. It was a 3 manual reset only in 2006, and admins cannot do that. Reset must be occasionnal !
Your score system is so hard to do, and finally not interesting a lot. Why can't do a basic system of score ?

Maybe we can transform the actual score in a championship score ? No more score. The winner gain some position in the scoreboard. The scoreboard at the end of the party show the position in the scoreboard and not the points.

yeku's picture
Last seen: 11 hours 56 min ago
Modo
Posts: 420
Points: 421
east amloo wrote:
A periodic semi-reset of points (Each month would be good, as in the first 7-8 months of 2006)

It's not true. It was a 3 manual reset only in 2006, and admins cannot do that. Reset must be occasionnal !



Well, I didn't explain that well, in the first 8 months of 2006 the points were reseted and players lost some stars.

And don't worry for this system, that isn't finished. Next week I will explain it better ;)

chris's picture
Last seen: 48 weeks 1 day ago
Admin
Posts: 852
Points: 848

Some answers and comments on the last posts

There are several reasons to change significantly the score rules:

  • Have something more simple, understandable by everyone and easy to calculate by mental arithmetic: when you see the score screen at the end of the game, you should understand the formula directly (it will need more details in the score table screen). You know then what you should do in the next games to win more points. That's why I would really like to have only one formula with simple things inside even if it's not perfect for every case.
     
  • Avoid periodic reset, as far as possible, because if a reset becomes necessary, it means that the rules aren't so good, and many players don't like reset (especially points hunters, and admins also ...)
     
  • Something more attractive for new players, with easy points earning at the beginning and more difficulty when the score grows. The old scores rules have many advantages, but are not attractive enough for new players, and very difficult to undersand, even for experimented players.
     
  • Give some novelty to RS. Why should we do a reset, if we keep quite the same formulas (except one real and full reset as pinpointed by Amloo)? This new score system wont boost RS audience by itself, but a better score rule or even only a simpler system can't be a disadvantage if fair enough (I'm OK to keep negative points for this purpose and make it harder to win points for bigger scores). As Amloo said, we should also give points not only to the winner.
     
  • Prevent cheating as far as possible. That's why we have less points for races (ex where the winner plays always the best path against new players who don't know the map) and should take imho the killed bots into account even in races (a 1vs1 race with no kills is the most risky cheating configuration we all know). That's also why we should have less points for the winner when there is less players in the game.

Back to the new formula : Score = nbplayers (for the winner only) + nbbotsyoukilled (x 3 for the winner) - 1% x yourscore / nbplayers

The new C coeff suggested by Yeku to add to the current formula bellow is a clever addon but doesn't change significantly the global working of the score system based on other players' score.
gain = C x [ coeff x mini(average opponents score, average score) ]
with coeff = (0.5 to 1 depending on the type of game) x nb of opponents/2
with C = (Px5)/(AP+0,05*WP) => there is maybe a little error in this formula wich doesn't give something between 1 and 5
The formula remains difficult to understand and depends on opponents' scores. I think it doesn't match many of the aims listed above. And technically it's difficult to access the number of active players and all the ranking when in a game in order to calculate this C coeff.
But another way to use the ranking can be the ranking of the player versus the other players inside the game: this is a very great alternative to improve the system and give more points to winners with low score against big ones.

Yeku said about the new formula:

  • Some kills in DM won't count : ex a bomb frozen bot on a slider to hell : Let's assume that we will be able to calculate 95 % of the kills with some new code and improvements, including a frozen bot on a slider to hell :D, a pushed bot wich goes by itself into a pit (to hell too of course :)). I have to work on it but I'm optimistic.
  • There is not enough points in the races versus death matches : I agree (to be improved in the formula)
  • Kills shouldn't give points in races : Hmmm, OK I agree that it can't be the only point source and that we must upgrade the formula, but take the kills under account in a race is not a bad idea, it prevents a little from cheating (see above) and balance the game between speed race and fight (a race can be won also with kills only and no validated marks)

Well now, back to work for a better formula including as much as possible those comments!

bye
Chris

dharokan's picture
Last seen: 1 week 1 day ago
Modo
Posts: 779
Points: 783
I can't go deeply into the formulas right now (ewww my head! :smoke:).
And I'm surely not the best for that anyway.

But I have 3 points that I want to emphasize:

1. Ranking needs to be part of the scoring in some way.
If a newbie beats an experienced player, he will be damn proud of that. And he expects this to be regarded by the score results, too.
This is (imho) one of the most important aspects to motivate players. They try to get better, to get some practice in order even to beat the "good ones". If the scores would ignore that, whey would surely be disappointed.

Maybe your yellow suggestion to take only the ranking fo the players in the game could do the job. I'm not sure though...

2. Races need their own system.
Races that are won by tagging all the marks can't be judged like a deathmatch. 
So I agree in ignoring the number of kills for them.
But to make it more interesting: Why don't we calculate the score depending on the tagged marks? For example we could use the difference of the players marks to the average number of tagged marks in that race. This could be a nice multiplier.

