This is an extremely interesting topic to raise, Brad. Introducing a rating system for RPS players is something I'd be in favour of (both personally and as head of the SFD) *if* it's implemented well, and done for the right reasons.
I personally would prefer that if we do it, we base it on something close to the English Chess Federation rating system, rather than the type of ranking system that tennis uses. In tennis, ranking points are effectively just rewards for good results, and while they are semi-useful, they're often far from accurate, especially for players who don't play lots of games each season.
I'll explain how the ECF rating system works, and why I like it (readers who are easily bored might want to skip this bit!):
The ECF system is somewhat similar to the Elo rating system, used in various other sports as well as chess, but the ECF version is much easier to calculate. It works like this:
Any game played in a registered competition (nearly all regional tournaments and leagues, as well as most intra-club tournaments) under standard conditions counts towards a player's rating.
When players win, they score their opponent's rating plus 50 points. (However, if your opponent?s rating differs from yours by more than 40 points, it is taken to be exactly 40 points above (or below) yours. This prevents players improving their rating simply by losing to much better players.)
When players lose, they score their opponent's rating minus 50 points.
At the end of each season, your average score per game is your new rating. (If players have not played a minimum of 30 games in a season, previous seasons' results are included in the calculation, up to a maximum of 30 games or 3 seasons.) Players only need to have played 9 games in the last 3 years for their rating to be published. (Usually the rating is accompanied by a letter to indicate how regularly the player has played, e.g. A = played more than 30 matches in the past season.)
There are some other details, but these are the main points.
I think this system would be a good one for WRPSS to adopt for these reasons:
It's statistically meaningful: If your rating is 125 and your opponent's is 100, you should win 75% of your games against that opponent. (((100 + 50) * 3 + (100 - 50)) / 4 = 125). (We can adjust the precise figures involved when calculating the initial ratings, to ensure that there's a sufficiently wide range, e.g. a 100+ point range rather than a 10 point range).
It's easy to calculate: Players can follow their progress as they go, and accurately estimate what their true current rating is based on their results for the season. This also helps it gain acceptance, as players understand the rating system and can clearly see that it's fair. Also, the simplicity of the calculations will be appreciated by whichever lucky WRPSS Department gets to implement the system and maintain the ratings.
It's suited to club play as well as tournaments: The system works for league matches, one-off tournaments, and other settings. Players can be regular players, or ones who only go to tournaments, the ratings apply to both. (One of the problems with the otherwise excellent www.rpssuperchallenge.com
ladders, in my opinion, is that the rating system is overly biased towards the players who simply play the most games.)
It's fair: It works by rewarding points based on your performance and your opponent's rating. Players don't receive extra 'bonus' points for playing in or doing well in major tournaments. The glory of success at a major tournament is its own reward. Yes, of course, players who do well in tournaments will still see their ratings increase - but this is because they have won several matches against good opponents.
It recognises effort: Someone who's knocked out in the second round will still see their rating increase, if they beat and then lost to two star players, but if they beat a novice and then lost to another beginner, their rating will decrease. That's fair, and addresses the point you made.
It's inclusive: In a tournament, everyone gets rated, from the players who were knocked-out in the first round to the champion. And players don't have to play many matches to gain or maintain their rating. (Though this is where Tri-Cut Tournament format or Swiss-style tournaments could help, by allowing all players to play several matches, helping them reach the requirements more quickly.)
It's extensive: If we went with awarding points only for the major tournaments, this would provide useful rankings of the top players, and allow some scores to be settled (and other rivalries begun), but the rest of the RPS community, like amateur tennis players, would feel left out in the cold. In my opinion, the rating system should extend wide enough to include club tournaments and leagues, and if that's the case, a system like the ECF one would work well.
It respects the best: If you don't take on the top players, you won't be able to work your way up to a top rating. Players who just play in their club tournaments can rise to a certain level, but to reach the pinnacles of the ratings list, you need to play in top competitions, against the best players.
If the WRPSS does decide to roll out a rating system in due course, setting it up would be fairly straightforward. Just collect results for the first season or half-season, and then begin publishing players' ratings based on those results. Or even simpler and quicker, results from the 2004 and 2005 Internationals (and other endorsed or sanctioned tournaments with accurate records for some or all games) could be used to calculate the initial ratings for a significant number of players, so that we 'hit the ground running'. As long as tournament organisers have kept accurate records of players' names and results, this should be fairly straightforward to do.
