I have to say I was being very parochial and only considering chess. Games like backgammon with an element of chance mean that the ratings have less significance (I know skill level should mean "more likely to win").
And not all ratings boosters are bad players!
One thing i have found though.. playing honest players the relative rating positions work in regard to win loss ratios....
Last point .. players who occaisionally make really dumb moves are clearly not relying on help!

I think vacation days should suspend any timing clocks just as they do for a game ... and maybe it is just a requirement to move in any game (thereby showing taht you are active) would suffice, but once the time is up (30 days if only a move required) then the decay begins. Of course there could be special circumstances for suspending accounts and ratings but Fencer is such a great bloke I am sure these few cases could be dealt with adequately.

What matters is for the ratings to be relevent so effectively none of the catagories of "cheat" is important as they are inactive, playing themselves or providing practice for the lower ranks. What I think is more important is inflation caused by the floor, to balance this the decay may need to be tied in with the number of different opponents.

OR another suggestion - "flag" ID's that do not finish a game (in 30 or 60 days), and in the rating list - have 3 different settings.

Provisional - established - active (with active just being the people who had finished a game within the last 30 or 60 days)

That way if a player takes a vaacation, or takes a break from the games, he will not lose nothing plus his "true" ratng can still be seen.

Fencer's idea sounds like a good solution! "Hidden" BKR of inactive player's again becomes "Displayed" BKR when, and if, they become active again.

Thought about it some more and I think 90 days should start the decay process and it then continues each 30 days ... and percentages are a better option.

If someone is playing, and they are "bottom feeders" (never heard the term before but I think its great!!!!) then so be it ... they eventually plateau anyway and how interesting can they be! For those who use programs, variants often sort them out as only rudimentary programs exist for things like atomic chess and I have never heard of a game engine for tablut or tank battles. Looks like I am saved from fritz (but not myself :)

how about games like anti backgammom ect these games can take forever to finish a game , i dont really understand the point to all this really , it would mean me playing every type of game on my list all the time or i would lose ratings , it would be ok if i only played a few types of games but i play quite a few ,, if i have got all this right it would discourage (sp) me from starting any new type of game

I'd say 90 days is far too long and agree with the suggestion that games in progress suffice.

Those who have many games going in which both players move fairly often would, of course, have no complaints. However, those who legitimately play few games in which both players move infrequently would be adversly effected. Some games are set at 30 days per move!

BBW: That's very good idea. I can lower the player's BKR but keep the original value for later use, when he finishes another rated game.

Well I personally don't think there is anything that can be done to help against people who use programs - (unless they admit to it - then their rating should be removed or something)

For people that "sit" on their standings when they get so high - well 30 days to complete a game seems like a short time since all it takes is the other player to purposly play slow. But maybe after lets say 90 days... that would make more since. (Or possible rating lower on "paper", but when a game is finished, the old "correct" rating is once again put in place.)

The general discussion is revolving around the issue of accuracy in ratings, standings, and averages. In the end people who use Fritz to mess with things may be more of a problem than multiple i.d.'s, bottom feeders, and those who sit on their standings when they get unrealistically high.

I think that 30 days would be a very short amount of time.

... but I'm not sure what this discussion is really about. Are we worried about players on the rating list who have not played for awhile, or ???? (i'm sort of lost)

I would not want to feel compelled to finish a game per month to protect BKR. This may also hinder players of many game types. Maybe "hiding" inactive players is a good idea. Ratings boosters are probably not worth worrying about unless flagrant and disruptive. Just beat them and live with the few extra points you probably didn't deserve. I would like to think most of us, as Noholdsbarred stated, simply use BKR as a means of engaging opponents of a desired skill level. Someone who only plays lower rated opponents should have that option, if they are that afraid of equal opposition they are probably not THAT good and will drop many points every time they occasionally lose.

its down to one day now ;o)

I like the idea of rating decay but am a little concerned for pawns who may play a number of types of games but overall not many games of each type ... I think the decay period probably needs to be longer and possibly logrithmic, say 100BKR if nothing in 90 days, and additional 200BKR if nothing in next 90 days, 400 for next 90. There would also have to be a consideration of the minimum rating. Another idea would be for the decay to be percentages of BKR ... 90 days 5%, 180 days 10%, 270 15% ...

I think the details still need to be worked through but I like it!

I think its a good idea, Filip, but that would mean possible problems with high rated pawns, who have slow moving games, there is a slight chance they wont have a game finish in one month.
Yet, I suppose that would be a good incentive to upgrade and have more games and so to have at least one game finish.

