NOMOEO: It's Czech. I'm sure one of our many Czech players will help you.

ColonelCrockett: No he didn't, but that's not the point. Perhaps in your culture it is acceptable to make unsubstantiated accusations, but in mine it is not. We are a multicultural community here and in my opinion we need to act accordingly if we want to get on.

chessmec: Why "Of course"? Please tell us how you are so sure. In fairness, if SOCRATES has evidence that some players are cheating then he should show us the evidence. I'm sure the rest of us will all want to know. If, however, he does not have any evidence then, again in fairness, he should withdraw his accusation. By the way, your opinion does not count as evidence!

S O C R A T E S: What evidence do you have to back up this accusation?

ColonelCrockett: In which case you will like Chess for Tigers by Simon Webb, ostensibly a children's book but most adults benefit from reading it.

ColonelCrockett: In my opinion, it depends on who you're playing. I'm a firm believer in playing the man not the board and over the board in matches (rather than here) that can be very effective.

I like to play 2...a6. That's partly because I used to play the 2...a6 O'Kelly variation in reply to 2.Nf3 throughout the 70s and 80s, even against people like James Plaskett, transposing into a Kan if white played c4. An early a6 is often effect as it often gets White out of his normal thinking. That can be good against bookish players or players stronger than you. Against a weaker player you might keep to more classical lines. Whatever you do, don't end up playing against yourself - always a danger with very tactical lines against a weak player when a simpler alternative would have done.

I'm pleased to see that 2...a6 still gets good stats here:
>http://www.chessgames.com/perl/explorer?node=33281&move=2.5&moves=e4.c5.Nc3&nodes=21720.32033.33281

though the sample is rather small. In any case it is NOT a refutation of 2.Nc3. I doubt there is one.

WhisperzQ: What I was proposing is already an exponential system since the amount varies according to your level. Actually increasing the slope over time I hadn't planned for and I'm not convinced it would work, But there's an alternative: have a fixed drop per month. That means that the percentage increases over time. Perhaps 2 BKR per day?

It would start only when you have a rating, i.e. have completed at least one game in game type. You could also bottom it out at 1300 and not subtract points below that since that is the hidden starting BKR.

jfa: Ha ha! It may happen. My history is of playing for 2 years then not playing at all for another 5 years! And then the cycle repeats. But don't hold your breath. :)

(By the way 0.3% per week is about 0.07% per non-vacation-non-weekend day)

jfa: The reason I prefer the falling rating system over the removal system is that there are many players who, like me, like to concentrate on just one or two games at a time. So I might spend a year on chess and tablut and then another on another combination of games like tablut and backgammon (or more likely Go or Xiangqi if they ever appear). That's the way I play because that's how my mind works. I prefer obsession to diversity! :) I wouldn't want my ratings removed just because I'm having a sabbatical on one game to concentrate on another, but I'd be willing to take the hit from a falling rating.

danoschek: Yes, deductions should be stopped on a vacation day. But I would do that unconditionally, whether or not it is an autovacation. That is because you buy your vacation limit with the membership fee and it doesn't matter whether that's planned or auto.

jfa: That's why I say we should keep the existing system too. That would tell you your relativity. But the RANKING system would be based on these falling ratings.

Actually, I'd be in favour of introducing a "falling rating" system, not necessarily instead of the existing system but maybe as an addition to it. It would work like this. Existing BKR would calculated just like now from the outcome of games. However, additionally an amount would be DEDUCTED regularly from ALL BKRs. This could be a percentage like 0.3% per week.

This would mean that you have to keep playing and keep winning in order to keep a high rating. If you don't keep playing then people pass you and your 2700 rating will fall to 2300 over the course of year and down to 2000 over two years.

This is like negative interest in a bank account! The interest rate could be adjusted of course - 0.3%/week is only a suggestion - and the time it's applied could be daily instead (by a smaller amount). But the principle is still there that you couldn't rest on your laurels once you got to the top.

What does everyone think?

By observation, the rating system on this site is done by a very simple lookup table and not by the application of a statistical formula (although perhaps the lookup table had its origin in formulae). It seems to depend on two things: (a) your own BKR - not exactly, but instead where it fits into a few very broad bands, and (b) the difference between your BKR and your opponent's - again not exactly but in wide bands of 400 BKR or more. An inexact system like this will generate some odd effects but I think it's still fit for purpose.

ColonelCrockett: Yes. Read here.