Updated ratings could be published annually, semi-annually, or after major tournaments. Publishing ratings online on the WRPSS website minimises costs and maximises exposure. Match results could be submitted to the WRPSS either by post or online (probably by giving authorised tournament officials a password to ensure that all results submitted are genuine).
What would it mean for players? Well, players don't *have* to reveal their ratings to their opponents, though it's considered bad form not to. As long as the players record their names on the tournament entry form correctly, the WRPSS will award the points correctly, since we will have records of all players' grades. A player could enter under a false name so as not to lose rating points if he or she performs poorly, but then he or she would miss out on improving his or her rating, so that's probably not a major concern. Throw by throw records are not required, just the match results.
The biggest issue, perhaps, is whether players could collude to artificially inflate their ratings. However, restricting eligibility to officially endorsed or sanctioned tournaments and recognised club competitions should minimise that possibility. We *could* open it up even more by, for example, allowing WRPSS members to submit results of games with other WRPSS members (perhaps if they both email to confirm the result, to avoid disputed results). However, this does open up the opportunity for members to 'throw' games, and though WRPSS members are as honest as the day is long, it's always nighttime somewhere in the world.
The second-biggest issue is whether the ratings *would* in fact be useful. There are two reasons why people might challenge this claim. One is that, "Since RPS is a game of chance, everyone's ratings will even out at around the same level." While it's probably true that there's less range between good and weak players in RPS, there is ample evidence that skill plays an important part, especially at top levels. Furthermore, the scale can be adjusted as mentioned above, to more clearly highlight the differences between mid-range players. And the ratings would provide proof positive that RPS is a game of skill rather than luck, finally silencing all critics of the sport.
The other reason people might question the usefulness of the ratings is that since many players only play a few games each season, their ratings in their first season or two are liable to change rapidly. To address this issue, the letters that accompany the ratings would indicate which ratings are most reliable (more weight being placed on A or B-quality ratings). And since the top-level players play more games, the accuracy of the rating system at the top end would be very high. Also, remember that there is a limit on how many points a player can earn from a match, based on the player's own rating. A low-rated player who beat custardchuk in one match, or even three matches, could not suddenly get a very-high rating, however high custardchuk's rating is (or isn't).
Though there are some issues with how rating systems affect player behaviour, these apply mainly to online games (where players can select their opponents and only take on players they judge to be 'overrated', for example), so this point does not apply to organised tournaments or leagues. Perhaps the most important question in the case of RPS is whether players would come to place too much importance on ratings. Some great RPS Masters do not seek the fame that comes with tournament victories, and instead choose to pass on their wisdom to selected students. These Masters would not appear on the ratings list, so can it truly be said to show who the 'best' players are? And yet, no list could truly do that (suppose one did hypothetically manage to list all RPS players in order of skill; by the time you completed the list, the order would already have changed as some players improve and others lapse). Despite this qualification, the ratings system *would* provide us with an accurate ranking of which competitive RPS players have currently proven themselves to be the best in competitive play (where competitive play = RPS tournaments, leagues etc.).
One remaining issue is which names the ratings would be published under, the players' stage names or their 'real' names? This choice would probably be best left to the players themselves (some may not wish their 'real' names to be used). Though this would likely result in several players being listed under the same name ('Fist of Steel' for example), that shouldn't be a problem as long as the submitted results include the players' 'real' names (confidentially), so that their results are correctly matched with their 'Fist of Steel' persona rather than someone else's.
I can say that the Strategic Forces Department would be happy to manage the ratings calculations, or analyse past International and other tournament results to compile a hypothetical ratings list from those results as proof of concept. (Since completing the R&D on the Tri-Cut tournament structure, we have some number-crunching lackeys available for new projects, and I'm sure the results would make interesting reading.)
Lastly, although the main motive for introducing a ratings system should be the valuable new service to the RPS community that it provides, there's one final point to consider: creating and publishing a fair and full world ranking system for RPS players would ensure that players, public, media and upstart RPS competitions alike will always remember that the WRPSS is *the* home of Rock Paper Scissors.