I added the following comments to the chess discussion page:
I am not really a fan of wikipedia, although I do find it useful from time to time. I tend to find the process laborious and prone to all sorts of nutters, trolls, propagandists, preachers, haters, self-righteous, etc. I find it occasionally useful, but am wary of relying on it without checking things from several other sources, and couldn't usually be a**sed to correct it or improve on (for fear of being dragged into the whole Bart Simpson, "Never argue with an idiot, they will only drag you down to their level and win on experience" experience.) Nevertheless, I do see the value that an informative, detailed, correct and NPOV article on ECF grading would have. If anyone here felt the urge to do this, then plaudits from me.http://en.wikipedia.org/wiki/Wikipedia_talk:WikiProject_Chess#ECF_grading wrote:The following text comes from the ECF grading article:
'Compared to the 'four-figure' rating system used by FIDE and other national federations, based on the Elo formula, the ECF formula is much less sophisticated. As a result, however, it is quite easy for a player to keep track of their own ECF grade from game to game whereas an Elo is a little more difficult to calculate. Conversion from ECF to Elo traditionally uses the formula:
ECF x 8 + 600 = Elo
This substantially deflates the predicted ELO grades of sub-master strength players compared to observations, so for players graded ECF 215 or below the ECF have issued the formula
ECF x 5 + 1250 = Elo
to replace it. This substantially inflates the Elo grades of much weaker players, but such players are less likely to need an international grade.
The current ECF board prior to their election expressed a desire to move to the Elo system within a few years subject to approval from ECF members.'
This seems pretty poor to me for the following reasons:
1. Why does one suggest that the "formula is less sophisticated"? RWB Clarke spent some considerable effort producing a grading system that was both simple to operate and yet of statistical value. I would suggest that the formula is, perhaps, "less complicated".
2. The translation formula "ECF x 5 + 1250 = Elo" is somewhat short of the truth. That formula applies only to the conversion of ECF grades to FIDE ratings. The old "ECF x 8 + 600 = Elo" still applies to the conversion to national Elo ratings. There is a strong distinction, although the writer of this article appears to be unaware of it.
3. The formulas given are no longer valid, given that the ECF grading system has now been subjected to a, somewhat controversial rescaling exercise. Formulas should be given for three date ranges, traditional, recent but pre-rescaling and post-rescaling, to both FIDE and national Elo systems, with note of applicable grading ranges, where that is relevant. It would also be useful, if a conversion section is given, to mention that various national 4 figure scales tend to be have higher or lower numbers than those given by the formualae (e.g. USCF numbers are generally thought to be 100 points higher than suggested by the formulae.)
4. I think the words "inflate" and "deflate" are out of place here. They generally refer to a perceived drift in the ECF grading scale, before the recent re-scaling. And anyway, the real reason that the formulae were revised was that FIDE ratings below 2300 had tended over the previous 5-10 years to inflate, given the effect of dropping the entry level from 2200 to 2000 (and lower still even later).
I think the article also misses any sense of history, failing to mention the work that RWB Clarke put into. It also misses any sense of statistical underpinning to the ECF grading system. It also lacks any statement about the properties of the scale. And finally it leaves out the fact that for half of its existence, it was not a three figure scale, but a two figure scale (1a down to 6b and below).
Regards,
Paul McKeown.
Regards,
Paul McKeown.
