Surrey's motion on grading for Council

[Mike Gunn]

Surrey County Chess Association wants to place the following motion on the agenda for ECF council: "Council disapproves of the practice of treating juniors as ungraded players and moves that it be discontinued.". We need the support of one other County to get it on the agenda. Please contact me on mikewxyz(AT)googlemail(DOT)com if you can deliver this.

[Alex Holowczak]

Can I suggest making the wording "all juniors", not "juniors". If a junior is ungraded anyway (i.e. a new player), it's perfectly sensible to consider them as ungraded.

[Mike Gunn]

Alex, thankyou, that improves the motion.

[Sean Hewitt]

Mike,

It might help if you can say why Surrey disapproves of the practice and what counter measure to the deflationary effect of junior grades you would put in place instead?

As has already been observed the junior increment is a one size fits all "solution" which actually solves nothing.

Sean
[Chairman & ECF Rep, Leicestershire]

[Mike Gunn]

We're not saying what the solution is, the point of the motion is to express dissatisfaction with the current approach. You are throwing away useful information about players by throwing away old results (albeit 20 results can now be used rather than 9). Hopefully further discussion and study will lead to an improved approach.

The basic problem is that nine games are too few to get a reliable estimate of a player's grade. This is why Clarke originally calculated grades on the basis of 30 games (or more if played in the current year). As a county grader I reckon about 5%-10% of players each year are new ones (of course this figure will depend on the type of event you are grading). The system can probably accommodate initially dodgy estimates of the grades of this number of players (and their grades will get more accurate as the years pass and they play more games). On the other hand, treating all juniors as new players is potentially a disaster, particularly as the estimation method for new players doesn't seem to be very stable. For example if you take the (now) infamous case of Mr Taylor of Morley College, I calculated his grade as 196 based on the games published on the London league website (ignoring the ungraded player and scoring +50 for a win). His original published grade was 216 (although it has now come down to 203). It is difficult to see on what logical grounds he should be awarded a grade of above 196 on the available information - you can't really squeeze more useful information out of the data.

[Sean Hewitt]

To be honest Mike I haven't seen a coherent explanation of what the current approach actually is.

I think it's difficult to put a resolution that doesn't offer an alternate approach though - even if that new approach is generic rather than specific.

[Mike Gunn]

Sean, the implication of our motion is that we return to something like the old system, perhaps accounting for improving juniors in a different way. For example let's suppose a junior plays 10 games a year. You could apply weighting factors of 0.5, 1 and 1.5 to the junior's points totals from 2 years ago, 1 year ago and last year (before adding the 10 points/whatever junior increment) This would give a grade which reacts more quickly to rapid improvement. (Note, I haven't thought about how to generalise this to different numbers of games in the last 3 years.) Obviously moving to such a system would require some prior study, different weighting factors could be used. A scheme like this would be relatively transparent and would allow non-juniors to calculate/ keep check of their grade in the normal way (if they want to, I know some who do).

(I don't think Council is the body to determine grading strategy in detail, but how else do we test the strength of opinion/ feeling about what is going on?)

[Kevin Thurlow]

I agree with the general thrust of the Surrey proposal, but you probably do need to suggest an alternative procedure - I am surprised nobody said that at the Surrey meeting where it was discussed. I don't see a problem with the age related increment. After all the grade can only be a measure of last year's performance in the games submitted (allowing for typos etc) - it cannot be used to predict the result of every individual game in the future. Even Asimov's psychohistory didn't do that...

I know tournaments that award a "best performance" award based on grading scored - imagine someone contacting the tournament organisers the following August to say that he should have won it because he played a junior who was retrospectively 20 points higher!

[Sebastian Stone]

Rather than age, why not experience?

eg. Someones who has been playing 20 years is less likely (until old age becomes a factor) to change dramatically than someone who has only been playing for 2 or 3 years.

[Alex Holowczak]

Rather than age, why not experience?

eg. Someones who has been playing 20 years is less likely (until old age becomes a factor) to change dramatically than someone who has only been playing for 2 or 3 years.

