The current script that I used was wrote by Prom, it is broke with all the changes to t-edit and since I am not running public games currently I just had no desire to fix it. But I am sure you could easily write a script that could have options for the items you want to score.

Couple of points here. One, I would like to add some new rating systems to support a variety of game types, as we've talked about in the feature request forum. Two, I'd definitely like to consult with ops like T0yman who've written their own scoring system, to build on their efforts.

And three, T0yman, I'm curious what kinds of things have broken scripts with my TEDIT changes. There are some changes that are major, like removing the edit screen for toggling flags (like active/inactive, dampened, etc) and replacing them with individual toggle fields. Those are going to break scripts. But there are a lot of cosmetic changes, just reorganizing the screens and cleaning them up, and it was my hope that keeping the text of the prompts the same would be enough to avoid breaking scripts, even if I moved an option to a different place on the screen. I'd just like to know how things broke so I can consider if there are any ways for me to achieve my goals for cleaning up TEDIT without causing any unnecessary disruption to gameop's scripts.

Reorganizing the screen placement could cause some issues with the scoring script if I remember it right. I believe I keyed off text that would then allow me to grab what I was wanting. For some reason in TEdit we have to use waitfors that we normally don't have to in normal game scripting. I doubt the script would be hard to fix, just take time to figure out what it is wanting, but I am rewriting all of my game scripts at the moment. If someone is needing it, I can work on it.

               / Promethius / Enigma / Wolfen /

"A man who has no skills can be taught, a man who has no honor has nothing."

One of my goals is to provide a way to read and write game data and settings directly through a simple text interface, without having to script the editors themselves. With something like that in place, nothing I change in TEDIT will effect server-side scripts, and I can focus on optimizing them for human use. Plus, I think a script interface to TEDIT will greatly expand what you can do with server-side scripts.

That would be very nice. Moving through Tedit is very slow and the rankings took a long time to run. I haven't ran that script in probably a couple of years. I think I can post the scoring values used as they seemed to work out pretty well. At least give people something to think about.

k, kind of nasty as it is a code snippet, but easily figured out:

               / Promethius / Enigma / Wolfen /

"A man who has no skills can be taught, a man who has no honor has nothing."

Different game types would each need their own scoring system, and different objectives. i.e. Classic, Money Run, Capture the Planet, Colonization, etc...

Each game type would need it's own point system. The point system should be flexable for custom game design. i.e. points could be awarded for - Experience, Alignment, Kills Minus Deaths, Planets Created/Owned, Planets Captured, Deployed Fighters, Fighters on Planets, Product on Planets, Cash in Citadels. It should be posiable to score planets by planet type and citadel level, and the Death Peanality can be set independent of the Kill Points.

Actually I think points should have a set value. Reason is, I hate to say it, you may have some gameops who create games for extra points just to boost the players on their site to the top. This is assuming we tie all the sites together for a "global" score. Each board needs to be the same.

I think the League would need to have a small subset of popular TWAs with predefinded point systems for each style of game. You wouldn't want to award points for Kills in a building type game, and you wouldn't want to award points for planets in a classic domination style game.

Player rankings is more about the winners and losers, than the actual point system, or style of game. Each game will needs its own rules, objectives and point system, but actual player ranking is all about winning or losing regardless of the game details. The Reverend designed the Rating System for twleague.com, and he sent me a copy. It is a fairly complex mathematically system based on Chess rating rules. In brief, a players rating will go up if they win, and the rating will down if they lose, while a runner up's rating will remain the same. The system is designed to reward winners more than it punishes losers, and the winner's rating will go up more if they beat someone with an equal or higher Rating.

Really most of the scoring was fairly quick with one exception, sector figs took forever since it had to run through all sectors to collect the data. I run into the same problem with my Sector Fig Counter/Sweeper script, counting 30K sectors takes about 30~40 minutes.

About the Rating System

The rating system is designed to be mathematically stable. All new players start out with a rating of 100. The overall average rating will always be 100. This means that the rating for any given player will always be comparable to the rest of the players in the system. Even if player A does not play for a year and all other players advance or decline during that period, player A's rating will still be comparable to everyone else's in a meaningful way.

The rating system is designed to reward underdogs for winning. If corp A has an average rating of 90 and the rest of the field of players has an average of 150, then corp A would receive more points for winning the game because they are not favored to win. Also, if the max corp size is 6 and corp B wins with only 2 players, then corp B will receive more points because they had a small corp. Conversely, better players will receive less points for beating lesser players and will lose more points for losing to lesser players.

The rating system is designed to award more points for better attended games. If a game has 50 players signed up, it will be worth more than a game with 5 players. This concept extends to corps also, so for example, if game A has a max corp size of 2 and 60 players, then it has an average of 30 corps. Game A will be worth more than game B with max corp size of 6 and 60 players, having an average of 10 corps. So given two games with the same attendance, the game with a smaller max corp size is worth more because there are more corps vying for the win.

The rating system is designed to divide winning and losing points according to each player's rating. Higher rated players will lose more points for losing and win less points for winning than lower rated players. Lower rated players will win more points for winning and lose less points for losing than higher rated players.

The rating system is designed to reward winners more than it punishes losers. For example, if there are 10 players in a game and 4 win, the winners may earn around 3 points each, while the losers will only lose around 2 points each. The total points awarded is equal to the total points taken.

The rating system is designed to recognize runners up. Players on the winning team that are eliminated are considered runners up. Players on the second place team that are still alive are considered runners up. Runners up receive 0 points for the game, i.e. their rating will not change for losing. If there are less players than four times the max corp size, then runner up rules do not apply for second place players.

The rating system can be exploited in certain situations. First, the number of losers must be greater than or equal to the number of winners for the rating system to make sense, i.e. each loser could lose more points than each winner. For this reason, games with less players than twice the max corp size should not be included in the system (i.e. there should be about two full corps in the game). Second, if a game has a max corp size of 6, but all the corps in the game are solo players, then the rating system would reward the winner as if he had won as a solo player against 6-man corps - a huge bonus. In this situation, the max corp size should be considered as 1 because all the corps in the game were solo players. Sysops should be aware of these types of exploits when designing their games.

The Equations

NP = Number Of League Players In Game On Day 7
MCS = Max Corp Size
WCS = Winning Corp Size
NF = Normalize Factor
NCF = Normalize Corp Factor
NW = Number Of Winning League Players
NL = Number Of Losing League Players (NL >= NW)
WR = Individual Winner's Rating
LR = Individual Loser's Rating
RFL = Rating Floor
SWR = Sum Of Winners Ratings (using RFL)
SLR = Sum Of Losers Ratings (using RFL)
AWR = Average Winners Rating (using RFL)
ALR = Average Losers Rating (using RFL)

constants
NF = 2
NCF = 6 <-- new factor to normalize for the "solo game bonus"
RFL = 20

corp size factor
CSF = WCS^2/(MCS*NF) <-- old csf
CSF = (NF / NCF) * (MCS^2 / WCS^2) <-- modified corp size factor

the game is worth TP points
TP = NL * NW / CSF <-- old equation
TP = NL * NW * CSF <-- modified total points equ
TP' = ALR / AWR * TP

each winner earns W points
W = TP' * (WR / SWR)

each loser loses L points
L = TP' * (LR / SLR)

RUNNER UP RULES
1st 1
1st #SD# 2
2nd 3
2nd #SD# 4
lost 5

Category 2 and 3 players are not included in the calculations.
Category 2 and 3 players don't gain or lose any points.
Category 2 players still get a win
Category 3 players still get a lose
Category 3 and 4 players will be considered category 5 players
if there are less than 4 X MCS league players in the game.

the rating system is based on the chess rating system, modified for more than one opponent and corp size considerations. it's mathematically stable so that player rankings are always comparable over time. there are a few exploits that can be easily avoided, as described above.

That's outstanding work, Rev. When I coached a high school chess team in 2000-2001, I implemented the Elo system for a program I wrote to track the ratings of our students. I've considered ways to apply it to tournament games, including TradeWars, but have always been caught up on how to expand it to a system as complex as TradeWars, with its multiple teams, corps, etc. Did you work from another source that expanded Elo to support multiple players/teams, or did you do this work entirely on your own?

i did it on my own. hehe. it went through a few revisions to accomodate multiple opponents and team sizes. i would be interested in your feedback if you find it interesting. i've been away for it for a while so i'd have to figure it out again if you wanted to discuss it in any depth.

note: use the "modified" equations above... dunno why i kept the old equations in the text except for historical clarity.

I definitely want to take a deeper look at it. Unfortunately, now isn't the time. It's on my list of proposed features, though, so I will follow up on it, sooner rather than later if people vote it at a high priority (once I add it to the voting list). Right now I'm focused on the truce mode rules and general bug fixes.