What was the reason????

This point can be seen when we imagine an extreme situation.
Let’s say we use a K-factor of 1000, then when any player wins a few games in a row he can suddenly be the highest rated player in the world.
When we consider a smaller and more realistic scale, lets say the K-factor goes from 10 to 30 for high rated players. Then one good tournament of any top 10 player could make him the number one on the next rating list, which would be a bit awkward.

Another argument that Macieja used in favor of a higher K-factor has two sides. He states that according to the simplest mathematical model there should be more 2700 players then there are in reality, and that this issue would be resolved with a higher K-factor. This is not completely true, due to the dynamic nature of the chess world. There are new players entering the system all the time. The simples mathematical model predicts that all ratings would eventually become normally distributed, and that the best players would have ratings that are on the right side of this distribution. However, because new players enter the Elo system all the time, there will always be players that are still under-rated. The day these under-rated players achieve their real high Elo rating, new players have entered the system that are under-rated. Therefore the actual distribution of Elo ratings will never be exactly like the normal distribution that is assumed by the model.
Basically, players who are strong will have a low rating untill they reach their true skill, and players who are weak will have a “high” rating untill they reach their true skill. So there will always be under and over-rated players, which is why the tails of the distribution will always be thinner then they should be according to the normal distribution.

To summarise, increasing the K-factor increases the speed in which players can change their Elo rating, but it will also introduce higher fluctuations which can create awkward situations in the top of the rating list.

Cheers,
Daan

If they mean that “less chess skill is required to maintain a certain rating over time” I will leave it without a comment, as I have not seen (or made myself) any proper study. It is quite possible, though the answer may depend on the rating possesed and also may depend on the period. We could have both inflation and deflation in the same period on various rating levels, we could also have both inflation and deflation on the same rating level in different periods.

If they mean, that the avarage rating of chess players (even those rated over 2203 - to avoid the problem of the decreased rating floor) is growing all the time, I would not be so sure, basing on my own studies.

The main problem is that many people don’t understand why the number of chess players rated over 2600 (or 2700) is growing and this is what they call “inflation” o even “unexplainable inflation”, thinking that the whole system has collapsed. Obviously, I cannot agree with this.

Remco,
I will try to write a bit more on the K-factor and on the rating inflation soon.

Thank you for all the comments.