I thought the main reason the Dual system sprang to life was to bring Quick ratings somewhat more in line with Regular ratings. How well has that worked?
In 2000, when the Delegates approved dual rating, there weren’t enough quick rated games for any differences between quick and regular ratings to be obvious.
Dual rating did increase the number of quick rated games. We rated around 40,000 quick games in 1999, and regular rated around 418,000 games.
In 2002, with dual rating implemented we quick-rated or dual-rated over 115,000 games. Since 2005 we have quick-rated or dual-rated over 300,000 games a year. (We are currently averaging around 480,000 games per year that are regular or dual rated.)
The differences between quick and regular ratings (especially the extreme cases) didn’t become an issue until a few years ago, and so far no steps have been taken to change that, because we still don’t fully understand what causes those differences, so it isn’t clear what steps are appropriate to correct them. (I’m still waiting for the Ratings Committee to give me a list of things to test on our testbed server.)
Moreover, as I have said more than once, if the whole point of having quick ratings is for them to be the same as regular ratings, why bother?
And, more than once, there is a good answer. The idea should not be to make everyone’s quick rating identical to (or close to) his regular rating. The idea should simply be to make the average quick rating similar to the average regular rating.
Some people are (relatively) better at quick than regular. Some are vice versa. Bringing the averages into line would allow us to see at a glance which players are which.
Everybody plays worse quick chess than regular, in an absolute sense. When I say that a given player may be “relatively better” at quick, I simply mean that that player’s quick play suffers less, compared to his regular play, than is the case for the average player.
Bill Smythe
Bringing the averages into line once does not guarantee they stay in line, Bill, especially if we don’t know why they diverged in the first place.
Surely we don’t want to have monthly/quarterly/annual ‘fiddle points’ in the quick system.
This issue is not worth more aggravation than it has caused already—but for the record here is a Quick Chess event from 1998 in which I played, as did Hikaru Nakamura and other patzers. Regular ratings were used for pairing purposes; grumbles about QC ratings being way off were already heard by then.
uschess.org/msa/XtblMain.php … 0-12119310
Going out on a limb I would guess that I was not stronger than Hikaru at any time control by then—most certainly not G/10. The issue of talented juniors dragging their QC ratings behind them, like Natty Bumppo’s canoe, did not just start a few years ago.
My take is there are too many such cases for them to be classified as “out-liers”…but let’s see the statistical analysis Mike Nolan and the RC produce. Till then, the majority of tournament players will scratch their heads when you start talking about Quick, Dual, increment and so on. As long as their Regular rating stays ‘pure’ the rest is head-scratching stuff to them…
I’m all for trying to figure out how quick and regular ratings became so divergent in the first place. But I’m not in favor, if the cause of that divergence should turn out to be impossible to nail down, of simply throwing up our hands and doing nothing.
There are, of course, two problems: (1) the two systems are currently hugely divergent, and (2) whatever phenomenon made them that way is probably still at work.
Problem (1) will have to be handled by some kind of one-time adjustment. Then problem (2) must be tackled also. Either alone will not be sufficient.
Since problem (1) is mired in ancient history (data no longer available), its precise causes could be tough to figure out. Problem (2), however, should lend itself to the usual thorough scientific study by the ratings committee. Either way, let’s get busy. I’m tired of hearing that, just because we can’t figure out exactly what caused the problem in the first place, we should sit back and do nothing.
Bill Smythe
I’m not advocating doing nothing, but I am opposed to doing SOMETHING without first studying its impact. And the more we know about the underlying causes of the problem, the more likely we are to be able to fix it, as opposed to making an arbitrary (and unscientific) change in the sincere hope that it works.
As to whether or not we have the data, as I recall quick rating was the reason the USCF went to the dBase/clipper rating system programming in late 1991, so I think we probably do have all the data (eg the games themselves) for quick ratings. We may not know the exact formulas that were being used at various points in time between 1991 and 2005, though.
There is evidence that there is still some kind of divergence occurring, players who began quick rated play in the past few years tend to have divergences, too. However, that may be little more than self-perpetuation of the existing divergences, since we have a large pool of players with lower quick ratings than regular ratings, and those new players get their quick ratings based on their performance against those who already have a (divergent) quick rating.
True, which is the main reason a one-time adjustment is necessary, as a first step.
To keep the ratings in line once they are brought there through a one-time adjustment, I suspect it will turn out that the main problem is inactivity. Perhaps old quick ratings should be allowed to “fade” with time, for example by using higher K-factors for players with old quick ratings but new regular ratings, or by counting old provisional quick ratings as being based on fewer games than they actually were, or by using a mixture of quick and regular ratings as the pre-event quick ratings, with the mixture being determined by recent activity (a high ratio of recent regular activity to recent quick activity would mean a high percentage of regular rating would be mixed with a low percentage of quick rating to determine the pre-event rating).
I agree that the effects of possible changes should be studied, to the extent possible. There’s no point in implementing a pure “guess”.