Also maybe require that games be under way rather than completed as a person's opponent can stall, use vacation days and dictate the pace of the game if they want to.

I would say that a great idea Filip. How would you get round the problem of top rated players just taking on the very lower rated ones though ?

What about to subtract 100 points from BKR of user who didn't finish a rated game within a month? [or 30 days]

Rating decay is a great idea.

With a new game I can think of three methods,one is to use your beta testers (established players) to set the baseline. Another to use the system as is until a base is established. A third would be to log results for the startup period and run an offline analysis to baseline initial BKR's.
Going back a bit I also think Ratings should have a half life (decay) to make people work to keep them and to offset the influx (i believe the current algorythm has a positive bias) if not then it should have.

Or we could just bribe you for BKR :)

<So that when two unrated players compete (as any game would be on a new game) both player's bkrs could be affected. A couple two-game tournaments would soon supply a base of established ratings. :>

It could still be the same as now Fencer, just only affecting the unrated players hidden bkr and not the established players one.

noholdsbarred: Interesting idea, but when I add a new game [so nobody has a rating in it], how could anybody get a rating if only established ratings are counted for calculations?

I agree, maybe that suggestion noholds, should be brought up on the feature requests board now :o)

That might not actually be a bad idea.

Excel lies....no account has been taken of the players who played lost a few then departed fbefore ever entering the lists...
I wonder if changing the system such that unrated players did not affect the ratings of the people they played until they had established a rating for themselves. At the very least it would discourage the one day ratings boosting campaigns since they would have to compete and complete against 5 already rated players before being able to have an effect.
The time taken to establish such a net would surely discourage the more casual offender.
It would also encourage even the most sensitive ratings protector to play unrated players... thus giving them a genuine start value.

You must remember too, that the MIDDLE bkr, isn't necesarilly the AVERAGE bkr.., but probably pretty close..

Oh, I see Kevin figured out the average for chess.... and in this case, I guess it's pretty close..

I would also add, that the average is most likely different for most games. Depending on the number of skilled players as opposed to novices that are here on this site.

I just put the chess ratings on Excel, and came up with this:

Total rating points: 407834
Average Rating (252 players): 1618.388889
Total rated games played: 7227
Average games per player: 28.67857143

Not steve at backgammon :):):) Just ask the angels lol

I know there are different averages mode median and medium .I just wondered what we classed as an average player:o)

I find most of my ratings on this site are significantly higher than elsewhere, but that doesn't verify the inaccuracy of the system.

And you'd think there'd be more people towards the bottom of the average (especially provisional ratings) than above, so the median is not necessarily even close to the average.

I think the rating system has been skewed by "ratings boosting". However I think that this has been spread across the board so the average has lifted.
I know that at least one player on my current list has an artificially enhanced rating. Gained against players who never stayed on long enough to gain a rating their 1300 has been added to the pool without their being added to the divisor (rated players list).

I agree above 1600 to me seems to be above average.
I dont know about the FIDE system

Of the chess players with an established BKR, #125, out of approximately 250 total, is rated at 1675. This seems a little too high to be an accurate average and a greater sample number will probably prove this out. Your 1500 estimate may be close to an accurate average although I believe FIDE is about 1300. Maybe Brainking is a little more forgiving.

I was thinking the 1300 would have been for a reason. And it is around half way up the scale too. But 1500 looked more like a good average player number to me

Steve R : I would think that most players fall into the 1300's which would account for the provisional BKR. There are, for instance, approximately 950 rated chess players with player #475 rated at about 1400.

What BKR value would we take as an average players amout, Would you say half way between 100 and the max, or is there a number that is just above or below that.
Or has this never been looked into?

We have been looking for a worthy pawn and think that we have found one that truly needs some help. This lady is the single mom of a small child and a membership would be a luxury that she really can't justify at this time with her fixed income. We are looking for someone who would be willing to help bring a little happiness into this lady's life.

She has played here since November 2002.
Has completed 281 games
Has 16 current games
1 tournament

If you are interested in participating in this program and sponsoring this lady please send me a personal message. Thank-you

Thanks Harley! Thanks Lamby!

You can see the interviews so far at BrainKing.info.

Cole, when the interview is complete, a link is usually posted on this board

Where do I go to read the interview?

<Why all the cringing? I'm rather looking forward to that interview. :>

yep I agree, It took almost 3 times longer than normal to access my profile(which is slow to access to start with)

its been slow here at times depends what im trying to acces i think

I have had many ntl probs recent months, was not too sure if it was them again

It depends. You know, database server performance goes up and down, up and down, up and down ;-) It needs more RAM.