Hello Iconoclast,

well I agree this would be different from what the system is today, but if you look at tennis for example, where there the ranking is assessed according a number of tournament each year (grand slam tournaments, master tournaments, less prestigious tournaments), the ranking is computed by adding the performance done by the player during the last year.
And the result is that the rating is a good way to rank players and to predict the next results.

I precise that I am not playing tennis at all
I find it somewhat strange that with the current system based on ELO points you can stop playing for several years and keep your rating unchanged. So to answer your question I do not know if your ranking should become zero, but to me it should decrease for sure.
Anyway, I am not a statistician and have unfortunately a limited insight into these questions.

Decide what you want to know, then and only then can you say what a rating system should look like. So, Grumly, what do ya want?

There is a point at which a rating should not be published.

There is a point at which a rating should not be published.

Expand this statement please.

For example, if a player is not playing for period X, and is not paying federation levied fees, then the task of maintaining visibility of his rating in the public domain is a service that is not being paid for. Therefore, his rating should no longer get published.
That is the way of the commercial world.
...
There is no such thing as cost-free maintenance, thus there should be no free-publication for inactive or unfinancial players.

I cannot agree. Cost of having data (a rating number) stored and reduced by x% a year is zero. The costs are for the entry of data and its distribution. Therefore, a player should be charged to have his games rated and organisations that want to see/use the rating should be charged for the information - and the rating should not disappear.

[Bloody economic rationalists usually espouse logic-free and dollars-for-me plans. Kill 'em all!]

This shallow view that you adopt, that the numbers themselves come for near-zero cost, is disappointing given the knowledge you obviously have of the ACF rating procedures.
Storage of data is not the main cost element.

In fact, the main cost element is the expenditure of the ACF personnel who meet on teleconferences to manage ratings and related procedural infrastructure.

It well documented by you that you have low respect for some of this, but the money is spent; that is reality. The money is a cost; the cost needs to be recovered.

Giving away the product for free, for inactive players, is not good business.

That truism is the way of the world; capitalism.

It would be good if you turned your attention to the significant issues I raised in my first post.

Again I am not saying the solution I propose is perfect and I am sure it would be flawed in other ways. My point is just to say, and I agree with you Iconoclast that there are many valid ways to rank people and that some other solutions have been tried (in tennis for my example) with good results as well.

Trevor, Im not sure what tournament "this list" is from, or what rating system it is using - lets have a look at the top of the field...

1 Cannon, David j1 1263 7
2 Raine, Marcus a 1714 6
3 Chew Lee, Max j1 1294 5.5
4-9 Chmiel, Rad a 1423 5
Bergmanis, Olgerts a 1238 5
Lawson, Shane a 1258 5
Flude, David A a 1396 5
Guo, Zhi Lin j3f 612 5
Croft, Roger a 1340 5
10-13 Tsai, Peter a 1190 4.5
Gluzman, Sam j1 961 4.5
Chan, Kris j3 585 4.5
Lee, Isaac j2 623 4.5
14-24 Ying, Jimmy j1 1064 4
Lee, Dmitry j1 987 4
Guo, Steven j1 1005 4
Foo, May-Yi j2f 782 4
D'Onofrio, Gus a 1283 4
Lau, Andrew E j2 641 4
Lai, William j2 733 4
Reid, Euan j3 570 4
Chew Lee, Adrian a 821 4
Lee, Ewan j5 201 4
Ebeyan, Thomas j5 171 4

Rad Chmiel is 1423??? He is heaps better than that - his ACF and FIDE rating are much higher than this.

Olgerts Bergmanis at 1238 is a joke - he is about 1500 even now and was once over 2000 - how he is an 800 point weaker player now is beyond me.... Lawson at 1258? He is over 1500 strength easily - his ACF rating says so, as does his play (Ive played him numerous times) Flude (another one Ive played before) is heaps better than 1396 - even on a bad day. Of the players I know well enough, only Raine at 1714 even resembles a sensible rating - and that is Raine on a bad day - he is usually much better than that - often over 2000. In this event he even scored 6/7....

I dont know much about David Cannon - but if he scored 7/7 in this field, how on earth is 1263 an accurate rating??? Since when does a player rated 1200 win open events with a picket fence score?

Ive only listed the players who finished with over 50% in this tournament, but have a look and you will see that there are players rated 171 and 201 finishing with 4/7! Some of the worst chess players Ive ever played were rated 900-1200 on the ACF rating list and they would have no hope of scoring 4/7 in this event. If these players (171 and 201) are 800 points weaker than them then they would hardly know how the pieces move!

