At this time, this page is for the purpose of working out a valid Ladder scoring system. In the future, it will probably be used as part of the detailed documentation of any such system that gets implemented.
Several implementations are being tracked here.
The first implementation of this ladder system was in 1998 by ??? for the purposes of tracking the overall player skill and reliability on games played at ???. The Skill Points system was designed to resemble the US Chess Federation (USCF) scoring system ( https://en.wikipedia.org/wiki/Elo_rating_system ).
This system worked as follows:
https://groups.google.com/forum/#!topic/alt.test/a6xFQTQCKDA
In March of 2013, Big Beefer implemented a ladder system, similar to the RCC ladder system, to score the Die Hard games.
http://planets.nu/discussion/die-hard-ranking-ladder
https://planets.nu/_library/diehard/diehard.html (click on Ladder at the top of the page)
https://planets.nu/_library/2013/4/allgames.htm
In July of 2014, Dotman created an early prototype Ladder scoring system for the Planets.nu site. That system was closely based on the RCC implementation. It had all the strengths and weaknesses of that system, but it worked with 11-player non-Campaign games from Planets.nu.
One significant change is that players who resigned or were dropped were not put at the very end of the list. They were put in the lowest unoccupied slot.
https://planets.nu/#/activity/1427142
https://planets.nu/_library/2014/7/nuelowtrans.html
In April 2015, Azzazzello started the ball rolling for a ladder-based secondary scoring system for Planets.nu. The evolution of that system is what will be discussed in the remainder of this document.
http://planets.nu/discussion/i-am-starting-a-campaign-to-lobby-the-gods-of-this-site-to
https://planets.uservoice.com/forums/136520-general/suggestions/7487350-implement-big-beefer-s-die-hard-ladder-system-for
http://planets.nu/discussion/ladder-system-update-this-is-a-followon-to-the-ladder
The primary scoring system at Planets.nu is an achievement-based system, and is purely additive. While it was designed to reward Quality of play, it is also possible to achieve a very high score with Quantity of play. A ladder-based system almost completely removes Quantity from the equation. While it is additive, it is also possible to lose score.
This is a Work in Progress. Everything here is subject to change. Information is placed here from the thread when there are a majority of posters who agree to it. In addition, when something is determined to be needed, all the prerequisites for that thing are added.
The core issue with the Planets.nu scoring system is that it's purely additive, with nothing to counteract the accumulation of points. This means that quantity of play is rewarded as well as quality of play. In order to most accurately reflect the quality of play, it is necessary to have accumulation of both positive and negative values, in equal amounts. By using a zero-sum scoring system, the problem where scores increase without limits is virtually eliminated.
By also providing an Average Strength Modifier, the ability to have scores that increase without limit is completely eliminated. This would be very difficult to implement in a system that's not zero-sum.
The following per-game information is required for each game to be added to the Ladder score:
In addition, for each player slot of each game, the following information is needed:
The majority of people understand that every game possible must be included in the Ladder score. This is mainly due to the relatively low number of games that most people play. It takes a certain number of games before the score is valid, and anything that extends that time is counterproductive.
Because of this, it is necessary to determine ways to merge different kinds of game into a single score.
At this time, the games to be considered are as follows:
| Game Type | To Be Considered? |
|---|---|
| Beginner (all except Mentor vs. Midshipmen) | Yes |
| Campaign Classic | Yes |
| Campaign Melee | Yes |
| Campaign Standard | Yes |
| Championship | ??? |
| Classic | Yes |
| Melee | Yes |
| Mentor and Midshipmen | No |
| Mentor vs. Midshipmen | No |
| Player-generated Public | If 8 or more non-AI player slots |
| Player-generated Private | No |
| Senior Officer Classic | Yes |
| Senior Officer Melee | Yes |
| Senior Officer Standard | Yes |
| Standard | Yes |
| Team | Yes |
| Training | No |
The scoring is a simple accumulation of the subscores. Old games are not automatically aged off from a player's score.
There are two subscores in every game score, as follows:
Placement Score — This is the primary scoring mechanism. Its purpose is to indicate where a player places in the finish.