Bill Smythe
Time to implement glicko-2 for Quick?
I think players care (and understand) more than E Mark claims, but not sure if an entirely different calc system is needed.
I know Mark Glickman would like to see us implement some form of Glicko-2. 
Here is what I have observed:
Players care about their Regular rating; some care more than others, but nearly all care to some degree. Some players care about their Quick rating, though to a far lesser extent than their Regular. Some players avoided QC altogether when it came along; some of that group now avoids Dual-rated events for the same reason. (If Dual-rating is so popular why do we see G/61 events?)
Tournament players know about QC ratings and some know about Dual ratings—though I know guys who go Way Back who scratch their heads at the idea of Dual…but the ones who care are out of luck: Ever since the Dual system came to be, some events that should have been Dual-rated were rated only as Regular. Now we will see the opposite, since the policy that games with delay or increment of 16+ seconds can only be Regular-rated has not gotten around to the boonies. For that matter, it has not gotten around to many TDs; I know one bright young Senior TD who did not know about it until I told him and sent him a link to the updated rules doc.
That long preamble frames my point: If we’re gonna have Dual-rating based on main clock time and also have Regular-only games based on delay/increment time, let’s be consistent about it. (It’s like prize-money floors used to be. Some who qualified got the floor, others did not.) It was not consistent before the new increment/delay rules came along; it will be less consistent now.
As to how much players care, here are two contrasts in point: There is a club I know that runs one-day G/30 or G/40 events almost every week. Not one of those events has ever been Dual-rated, that I know of. My impression is that the main TD—a well-informed former USCF employee, who has directed hundreds of tournaments—prefers that the games be Regular-only and the few players who notice and care agree. Note I say “few” players who care…out of more than 100 who have played rated chess there the past nine years, including a few Masters and some active, savvy vets of the Mid-Atlantic Swiss circuit.
Then we have a popular chess academy run by a very strong player, that holds chess league nights twice a week, with a time control of G/60, Inc-30. These events draw a number of strong, active players who know the score when it comes to ratings and rules—at least the rules they have heard of. Until I informed the main TD that these games should only be Regular-rated, every one of them had been submitted and rated as Dual. This guy will likely be an NTD someday. No one I know is more earnest.
No player at the academy league night complained that the games should have been only Regular-rated. I will bet not one of them knew the 16+ delay/increment policy. No one has ever complained, (at least to USCF), that the games at the other club I mentioned should be Dual-rated. In either case, all it would have taken was one complaint to the home office to ‘fix’ things…but no one complained.
They did not complain because in all cases at both places, players’ Regular ratings were affected properly. Had that not been the case, you can be sure the howls of protest would come but quick…in contrast, the fact that Quick ratings might not have been properly impacted—in either direction—did not inspire one howl or protest that I know of.
That’s the point: Regular ratings matter a great deal. Quick ratings matter a bit, to some. Dual ratings annoy some players, make others scratch their head and some still can’t be bothered to figure out the nuances. Those reactions will all increase, perhaps in ways that amuse the perverse, as word of the increment/delay-based Regular-only policy spreads.
If G/60, Inc-30 explodes in popularity in the next year or two, as it might, confusion will be king and the Dual system will lose any credibility it has, thanks to its wildly inconsistent application.
Maybe it would be simpler to just scrap the Dual system…Since that won’t happen soon, it is worth some effort to spread the word on the ‘new’ Dual/Regular/16+ seconds delay and increment rules.
Aside from one point, which I agree with –
– this strikes me as much ado about nothing. Players who don’t care about their Quick ratings are free to ignore them. If they want to be childish, they’re also free to avoid any event that affects Quick ratings at all. But the gap between Quick and Regular ratings reflects the easily observable fact that some players can play as well, or almost as well, under a fast time control as under a slow one, while others play significantly worse, and since – lest we forget – Elo ratings are meant to measure probability of victory, not personal worth, it therefore makes sense to preserve Quick and Dual rating for the sake of fairer pairing in rapid events.
In fact, thinking about this, since Dual rating is presumably meant to reflect the fact that games between G/30 and G/60 are “regular” (slow) in some people’s minds (e.g., a hyperactive 11-year-old) and “quick” in other people’s minds (e.g., an old-timer who came up under 40/2, SD/90), and thus there’s a “blur zone” of overlap in what rating is affected, what might be good for consistency is if TDs got into the habit of pairing G/30 tournaments using Quick rather than Regular ratings, since that time control represents the faster boundary of the zone. It would make sense to few people to pair a G/60 using Quick ratings, so why should the default pairing system of a G/30 be different from that of a G/29? Genuine ambiguity should only occur near the center of the zone; at the ends, I would think, it should be fairly sharp.
There would be rioting in the streets.
G/45 is a very popular time control for adult (open) chess tournaments here in Northern California, in addition to G/30 (scholastic) and G/60 (adult/junior). AFAIK all of these events are dual rated, period. I wasn’t even aware that the TD could opt not to dual rate.
