Most of us got the email. No opinions?
I guess I don’t understand what an ‘arena’ event is, or why it creates game conditions that are not equal, like an odds game does.
The Delegates were emphatically against rating Chess960 in with classical chess, and there doesn’t seem to be much interest in setting up separate Chess960 ratings systems.
Lichess has “arena” tournaments, which don’t run in “rounds”, but simply have an overall time period. When you’re done with a game, you go back into the pool of available players. In general, it does 1-2, 3-4 pairings with the available players. It can be set to have bonuses for winning streaks and bonuses for “berserking” which means playing with 50% of the standard time + no increment. The scoring system tends to reward people who play fast (and particularly who win fast) but also tends to reward players who agree to quick draws when a game doesn’t seem to be going anywhere so they can get back into the pool.
Hmm, interesting. I don’t see why an “arena” event couldn’t be rated, except for the “berserking” part, which might cause some games to be regular-ratable while others are quick-ratable, or something like that.
Of course.
Bill Smythe
Berserking is an individual choice (though both players could independently choose it) so it’s not all that different from someone showing up very late for a game (except they get extra credit if they win). The main problem is that there is a strong incentive is to end games quickly—by resigning or offering draws—to move on to another game.
Both players must have identical time, or the game is not ratable.
There’s also a rule for blitz events that all games in a section must have the same time control.
All games in a section must be ratable for the section to be ratable. Otherwise, organizers or players might try to pick and choose which of their games are rated.
Either of those issues could render a section unratatable.
Some players may face each other more than once in an arena.
Each player completes a different round schedule, even without late entries. Player A faces player B, but it might be A’s 3rd game while B is already on his 5th game. This will be one messy crosstable with lots of -U- to match up rounds.
The US Chess ratings system has no prohibitions against player meeting an opponent more than once in an event, that is a pairing paradigm. (Ladder events often feature multiple games between people on the same rung or adjacent ones.)
If there are too many games between two individuals, those games may be treated as a match for ratings purposes (which limits the total number of points that can be gained or lost over several time periods), but otherwise they remain ratable.
And as to messy crosstables, we’ve certainly got enough of those. A few years back we tried looking at events where less than 50% of the possible games in an event were actually played, there were too many of them for that to be a useful tool for finding possibly suspicious events. That appears to be the case with locating undeclared arena-style events as well.
Do Arena events have time controls, or is it just, say, two hours to finish as many games as possible?
Alex Relyea
What about cross-round pairings? Doesn’t the U.S. Chess software already allow these, even if they are not marked as such? What if there are gazillions of apparent cross-round pairings? If A defeats B at one point in the crosstable, it should be important only that B loses to A somewhere else.
Bill Smythe
My understanding is that the individual games are played with clocks, most likely at blitz time controls. Are there any online servers that don’t enforce time controls on their games?
Cross round pairings involve two players in the same section, so I think you really mean cross section games. Even those still have to be in just one section in the rating report, whether that’s the original section for player A, the original section for player B or (IMHO the preferred way) in a separate ‘extra games’ section. How it’s handled in the original sections is pretty much up to the TD (usually as unplayed wins, losses or draws), it sort of depends on how they want tie breaks to work.
Since part of the ratings algorithm uses Player B’s intermediate (step 4) rating to determine player A’s final rating, the players have to be in the same section.
No, I was thinking of a huge, single-section tournament, like an arena event. All the cross-round pairings in such an event would be in the same (the only) section.
Of course. But it isn’t always necessary to have an “extra games” section. If, for example, both A and B are given full-point byes (in sections A and B respectively), the TD might offer them a game against each other. If A wins, add B’s name to section A’s wall chart and put the played result there, while still listing B with a full-point bye in section B. It’s neater that way, and both players get 1 point in their respective sections.
Bill Smythe
No, I think Mr. Smythe means “cross round”, so that if I’m playing my tenth game in the arena and he’s playing his fourth, can the games be matched?
My answer is that it doesn’t matter because the software could enter zero point byes in rounds 4, 5, 6, 7, 8, and 9 (or as appropriate) and pair us in round 10. In practice, an arena with a moderate number of players, say twenty, could easily have hundreds of “rounds” with almost all games unplayed.
Alex Relyea
I guess my main point is, maybe arena events should be allowed to be U.S. Chess online rated, as long as either all games are regular-ratable or all games are quick-ratable or all games are blitz-ratable – no mixing of different controls in the same tournament, and of course no time-odds games or other shenanigans.