Average Strength Modifier — This is a secondary scoring mechanism. Its purpose is to indicate the skill of the opposition relative to the player's skill.
There are several algorithms that are needed to implement the Ladder scoring system.
The largest change that needs to be made from the previous implementations is that a different method must be used to generate the Placement Score (called the Rank Modifier in RCC). This is primarily due to the ability to have scored games with 8-35 players. Extending the RCC Rank Modifier table from 11 players to 35 players would yield a 1st-place score well in excess of 250,000 points, which would completely overshadow any possible scores from 11-player games.
It has been suggested that the scores be based on an equation. This is the method that is currently planned to be used. Because there is so little demonstrated difference between last place and next-to-last place, the huge last place penalty that the RCC scores imposed will be replaced by a more gradual curve. At this time, the equation to be used has not been determined.
Pro - This is fairly simple and flexible WRT the number of players.
Con - This requires a new score equation, as no simple (3rd order or lower) score equation has been found that correlates to the RCC scores. As the RCC scores excessively penalize the last place finisher, this isn't all that big of a problem, but it is an additional task that needs to be done.
It has been suggested that the players placing below 11th place all receive the score of 11th place. This has been refused.
Pro - This is simple, flexible WRT the number of players and uses the same scores that the RCC Ladder system used.
Con - This will move the scoring away from a zero-sum score. In this case, it will have a tendency to decrease the average score below the starting point.
Con - This means that the system sees no skill difference between last place and 12th place in a 30-player Giant Melee.
As it turns out, the Events log in a game is rather difficult to parse, and is missing critical data. Because of this, the sections below won't work, and a different method of placing these early exiting players had to be created and documented. This section is that documentation.
When a player exits the game prior to the end of the game, they will be placed into the lowest open slot in the score system. If multiple players exit the game in the same turn, they will be sorted by planet count (military score for games where that's the win condition), with the player having the lowest being placed first, and the player having the highest planet count placed last. In the event of a tie, the effected players will occupy a single slot.
When a player leaves the game, either through complete elimination, or the action of the Fight or Fail mechanism, they will receive the lowest open slot in the score system. Multiple players who leave this way in the same turn will consume a single slot, and will all receive the score for that slot.
Due to data availability and formatting, this can not be done. The replacement method is described in the "Player Early Exits" section above.
When a player is dropped for missing three turns, they will receive the lowest finishing slot that is not yet taken. This will be done after player exits due to elimination and FoF have been placed. Multiple players who leave this way in the same turn will consume a single slot, and will all receive the score for that slot.
Due to data availability and formatting, this can not be done. The replacement method is described in the "Player Early Exits" section above.
When a player resigns from a game, they will receive the lowest finishing slot that is not yet taken. This will be done after Dropped Players have been placed. Multiple players who leave this way in the same turn will consume a single slot, and will all receive the score for that slot.
Due to data availability and formatting, this can not be done. The replacement method is described in the "Player Early Exits" section above.
When a replacement player comes into a game, the score table will expand by one. If, at any time, the replacement player is dropped for missing turns, they will receive the drop score, but only if it is negative after all modifiers have been applied. Allowing a positive score change for a player dropping would allow replacement players to abuse the system.
If the replacement player resigns, or if the game completes before 20 turns have elapsed, they will receive no score, and the created slot will be removed.
If the player spends more than 20 turns in the game, they will be eligible to receive a positive score when they exit, if they play to the end of the game. If the replacement player's score would be 0 or negative, the added slot will be removed and the replacement player will receive no score.
As the replacement player has done a service to the game, they will not receive a negative score after adjustments, unless they are dropped from the game (a counteracting disservice).
The only known method to "game" this system is to play large numbers of games as a replacement player, preferably taking over positions that are well placed. It would be good if we could find a way to avoid this hole.
UNDECIDED
A returning player is one who exited the game (usually by being dropped) and wishes to return. The Planets.nu system will only allow a player to return to the slot they had previously.
The current suggestion is that if they missed any turns, they will count as a resignation or drop, followed by a return. If no turns were missed (rejoin on same or next turn), they will be considered as having a single contiguous play.