Michael Aigner
You are correct, the TD CANNOT CHOOSE whether an event is dual rated or not, the time control is what determines whether an event is dual rated or not. As a reminder, here are the rules:
- There must be at least five minutes of time per player on the clock.
- If there are 16 or more seconds of delay or increment time, then the event is regular-rated only.
- If #2 does not apply and if the total time per player is less than 30 minutes, the event is quick-rated only.
- If #2 does not apply and if the total time per player is between 30 and 60 minutes, the event is dual rated.
- In all other cases, ie, if the total time is more than 60 minutes, the event is regular-rated only.
A TD who intentionally mis-reports the time control information in order to change an event that should be dual-rated to one that is just regular-rated or just quick rated (or vice versa) could be sanctioned, as could the sponsoring affiliate.
There would be howls of protest from players who lost a class prize to a guy whose Quick rating was hundreds of points below his Regular rating.
If you want to try this out at your local club for G/30 quads with small or no prizes, go for it. For even small “open” weekend events with even trivial class prizes, it would be a recipe for disaster…plus if someone posted about it on this Forum someone might feel the need to frown and mention potential sanctions.
The TD cannot so choose; so say the rules. Properly so, I think, though time control boundaries are so fuzzy to some folks that I prefer not to mention sanctions until things reach the point of willful defiance.
The issue now is that the new rules regarding increment/delay of 16+ seconds—which makes an event Regular-rated only—have not been well-publicized and many TDs don’t know about them. This at the same time that 30-second increment games grow in popularity…which inspired the new rules to begin with.
The big problem is G/60, Inc-30. That’s about as fast (G/60) as a game with 30-second increment is likely to get. It is a popular time control already in places and will likely become more popular soon.
Some G/60, Inc-30 games are and will continue to be submitted for Dual rating. In most cases this is because the TD does not know that “Inc-30” makes the games too slow for Dual rating, per the new rules. (See specific example I cited in previous posts.) In some cases it might be because TDs try to submit such games through the online system at TD/A, but get error messages that fail validation. (See first post in this thread.)
Maybe the solution is to move the Dual threshold up one minute to G/30-G/59…
Ignorance of the rules is not an excuse, and I think the rulebook even cautions TDs to keep up to date on rules changes.
Updates to the rules have been posted on the website for several years.
The time control rules were changed in 2008 (effective 1/1/2009), and that information has been mentioned continuously on TD/A since then, including being bannered for several months, and I’m sure it’s been mentioned on the Forums quite a few times, and I think it has been published in Chess Life too.
I’m aware of one TD who was warned about submitting events with drastically different time controls from what was actually used in order to affect whether they were dual rated or not. (We found out about it because a player complained.)
We’re in the process of adding time control information to MSA crosstables, but that will only go back to events submitted since early 2005.
An email to all TDs and affiliates advising them of the dues and rules changes made in Irvine, and also mentioning the time control rules, is scheduled to go out tomorrow. (The copy was sent to several people for comments, so far nobody has notated any errors.)
You keep saying this, but I’ve never heard of it being done (actually once, but the TD deliberately fudged the time control so that it could be dual rated). You’ve also mentioned that there is one Senior TD who was running these events. Is anybody else doing it?
Alex Relyea
Fair enough, Alex. I do not presume to know the nuances of time control trends throughout the nation.
I do know that one fast-growing chess academy in NJ sponsors league nights twice per week. These events started as G/60-G/55, delay-5; they soon went over to the dark side of G/60, Inc-30. (I don’t know what if any analog equivalent they use.) The increment idea was embraced by the regular players, some of whom had used increment for the first time in a G/90, Inc-30 FIDE Futurity at the same site.
I know that players in a NJ club where I play and direct, (I live in PA but right on the NJ border) support G/60, Inc-30 for fill-in games that start too late to be played at our standard 40/90, SD/60. Some of them support changing the standard time control to either G/60, Inc-30 or G/90, Inc-30.
The push that got our club members to consider increment—rather than tune out the crazy fat TD who spends too much time on the Forums—was a gung-ho endorsement from the one player who regularly attends both our club and the chess academy I mentioned.
Little by little, increment will spread throughout rated amateur chess, one gung-ho endorsement and crazy TD at a time. Also, the lowered FIDE rating floor and the chance to hold Futurities with three-hour time controls for amateur players will make G/60, Inc-30 an attractive option. I think Sevan Muradian has mentioned that he used this time control, BTW.
So, I do not claim to see the future or know what happens everywhere now. All my instincts and limited-thus-far experience tell me that G/60, Inc-30 will catch on and become widely-used. If that happens, I am pretty sure it will be later rather than sooner before a clear majority of those events are rated as Regular-only rather than Dual.
That does not seem like a tragedy to me, and it seems other players see things the same way—though perhaps we need to wait until players who are not TDs or Forum geeks get the memo on the 16+ Rule…which will also be later rather than sooner.
I have massaged this point too many ways and rambled on too long in this thread. Enough said.