I would agree with that completely. The problem is, how do you know how long they have been playing? They might have played on the Internet for years, and then joined a club/Congress suddenly. So the grading system says they're new to it, but in reality, they're very experienced.

[Brian Valentine]

I have some sympathy with the motion, but I feel it is the wrong response.

This whole mess comes about from knee jerk reactions to some statistics that suggested a problem with the system. I want the ECF to stop making knee jerk decisions on something which is key to justifying members fees - a credible grading system.

The federation needs to stand back, understand the key deliverables of the system, test whether the system meets those deliverables and finds solutions to any weaknesses. Something does need to be done with the system for all ungraded players; fixing it with an ad hoc patch is likely to lead to unforeseen side effects.

The motion as it stands encourages a knee jerk response, in that there has to be an alternative in place. Watching how this project has been conducted so far there appears that the replacement will have its own problems, especially if rushed in.

I was hoping Kevin Thurlow was going to give feedback from the Thanet Congress on this issue. Maybe nobody else cares? This forum may be addressing a storm in a tea cup (the old term for a chat room )

[Sebastian Stone]

Rather than age, why not experience?

eg. Someones who has been playing 20 years is less likely (until old age becomes a factor) to change dramatically than someone who has only been playing for 2 or 3 years.

I would agree with that completely. The problem is, how do you know how long they have been playing? They might have played on the Internet for years, and then joined a club/Congress suddenly. So the grading system says they're new to it, but in reality, they're very experienced.

Meh, internet chess doesn't necessarily prepare you for the real thing.

[Roger de Coverly]

some statistics that suggested a problem with the system.

Three years down the line, do we have a clear idea on exactly what these statistics are? There's the "Grist Graph" which is the s-shaped thing plotting grading difference v %result. We've only ever seen one of these so as to whether it's the same picture across seasons, grading ranges and activity is generally unknown. I'm aware that they only have detail results back to 2002 but there are numerous crosstables for events such as the British Championships available prior to that so they could have sampled these for historic flavour. My point being that if the Grist graph shape is stable and ever present historically then there's no particular need to panic and destroy the historic continuity and in the case of ignoring junior grades damage the utility of the system.

[Kevin Thurlow]

Re -Thanet, for Brian... I did get one idiot who appraoched as I was doing the pairings for round 1, and said, "Can I ask a question?" I said, "Yes", and he then asked if I could explain the grading system to him. I pointed out that the draw was rather more important and suggested he ask the ECF, but by then one of his mates had dragged him away and berated him for his stupidity. I did hear people discussing grading and at least a couple of players didn't understand how they had gone up after getting rubbish results! There was certainly dissatisfaction from players who (like me) keep a record of results and monitor grading performance through the season. This actually included juniors who might play other juniors quite a lot (although many junior-only events are ungraded of course). My knowledge of statistics is fairly basic, but to treat someone who plays 100+ games during a season as a new player the following year is absolutely ridiculous. No system will work for everybody, but you cannot have a system for each player... The danger is then that you start adjusting the grade based on personal reasons, not arithmetic. The old-style junior increment seems a lot fairer.

The new grading "system" resembles the accelerated and Dubov pairings systems, which are so complicated, very few players will understand them, which is a huge benefit to the organisers as people are less likely to come up and complain.

I did find it comical that one junior organiser bleated that the juniors might go up too much and start losing games in higher sections and get depressed, and might give up. Well, if you're afraid of losing - you should give up!

Probably the highlight of the event was when somebody came up and fawningly told me how wonderful I was (which compliment I graciously and modestly accepted), then it transpired that he thought I was Peter Bayliss! He retreated in some confusion, just as he did last year when he did the same thing...

[Richard Bates]

I did find it comical that one junior organiser bleated that the juniors might go up too much and start losing games in higher sections and get depressed, and might give up. Well, if you're afraid of losing - you should give up!

What you find "comical" is just a perfectly reasonable argument. Juniors won't give up because they "are afraid of" losing. They will give up because they ARE losing, effectively being barred from playing people of their own true strength, whether by being forced into higher sections of tournaments or onto higher boards of club teams. With an ever aging chess population, far better "under-rated" junior players dominating the prizes of tournaments, than giving up disillusioned with losing all the time IMO.