UNDECIDED
If an alliance wins, all members of the alliance receive the top positions. The total points of these positions are averaged, and all members of the alliance receive that average.
In a Team game, with the exception of the winning team (handled above), all surviving members are ranked by their score (planet count or military score, depending on win conditions), then the scores of those members are averaged. As above, all surviving members of the team receive the average.
This is what the RCC implementation called the Difficulty Modifier. It is a better term for use at Planets.nu, as there's already a Difficulty Modifier.
The equation for this can be used as is. Some experimentation may be done to determine whether or not the 10% constant works best.
Refused
Incorrect information. Looking into either removing this or redefining it based on something other than the Planets.nu Difficulty Modifier.
The Planets.nu Difficulty Modifier is a single number that is supposed to represent the relative difficulty of play for different sets of game generation parameters. This score will be added to the player's score, with the last-place player losing the bonus, the winner(s) receiving the full bonus, and the rest of the players receiving or losing a pro-rated portion of the bonus, depending on their finishing position.
As the environment impacts all players equally, the Planets.nu Difficulty Modifier can't be used, and creating an Environmental Difficulty Modifier is beyond the scope of this project, it has been decided to not implement this. Too much work for too little return.
Refused
As an attempt to compensate replacement players for the benefits they provide to the system, it has been suggested that each time a player completes 5 games as a replacement player, having played at least 20 turns, their oldest negative score be removed.
This is a violation of the "zero-sum" philosophy of a true ladder scoring system. Any imbalance caused by this benefit should be put into a pool, so that players who are dropped from games can receive an appropriate penalty.
It is believed that the placement from the player leaving the game will be adequate bonus for the replacement player.
UNDECIDED
As it is expected that most players will improve with time, it would be beneficial to the system is there were a way to remove old negative scores. It has been suggested that each time an original player completes 10 games, their oldest negative score be removed.
This is a violation of the zero-sum philosophy of a true ladder scoring system. Any imbalance caused by this benefit will have to be monitored to make sure that it isn't causing excessive the system to become unstable. We may need to find a penalty to counteract this.
UNDECIDED
One problem that a Ladder scoring system shares with an additive scoring system is that players can use it to get to (or near) the top of the ladder, then stop playing, leaving a high slot occupied by an inactive player. In order to keep this from occurring, it has been suggested that the scores of inactive players be aged.
The aging method that has been suggested is that their score have a half-life of 2 years. This means that 2 years of inactivity will cause the difference between their score, and the initial score, to be halved. The mechanism to do this is to multiply the earned score by 1053/1054. This should be only be done for players with a positive earned score, as doing this with players who have a negative earned score would improve their score.
The suggested definition of an active player is one who is enrolled in a game that requires regular turns (any game other than a training game), or takes at least one turn per week in a training game. It has also been suggested that the score aging not start until at least 6 months if inactive time have elapsed. It may be appropriate to expand this definition.
This process is similar to the achievement score aging that was implemented with the new Badge system.
Refused
It has been suggested that top-rank players who are not in a Championship game, and refuse to play in one when offered, are undeserving of the top-rank position, and should receive a penalty when this occurs. The suggested penalty is 10-20% of each of their Ladder scores.
This has been refused by the players. Various reasons were cited, the strongest being that consideration of Real-Life situations is something that only the player can make, and they should not be penalized for having a life outside of Planets.nu. Other reasons are that the player may already be in as many as they feel they can handle, or that the player might not like that particular type of game. These are all valid reasons for refusing a Championship game.
UNDECIDED
A pre-game alliance has become a common occurrence at Planets.nu. Such an alliance gives the players a wider view of the game then they would have without such an alliance, thus improving their position in the game.
It has been suggested that the majority of alliances that are started with "Share Intel" prior to turn (TBD) are the result of a pre-game alliance agreement. Because of this, it has been suggested that a Share Intel (or higher) status that is given prior to turn (TBD) be penalized by one point per turn, for each such status. This penalty would, obviously, not include members of a team.
The turn to be used would be such that 99.99% of the Share Intel events between turn 3 and turn 20 happen after the selected turn.