3. I agree with Chris' idea about transparency
We could create a new game's result scoreboard that shows why the players get the points they get. But for doing so, the formula does not neccessarily need to be simple. Even complex calculations can be visualized by some nice information graphics. 

For example, you don't have to show the whole formula. It's enough to show the things on that the results depend. Some nice icons for the kills of each player, an icon for his ranking, for the number of players etc.. (or for the marks :unibrow:)
Such a result table can be very motivating, I know.

But the optical output should only be the 2nd step, imho. First things first :p

Ahoi,
Dharo
yeku's picture
Last seen: 11 hours 56 min ago
Modo
Posts: 420
Points: 421
Chris wrote:
with C = (Px5)/(AP+0,05*WP) => there is maybe a little error in this formula wich doesn't give something between 1 and 5


I edit all this:

it should have written this  C = (Px5)/(AP+0,05xWP)  and not C = (Px5)/(AP+0,05*WP)


But the formula is the same so:

I know, the formula would be very complex adding the active players and the position of them.

But, about the coefficient, it wors well. It always give a number between 0 and 5, but I told you that if it was less than 1 it should be 1. I explain you:
P=position in the ranking
AP=active players (the total number of players that appear in RS (who have played in last 7 days.
WP=Points of the winner.

You can see that P is always as AP or less.

  1. If P=AP and WP=0, there is the biggest possible coefficient--> C=(Px5)/AP=(Px5)/P=5
  2. But, only 1 player will have P=AP, so AP>P, so, the C will be smaller. And as more points the winner has, the C decrements more and goes to 0.
Examples:
Position=150 Active player=200 Points= 100
C=(Px5)/(AP+0,05xWP)= (150x5)/(200+0,05x100)=3,6585...

Position=130 Active player=130 Points= 0
C=(Px5)/(AP+0,05xWP)= (130x5)/(130+0,05x0)=5

Position=150 Active player=150 Points= 1100
C=(Px5)/(AP+0,05xWP)= (150x5)/(150+0,05x1100)=3,6585

Position=15 Active player=200 Points= 2100
C=(Px5)/(AP+0,05xWP)= (15x5)/(200+0,05x2100)=0,24-->it would be 1

Position=1 Active player=250Points= 3000
C=(Px5)/(AP+0,05xWP)= (1x5)/(250+0,05x3000)=0,0125-->it would be 1

Do you see it is aways between 0 and 5?

Because i said that:
Yeku wrote:
The higher coefficient is 5 and for the players that the coefficient is less than 1, they would have 1 (the score system would be a "skill-system".


Yeku
yeku's picture
Last seen: 11 hours 56 min ago
Modo
Posts: 420
Points: 421
dharokan wrote:

2. Races need their own system.
Races that are won by tagging all the marks can't be judged like a deathmatch. 
So I agree in ignoring the number of kills for them.
But to make it more interesting: Why don't we calculate the score depending on the tagged marks? For example we could use the difference of the players marks to the average number of tagged marks in that race. This could be a nice multiplier.


Totally agree with Dharokan. Races need they own system because with the same system you are DESTROYING the races. Imagine what would happen with the same formula: race with 6 players,some of them see that they can't win, so they start to kill other players to get some points. So the race becomes a deatch match.

And I would do what Dharo thinks about marks.

This is an example of the formula that I thought.

Gain= nb + [marks_marked x (15 for winner, 5 for other players) ]/ total_marks - 1% yourscore / nbplayers - botskilled

About reset of score: reset reset reset reset reset reset.....

1. Do you know that a player that in september of 2006 had 5000 points now he has 500 points?? Come on, after 4 years keeping his points!! Enough gift for them is to keep their accounts after many years without playing!

2. The games are for the active players and for the players that NOW play robostrike. So, benefit us doing the reset!

3. And.... new version + new graphics + new score system = everybody starts with the same points

chris wrote:

Besides, the score is not the only goal. We could maybe include other items in the stars ranking. One star would be for the score but the others could be about community activity, nb of played games, nb of friends, etc ...


About this I thought a thing: What do you think if each player has a reference link to get players? Thus, it could be know how many players each player gets and you could create a ranking about this in which the players with more references would have some specials awards: as be able to create more maps, add more maps to favourites (but I don't know which is the limit now), extra-points, to create some special map (as the old 20x20) and more.

All this would be a motivation to get new players.

dharokan's picture
Last seen: 1 week 1 day ago
Modo
Posts: 779
Points: 783
Just an idea that came into my mind about the reset discussion:

Why not include an automatical loss of 5% score each month, for all players?
That would not be too hard for newbies, imho.

Just an idea ;)
clem759's picture
Last seen: 2 years 35 weeks ago
Modo
Posts: 56
Points: 56
Joining Dharo on this idea, a percentage between 5 and 10 % and also losses enabled after reaching a certain number of points in order to not hurt the newbies.
east amloo's picture
Last seen: 2 years 22 weeks ago
Modo
Posts: 1258
Points: 1260
After one reset, no problem !

We need ONE RESET !!!! aaaaaaaa :mad::thumbsup: