Up to this point, I've put all my PRS effort into stopping the
proposed PRS by offering "no PRS" as an alternative. This seems no
longer constructive as Clint has clearly indicated he intends to
implement the proposed PRS despite the "vocal minority (named Darrell
Shimel)" opposed to it.
So, how about I take a shot at designing a better PRS that he may be
willing to go for instead.
Things a ranking system should reward:
Win %
Experience
Picking up dropped positions
Neutrals that decide early
Neutrals that help game balance
Things it should punish:
Transferring to a teammate
Hard drop where MiddleEarthGames has to find a replacement
It most also give new players a "decent shot" of eventually moving
toward the top.
I'd add that it shouldn't require a code change (such as a different
Victory Point system or different Victory Conditions), shouldn't be
labor intensive, shouldn't be complex or hard to understand,
shouldn't rank a selfish player above a good team player, doesn't
require player votes, and requires no subjective decisions by ME
Games.
To accomplish all of these things I'd use Win % and a "positions
played" adjuster.
WINNING %
Win % is simple. Track the last 10 games "completed" for each
player. Wins count as 1, losses 0, transfers -1, drops 2. Divide
the sum by games completed (1-10).
If you're playing 2 positions at game end and the team wins, then it
counts as 2 games completed, 1 point for each "game". This
encourages picking up teammates positions before they hard drop.
If you're playing 2 positions at game end and lose, it only counts as
one loss. Exception is if you started the game with both positions.
Then it counts as 2 losses. This is to encourage you to pick up
teammates positions, even if it looks like you're losing.
If you're a neutral, the game doesn't count at all unless you decide
by GT10, and are one of the first 3 to join the side. This
encourages early decision and game balancing decisions. Exception
would be FA1000 where all nations still neutral on GT10 would be
considered on the "neutral alliance" meaning the game is a win if the
neutrals win, but a loss if FP or DS win. Jumping to FP or DS after
GT10 means you could get counted as a loss, even if the alliance you
join wins.
Notice that you can have a negative winning % if you drop or transfer
positions.
POSITIONS PLAYED ADJUSTER
Notice that the simple winning % accomplishes most of the things I
think a PRS should do. However, it fails to account for experience
in the ranking.
To rank players with more experience ahead of those with less while o
giving newer players a decent shot of moving up over time, Winning %
would need to be modified by an time-weighted, positions played
adjuster.
I'd track "positions played" by season, and keep 5 years worth of
seasons (20). At the start of each season, count how many positions
each player is currently playing and then add 1 every time they start
or pick up a position within the season. One-week games count as 2
for positions played since you're processing twice as many turns
during the season.
To get the win% adjuster, sum up the products of each of the 20
quarters and (20-seasons ago)/20.
If a player is always in 1 game at a time (starts a new game the
season after the last ended), the adjuster would be (1 * 20/20) +(1 *
19/20) + (1 * 18/20)... + (1*1/20) for a total of 10.5.
Twice as many games gets double the adjuster.
A player that has been in 1 game per season but has only been playing
2.5 years (10 seasons) would have an adjuster of (1 * 20/20) +(1 *
19/20) + (1 * 18/20)... + (1*10/20) = 8.25. That is about 75% as
much as a player that had been playing 1 game per season for the full
5 years.
There is no way a new player could join 10 games in one season to be
considered as good as someone that has been playing 1 game at a time
for 5 years. However, a year of playing 3 games at a time, and the
new player could get an adjuster as high as a player that has been in
1 game at a time for 5 years.
PLAYER RATING/RANKING
So, we take each player's win % and multiply it by the positions
played adjuster to get the players rating. Sort by rating, and you
get ranking.
Three games at a time for 2.5 years (24.75 adjuster) with a 60 win%=
1485
Two games at a time for 5 years (20.5 adjuster) with a 70 win% =1435
One game at a time for 5 years (10.5 adjuster) with a 50 win% = 525
New player in first game = 0
Player that drops more than wins = -ranking
Same new player as above, plays 2 games for a year (7.4 adjuster)
wins first game (100 %) and jumps to 740 rating.
I believe this system would be MUCH better than the currently
proposed system. It rewards all the things that should be rewarded.
It punishes all the things that should be punished. It would only
require mod input at the change of season, at the start and end of
each game and when someone drops or transfers. Well, they'd also
have to do a bit of work on GT10 to see if there are any undeclared
neutrals and to see if more than 3 went one way (then would have to
figure out which were the first 3). So, a bit of work every game 10.
Best of all, it would give real meaning to things like "experienced
grudge team" or "fairly new player". ME Games could create "newbie"
games only playable by people with low positions played adjusters,
and "crusty veterans" games that can only be played in by people with
high adjusters. They'd be able to better fit grudge team opposition.
They'd be able to better match players to position dificulty. They'd
be able to balance open games. "Hey Joe Newbie, too many new players
on the FP team already.... Are you sure we can't talk you into a DS
position to get you a few teammates with some experience?"
It wouldn't be hard to start up. You'd ignore all previous drops and
transfers, as people were not given fair warning of it counting as a
negative (and data probably isn't available). Also, all neutrals
that were on the winning side would get credit for the win. All I'd
need to know is when games ended, who was playing at the end of each
game, how many turns the game ran for, and whether it was a one-week
or two-week game.
Given this data, and my programming experience (15 years in software
development), I bet I could get current rankings within a few days.
With a couple weeks, I could create a MS DOS or MS Windows app that
would allow the mods to maintain the data with less than 10 minutes
per game start, game end, player drop/transfer, or turn 10 processed.
If Clint is going to implement a PRS, I hope he'll wait long enough
to get some feedback on this proposed PRS system.
Thoughts?
Darrell Shimel
Anyway