The inevitable presence of a whole bunch of apparent cross-round pairings on the (single-section) wall chart should not present a problem, assuming the MSA software can already handle cross-round pairings anyway.
Bill Smythe
The current tournament/ratings programming only supports 32 rounds in an event.
It is possible to get around that by have the players under multiple pairing numbers, a technique that is sometimes used to enter double, triple or even quadruple round robin events where the number of rounds might exceed 32.
MSA has handled cross-round pairings in the past. I’ve had four round events where a third-round bye received a cross-round pairing that I treated as an extra (fifth round) game for another player (one way to get an unrated all four games after having a forfeit win), so you can just keep adding to a player’s rounds (if the player plays more than 32 games then clone the player).
That allows for players A and B to each play four games in the time it take player C to play one (all having one win by then), A and C to play each other while B plays D, C then plays her third through sixth games (being 3-3 by then) while B plays his sixth and seventh games (being 3-4 by then), and then in her seventh game C meets B in his eighth game. If you tried synchronizing rounds then A vs C would be round 5 (or later - probably much later for a large section) while B vs C would be round 10 (or later - probably much later for a large section).
Using cross-round pairings drastically reduced the number of rounds and spottiness of the MSA report at the cost of making it more difficult to see which games are going on at the same time (problematic even if you try to synchronize rounds between two players).
So from a reporting standpoint there isn’t any reason an arena tournament couldn’t be rated. The big problem is that the format of an arena tournament promotes quick results (three quick draws is better than one win in the same amount of time and a quick loss and four quick draws is better than a long draw and a long win) and thus distorts the normal percentage performance that a player may be expected to have against opponents of the ratings the player faced (seeing as 1.5-1.5 is better than 1-0 and 2-3 is better than 1.5-0.5). That distortion would seem to render as suspect any value the results would have on measuring a person’s relative strength in non-arena competition and is more appropriately either not rated or rated in a separate arena-only system (not enough demand for something like that).
It would seem that, in an arena-type event, it may be impossible even to define, let alone calculate, whether two games are going on at the same time or not.
Bill Smythe
That’s true of any event, all we know from the data provided in a crosstable is that person X played person Y and the result, we don’t know when the game started or ended or what else was going on. (I’d hazard a guess that there have been some simuls submitted for rating, based on how the crosstable looks.)
If an event is entered in swiss format, we might assume that round 1 was played before round 2, but that’s not guaranteed, and though we can usually figure out when there is a cross-round pairing, that doesn’t tell us anything about when it was played.
Further, I’ve had events in which we turned a swiss event into a RR due to low turnout, and just posted a list saying when X plays Y which person has white, and let the players choose when to play those games. (One such event got done an hour before the last scheduled round would normally have started.)
As I recall, Bill, you once floated an idea in which rounds 1 and 2 would be played simultaneously, then rounds 3 and 4, etc.
FIDE tries to ask when the games were played, presumably so they can check on their ‘maximum games or playing time’ per day limits, but their data format has some weaknesses in it.
If we’re going to ban rating area events, I think people need to know what they are and why they aren’t ratable. This thread has been a good start on that process.
We rate things like last round quick draws between the top 2 players, don’t we? Do THOSE games improve the accuracy of the ratings system?
People have even suggested different scoring and pairing procedures, for example, 3 draws = a win rather than 2, and those might increase or decrease the incentive to win.
That was for a 4-round tournament (or section) in which only 5 players showed up. (And there would be no “etc” – just rounds 1-2 and rounds 3-4.)
Arrange five tables (2.5 feet by 6 feet or whatever) like the blades on a 5-blade ceiling fan, with the corners touching, and the 5 short edges forming a small regular pentagon in the center. Each player sits in one of the five V-shaped areas formed by the long edges. Each player plays two games simultaneously, white vs the opponent on his left, and black vs the opponent on his right. All clocks are placed on black’s right, all facing the center small pentagon.
Pairings are as follows:
Rounds 1-2:
A vs B
B vs C
C vs D
D vs E
E vs A
Rounds 3-4:
A vs C
C vs E
E vs B
B vs D
D vs A
Double the time control. If the event was announced as G/30, play the games instead at G/60 since each player is trying to concentrate on two games. The tournament will still finish at the same time it would have with four individual G/30 rounds.
The TD stands in the small pentagon in the center, so that he can view all the clocks. The TD is not allowed a bathroom break until one of the games finishes, when that table can be removed to let the TD out.
Just kidding about that last paragraph.
Bill Smythe