A modest proposal to FIDE regarding ratings

Near the end of the thread about FIDE rating regulations, some issues have been raised regarding FIDE rating systems vs allowable FIDE time controls.

Just as U.S. Chess has three OTB rating systems, Regular, Quick, and Blitz, so FIDE has three, Regular, Rapid, and Blitz. The boundaries between the FIDE systems are different from those between the U.S. Chess systems, but the idea is the same, to separate slow from medium, and medium from fast.

The FIDE specifications for their three rating systems are:

Regular: mm + ss at least 60
Rapid: mm + ss between 10 and 60 exclusive
Blitz: mm + ss between 3 exclusive and 10 inclusive

(mm + ss means main time in minutes plus increment time in seconds. It is the total of the two that determines the rating system.)

The above specifications are contiguous. There is a clear and unique boundary between Regular and Rapid, and between Rapid and Blitz.

But FIDE throws in another wrinkle, by demanding that, for players in certain FIDE rating ranges, Regular mm + ss must be slower than that. Specifically:

  • for players rated 1800-2399, mm + ss must be at least 90, and
  • for players rated 2400 or higher, mm + ss must be at least 120.

This wrinkle is laudable from the standpoint of guarding the integrity of FIDE Regular ratings, but it leaves gaps in the specifications:

For players with FIDE ratings 1800-2399:
Regular: mm + ss at least 90
Not ratable: mm + ss between 60 and 89 inclusive
Rapid: mm + ss between 10 and 60 exclusive
Blitz: mm + ss between 3 exclusive and 10 inclusive

For players with FIDE ratings 2400 and higher:
Regular: mm + ss at least 120
Not ratable: mm + ss between 60 and 119 inclusive
Rapid: mm + ss between 10 and 60 exclusive
Blitz: mm + ss between 3 exclusive and 10 inclusive

So what happens when a player rated 2450 plays in an event where mm + ss is between 91 and 118? Depending on which section of the FIDE rules you go by, either (a) that player’s games do not get FIDE-rated, or (b) the entire tournament gets thrown out, and the TD is severely penalized for allowing that player in.

Needless to say, this is ridiculous. It would be much better if FIDE would close the gaps:

Regular: mm + ss at least 60, 90, or 120, depending on the player’s FIDE rating
Rapid: mm + ss between 10 and the above number, exclusive
Blitz: mm + ss between 3 exclusive and 10 inclusive

Now all the worries and discontinuities disappear.

This would require, however, that the tournament be able to affect more than one of FIDE’s rating systems. This, I’m sure, is bringing about howls of protest from some of you by now.

But why not?? If players FIDE-rated 1700 and 2450 are paired against each other, their game would be Regular-rated for the 1700 and Rapid-rated for the 2450.

And no, there is nothing inherently wrong with this idea. Since a 2450 usually defeats a 1700, there would be a large number of games where the Regular system would donate a small number of rating points to the Rapid system, and a small number of games where the Regular system would snatch a large number of rating points from the Rapid system. This “large” number of “small” exchanges would largely cancel out the “small” number of “large” exchanges in the opposite direction, so that neither system would be significantly inflated nor deflated.

What do you think of that idea, Baba Looey?

Bill Smythe

It’s a technical nightmare.

At the FIDE end, it would require some re-programming of the ratings code.

At the U.S. Chess end, it would require no changes at all. All that matters is the U.S. Chess ratings, not the FIDE ratings.

In the case of multiple schedules, with one schedule being FIDE-ratable and the other not, some kind of manual action is already required, by the organizer, to weed out the non-ratable games (or rounds) from the version of the rating report that goes to FIDE. I don’t know how this is presently being done, but I would guess it’s via a list of which games were played in the (pre-merge) 2-day schedule, versus those played in the 3-day schedule or post-merge rounds.

This same list could be used to identify which games were played at which time control. Usually, there would be just two schedules and two time controls, something like G/90 inc/30 for the 3-day and post-merge, and G/45 d/5 for the 2-day. This would equip the FIDE version of the rating report with all the information it needs to assign the correct FIDE rating system (Regular or Rapid) to each player in each game.

Simple, elegant, and logical. And no player is deprived of the right to have his game FIDE-rated, and no organizers need to worry about whether to allow high-rated players into their tournaments.

Bill Smythe

You’re being too kind.

Maybe don’t present it that way to anyone who cares about FIDE?

Is it, though?

Er, I’ll just say debatable.

Rating games over 60 minutes as if rapid (defined by FIDE as less than 60 minutes) is logical to you?

At the U.S.Chess end, all that matters is U.S. Chess ratings. The rest is a suggestion to FIDE about FIDE ratings.

Absolutely. If it’s too fast to be Regular, it’s Rapid. For a player rated over 2400, FIDE has already defined G/90 inc/30 as too fast to be Regular, so it should be Rapid. That’s as logical as it gets.

Bill Smythe

In the title you might want to replace the word “suggestion” with “proposal”.

Granted, it may make no difference in the amount of support received.

OK, I did that. Let’s see what happens.

Sometimes people just don’t want to consider new ideas.

Bill Smythe

The issue here is fairly simple. FIDE has set a minimum time control and a maximum daily playing time based on a player’s FIDE rating. If a player’s FIDE rating is high enough that one or both of those are violated by the planned time control and schedule of an event, then that player should not be playing in that event.

Has anyone considered asking our FIDE team for clarification from the FIDE office as to what action the organizer is expected to take when registering players for an event?

Still not correct. . .

