rank inflation

   

Note: words in parentheses and italics added in order to disambiguate rank from rating as explained by Bob McGuigan below.

A situation that is ALWAYS going to happen with real-world and Internet Go rankings (ratings).

One possible cause for this is that more players are getting stronger.

The only way to (artificially) stop this is to constantly reshuffle the points/fractals/system used to determine the rating points and therefore the rank (rating). IMO this is unfair, especially when a player has just reached, say, Dan level after years of trying, and then with a new system to have to lose this ranking to keep the number of Dan players down (the main complaint of rank inflation to most is that it "cheapens" the rankings).


Thank you, Tim, for creating this wiki and your clarification on rank - on go servers.

Undoubtedly, our general knowledge is improving. As a way to gauge this, I remember reading that a millennium ago only the best players were able to replay games from memory, which would indicate a natural inflation rate of ~ 2/century.

Is this what we're observing now? From some of the discussions I read I gathered the impression that the inflation rate was pretty dramatic on some servers recently - more in the order of 2/year. If this is the case, are we getting dramatically smarter?

I think the argument about "cheapening" depends on one fundamental decision: Are ranks (ratings) primarily an acknowledgement of an achievement, or are they a tool to make games interesting? The answer probably differs between amateurs and pros. Maybe we should give up trying to reshuffle ranks for amateurs and just readjust the dan-pro border. -- Sebastian 2003-09-10

HolIgor: The rank inflation is more likely to happen in real life than on the Internet servers. IGS was often critized for its deflation. People don't usually like going down in the rank (rating). In real life, especially in Japan there is a custom to issue diplomas to the amateur players. People pay money to have their ranks nicely printed. And if some rank is achieved once the player would never agree that in fact, on average he or she is weaker. That is the main issue with the inflation. In many real life ranking systems ranks cannot go down while the concentration or reading ability of the player has its ups and downs. Servers are much stricter in this respect (due to reporting rating rather than rank).

BobMcGuigan: One way to think of the situation in Japan is that the diploma rank represents, in some sense, best sustained performance level, not current playing strength. Thus there is a difference between ranks and ratings, the rating indicating the current level of play. By the way, in Japan professional ranks are also awarded with diplomas.

tapir: In self-ranked environments I guess the most important issue that leads people to overestimate themselves is the "i won against him = i am as strong or stronger than him", when winning 1 in 5 or 1 in 3 games actually means a difference of about two and one stone in strength, respectively.

Velobici: might be good to reword this page to discuss the issue more carefully as the page is entitled "Rank Inflation", yet part of the discussion focuses on (calculated/estimated) ratings as opposed to ranks (granted/awarded/certified).



This is a copy of the living page "rank inflation" at Sensei's Library.
(OC) 2011 the Authors, published under the OpenContent License V1.0.
[Welcome to Sensei's Library!]
StartingPoints
ReferenceSection
About