The Taffometer Proposal

The opinions expressed in this document are the opinions of me, the author, John Trevelyan. I do not represent any club, group, faction, or whatever, but I would like to thank those, with whom I have already discussed these views, for taking the time to listen, encourage, criticise, or whatever. I have no wish to offend, or blame, anybody in this document. If anybody feels offended, or blamed, then I am sorry. That was never my intention. These will probably remain my opinions. I have named several people in this document. I am sure that none of them will take offence.

I like to think, and it's probably true, that I have no personal axe to grind here. I've been in the Welsh team before, and I've bought a few tee-shirts. It doesn't bother me if I'm never picked again. I'm getting too close to sixty to imagine that any more such experience will do much: either for my chess or for the future of Welsh Chess. On the other hand, if the Welsh team needs me to make up the numbers, that's fine too. I haven't retired.

I will keep the main part of this document as short and simple as I can. There is more information available, for those who want it, by clicking the appropriate links. Naturally, whenever I refer to a player as "he" I really mean "he or she". I have no intention of murdering the English language. No doubt, in future, there might be lots of discussion. In order to avoid answering the same question many times, you can examine any previous questions and answers here FAQs

After the WCU Management Board Meeting, in August 2007, I decided to give my "Taffometer" idea, already some years old, some more thought. On the same evening I made one that worked. It has been altered many times since. More about the original reasons for, and evolution of, the Taffometer here. Also more about how it works, and how to use it. I now propose that we use it.

I accept that there is a lot of maths in my proposal. This is not a problem. I have written a computer program to do most of the hard work. The ELO rating system is also mathematically complex but, fortunately, we have computers do do the hard work. We tell FIDE our results and FIDE give us a new rating. Simple enough!

Note that most of the numbers in my proposal are also hyper-links. They are numbers that I consider reasonably OK. Clicking the link will explain why this number was chosen, and what the effect of changing it will be. Any of these numbers may be changed quite simply (HOW) at any AGM. They might also be changed before I propose this!

In order even to be considered, a player must have been active enough in Welsh chess (20 games per year). Also, he must also have a Taffometer reading above the minimum value: 12 percent.

For selection purposes, we calculate the "Total Eligibility Rating", for the want of a better term, of every player registered with the WCU. It's not a difficult task. A player's TER (Total Eligibility Rating) is calculated as follows: It is the sum of four numbers:

a) The player's ELO rating.
b) 100 times his Taffometer reading.
c) 200 times his "Junior rating".
d) 1.00 times his activity rating.

The Selection Committee's task is simply to pick the players with the highest TER.

Note that these numbers may well be different, for European Team Championship teams, to the numbers used for the Olympiad. The composition of the ideal teams for each event might well be different. [Why?] The numbers for the Ladies' events might be different again. [Why?]

That's the end of my proposal. It might seem complicated, but it's not. [Why?] I cannot believe that it is any worse than the current system. [Why?]

The rest of this document explains why certain numbers were chosen, and the effects of altering them. There are more numbers [here] in the explanation of how the Taffometer works.

MTR: Minimum Taffometer Reading. 12 percent. This apparently low number has been chosen to be consistent with the current system. A person who is eligible via all five methods is arguably more Welsh (100%) than somebody who is eligible by only one method (12%). At the moment, one method is enough. (back)

TRM: Taffometer Reading Multiplier: 100 times. This number controls how likely it is that a more Welsh, but weaker, player will be selected ahead of a less Welsh, but stronger player. Probably best explained by an example: Take two players (Player A: ELO = 2200 TOR=90%) and (Player B: ELO = 2240, TOR = 50%). Ignoring "Junior ratings" for a moment, both players would have a TER of 2290. Maybe we toss a coin. A number higher than 100 would mean that the more Welsh player A would be preferred. A lower number would favour the stronger player B. (back)

JRM: Junior Rating Multiplier. First, a players "Junior rating" is (100 - (4 x AGE)) percent. Players older than 25 have a Junior Rating of zero. A higher number here will make it more likely that a promising junior will be selected ahead of a burnt out has-been like myself. Assuming that I am 2200, it's a toss up whether a 15-year-old, rated 2120, would be selected ahead of me. A higher number would make certain that this junior would be selected before me, and a lower number would favour us old fogeys. (back)

ARM: Activity Rating Multiplier. Activity is measured in significant games per year. I think that the current rules say something like 60 games over the last three years. I think that somebody who plays lots of games, in Wales, every year should be more likely to be selected somebody who plays the bare minimum. I think also that older games should fade away gradually rather than suddenly disappearing after three years. The Taffometer, given the necessary data, can measure one's activity rating too. (back)

How to change the numbers: The current system is time consuming. Somebody proposes that we change the number 400 to 300, and somebody else proposes an amendment, and we vote on the the amendment first, and it takes all night. I suggest that, once we all understand the system, we allow every member to download a voting form (here) which allows him to say how he would like every number to be changed. Somebody collects one form from everybody who attends the AGM, possibly proxy or postal votes too, and possibly at the end of the meeting he says "This number has been increased by 5, The second number has etc. etc. etc." Very few opinions are changed after an hour-long debate. The AGM, and the management board, should be given the chance to get on with more important matters. Click here to see how your votes will affect the system. (back)

Why different rules for different events? In an Olympiad, if the Welsh team is doing badly, it gets paired with a weak team, from an island in the Caribbean or a small country in central Africa. The team scores some points and everybody feels good. It's better than a rest day. In the European Team Championships there are no weak teams except, obviously, for Wales. Not finishing in last place is a major success. It's possible that, in the European Team Championships, we should be less fussy about Welshness, and place more emphasis on ELO ratings. Leighton has told me that there are not so many weak teams in an Olympiad as there once were, and they are getting fewer. No doubt, he's right. Nevertheless there still are more weak teams in an Olympiad than there are in the European Team Championships. If Leighton is right, or gets more right as time goes by, then the numbers used for the two events should move closer together. They might well so so. I don't care. I might get a vote but it's not my decision. It's yours. (back)

Why different rules for the ladies? Not different rules, just different numbers! Why not? I might be wrong but I get the impression that the stronger Welsh ladies are exiles. The numbers with which we start, and not just for the ladies, should ideally select the same team that we have selected for years. The selection system might change dramatically but the team that we select should change gradually. Who votes on the changes for Ladies' events? Just the ladies? Do we blokes get a vote? I don't mind either way. I'll leave this question for the Management Board to sort out. (back)

Not more complex? Well, maybe a bit. Once my original Taffometer program was written, it took me about five minutes to type in the data necessary to calculate my own reading. Players who want to be selected will have to explain how Welsh they are, as opposed to how they are Welsh. This information must be made public, so that others may check it. Any player can download the Taffometer program, calculate his own rating, and send in the data. Or they can send me their (very basic) CV and I'll do it for him. Not too much work - a small price to pay. (back)

Why no worse? We already have a sort of Taffometer, although it's not called that. If you satisfy one of the five conditions, in the eligibility rules, you are 100% Welsh. Otherwise you are 0% Welsh. Not the most accurate instrument in the world and, with only two possible results, it never will be. It is probably un-repairable although many people have tried. Even so, using zeros for the TRM and JRM would make my proposed system virtually the same as the current system, but zeros are almost certainly not the best numbers to use! My proposed system is much more flexible and much more easily changed. (back)