What I think has happened here, is that 171 and 201 are not really that level at all - but much stronger. The problem is though, that kids improve quickly and take rating points off people who have higher ratings than them. For example, a 1400 player plays a number of kids rated 900 (but who are playing at 1500 strength) - the 1400 then loses rating points and becomes 1100 (even though he is still 1400 strength). Then in the next tournament he goes on to beat a number of 1300s (all playing at 1300) and they lose rating points too - and it brings the whole rating system down with everyone losing rating points or catching up by taking them off other people who are playing no worse than before.... But that isnt all folks! The aforementioned 900 rated kids (who are really 1500 strength) play new unrated kids who are probably 1000 strength - and of course the 900s all win and the new unrated ones get ratings of 400 etc. They then play other unrateds and beat them and they end up with 200 ratings.... Pretty soon we are going to have negative ratings!

There will be people out there who read this and dont agree with me, and thats ok, but what Ive just written is basically what a hell of a lot of people think.

There will also be people who say that such and such rating system is mathematically perfect - but so what? What that doesnt take into account, is the following:

1. People improve - they get better quickly yet people lose points when losing to them when they themselves havnt played any worse than before

2. People dont get worse at chess. Even after 10 years away from the game, Im no worse as a player (Im no better either) but I know what I knew back then and I dont "not know it" or "not apply it" now.

3. This rating system doesnt appreciate one main point: That it is much harder to win a game of chess than to lose one. To lose a game of chess, all you have to do is resign - its that easy. If that is too much effort even, just do nothing and let your time expire.... To win a game of chess (against anyone), you need to outplay someone who is trying just as hard to beat you as you are to beat them - not an easy feat. First you have to get an advantage and then you have to put them away.... Its a hard thing to do. Yet, despite this, we give equal time to wins and losses. If winning a game means +10, does that mean that losing a game means -10? Well, its a lot harder to win one than lose one... It really shouldnt be that equal...

Basically to cut the long story short, when you have heaps and heaps of improving kids playing in an event, all with ridiculously low ratings, they bring everyone elses ratings down with them while still staying too low themselves... Then, people stop taking the rating system seriously and it basically has no legitimacy in the eyes of most of the chess scene and people start calling for changes or simply running alternative rating lists.

What to do about it? Well there are a few ideas - Davids idea of a floor of 1000 or 900 or something seems ok, as does another idea I once heard of starting all unrateds at 1200 or 1500 - then people take points off them which counterbalances all the points they lose to underrated juniors. An idea that I thought of was to have no limit with the 336 rule or 300 rule - whereby if you play someone rated 400 points below you then it is rated as if they are only 300 below you yet if they are 700 points higher than you then it is rated as such for you - but only as 300 for them... The ratings tend to go up but that counterbalances the loss of points to underrated kids.

There are several other ideas I know about, but I think that the best of them all would be to simply scrap the whole rating system altogether and start a new one based on the ratings used in diplomacy: In their rating system, you simply cant lose points - ever. The catch is though, that the higher you get, the harder it becomes to gain points. So in a chess sense, a 1200 could gain points by beating a 1500, whereas an 1800 may have to beat 2 1900s to gain points whereas a 2500 may have to score a 7/7 against a strong field averaging 2300 to gain only 5 points, etc. It could be worked out carefully if people wanted to do it...

Anyway, thats beside the point - the point was that the ratings in the event Trevor alluded to are simply ridiculous in my opinion and make no sense whatsoever.... How can anyone really play chess at 171 or 201 strength??? Its simply an insult to the players themselves - who are obviously much stronger than that...

Just let me ask one question. For a player who barely knows the moves, playing in his first 7-round event, who has a little cry after his first loss, and a long cry after his second loss, and then goes home...what rating would you assign?

101 looks just about right to me.

I would assign the child a rating of 1,000, which would mean that they have never won a chess game. But lets have a look at the positives.

1-The child took the brave step of playing in a registered chess tournament
2- The child turned up to the chess tournament on time.
3- The child got some experience of using a chess clock
4- The child knows where Box Hill chess club is and will hopefully tell their friends about the experience.

and on the negative side

1-The child has an official rating of 101 which destroys all self esteem.
2- It is not enough that the child is psychological traumatised from realising that they couldn't win a game of chess, now they need to be marked as a really bad player.

Seriously, what is the point of the mathematical accuracy in this case?

I'm just wondering what the effect would be on an 1100 rated player knowing in particular cases he was only rated 100 points higher than someone who hardly knows how the pieces move .

...

Seriously, what is the point of the mathematical accuracy in this case?