Alex Relyea

We’ll see if it kicks off as much discussion as when Jonathon Swift used that title.

I wish all of you naysayers could understand that there is a very real problem here, and that only FIDE can fix it. If you don’t like my proposed solution, please come up with one of your own.

In other words, I wish everybody would either put up or shut up.

In Illinois a young GM has surfaced recently, whose FIDE Regular, FIDE Rapid, U.S. Chess Regular, and U.S. Chess Quick ratings all exceed 2600.

Who, among you, will stand up and volunteer to be the one who tells this GM, sorry, I can’t let you play in the “open” section of my FIDE-rated event, because your FIDE rating is too high?

For players rated below 1800, FIDE has, essentially, already defined Rapid as “too fast to be Regular”. I am simply proposing that FIDE extend this definition also to games played by players rated 1800-2399 (where mm + ss must be at least 90) and 2400 and up (where mm + ss must be at least 120). The effect would be, simply, that games involving these high-rated players would be FIDE-rated in their Rapid rating system rather than in their Regular rating system. I really don’t see why that idea is so shocking to you guys.

This would be good, as far as it goes, but “our FIDE team” should also push for this itty bitty little change while they’re at it.

Bill Smythe

I’ve often fielded programming requests where the requestor wanted what they thought was a massive change and it was only a minor tweak (sometimes even when the requestor is a programmer that worked with the programs that needed a tweak) or the requestor wanting what they thought was a minor change that turned out to be similar to adding a floor to a four story building, and doing it by inserting it between the first and second floors.

Without knowing how the FIDE programs were written there is no way of telling whether the “itty, bitty little change” falls into type A or type B.

That entirely ignores the issue of whether or not a game really should be rated half in one system and half in another (which would probably require approving a fundamental rule change).

I don’t know a lot about how FIDE’s ratings programming is done, though I did have some discussions in Turin with several FIDE officials on ratings, but I suspect Bill’s proposal falls more into the category than the one.

The reporting format would need to be changed, as it does not contain any fields that could differentiate between games that belong in ratings system 1 vs ratings system 2, if pre-event ratings are insufficient for that purpose. And I suspect that’s just the beginning of the changes.

I know for a fact it would be difficult to change the US Chess ratings programming to do what Bill suggests.

I don’t know the full reasoning behind why 2400+ players are required to have slower time controls, was this something higher rated players demanded?

Here’s why I think we need guidance from FIDE:

Suppose there is a player who is at 2399 on November 25th but an event that finishes on November 29th and is submitted promptly raises his rating to 2401. He is planning to play in an event that starts on December 1st. As a 2400 player for the month of December, he is now required to play at the slower time control. Not a lot of advance notice for either the player or the organizer here.

My favorite story on programming changes came from someone I knew who worked for a company that did payroll programming.

The IRS, in conjunction with the SSA, was contemplating changing the social security number from 9 digits (with thousands of known duplications) to an 11 or 12 digit number.

The estimated cost to change all the payroll programming at that one company was in the $30-50 million range. The insurance industry, which also uses SSIDs, was also alarmed by the proposal.

The proposal died a quiet death.

Having done some programming myself during my life, I too have seen both of these types of cases.

True. But not making this change (or something similar) has resulted in the untenable situation we are now in, where a high-rated player may have to be told he cannot enter a FIDE-rated USA tournament because his FIDE rating is too high.

I had actually dropped that part of my proposal in my most recent post above, realizing that simply FIDE-rating such games as Rapid for both players would at least solve part of the problem, while making it a little easier for FIDE to swallow.

I should add one more observation. If an organizer wants to FIDE-rate the top section(s) of his tournament, the problem will occur only if the section specifies an mm + ss faster than 120. So, for the time being at least, organizers should use 120 or slower (such as G/90 inc/30) for all FIDE-rated sections.

It seems that, in Illinois, a couple of recent or upcoming tournaments have, fortunately, heeded this restriction. Our 2600-rated GM entered the Sevan Memorial tournament last weekend, and is already registered for the Jane Addams tournament coming up this weekend (see King Registration). I only hope nobody will try to FIDE-rate a tournament with a control of G/60 inc/30 or something like that.

Bill Smythe

In my mind, you’ve already identified the correct solution:

Organizers should plan FIDE-rated events according to FIDE’s requirements.

I’d be willing to stand up and volunteer to ask the organizer why they designed an open event so poorly that a GM could not play because of too high a rating.

I would imagine that FIDE thinks that anyone who attempts to run ratings restricted sections using any rating other than a FIDE rating is incredibly foolish and deserves whatever problems they get.

Also, kudos to FIDE for not endorsing the idea that every single game of chess must be rated, so we’ll find a way to rate it even if it doesn’t make much sense. I’m fine with the idea that there are some controls that just can’t be rated.

Alex Relyea

I think FIDE assumes that organizers have full control over the entries in their events, that they aren’t just ‘open registration’ events. Thus an organizer who runs a FIDE rated event at a time control faster than 120 minutes is not planning to invite (or allow) any 2400+ players in that event.

There may be a championship event and a reserve event, for example, with 2400+ players in the championship event. That is consistent with how many European events are run.

I think the best solution is clear, if you plan to have a truly ‘open’ FIDE rated event, make sure the time controls are 120 minutes.

I think Bill’s proposal is interesting and shouldn’t be dismissed as quickly some in this thread have done. However, I think it would be much better if FIDE just made G/90 ratable in the FIDE standard rating system for everyone.