40/25, SD/10+5

I guess we don’t really disagree all that much. Certainly a playing TD would further complicate the problems caused by the double time control segments. Let him change the TC to G/30 inc/5 and then play in his own event if necessary to even things out.

Bill Smythe

Having heaped praise upon the rotating triangles, I’d now like to discuss color allocation.

First, please note that the rotating triangles include a complete round-robin within each trig (plus one inter-trig pairing for each player). Thus, 1-2-3 all play each other, and 4-5-6 all play each other.

IMHO, when assigning colors in the “rotating triangles” pairings, the following criteria are all desirable, in decreasing order of importance:

  1. No player should end up with three whites or three blacks (WWW or BBB).
  2. Each player should have equal colors (one W and one B) in his two intra-trig pairings. For example, if 1 has white vs 2, then 2 should have white vs 3, and 3 should have white vs 1. And similarly for 4-5-6.
  3. If possible, all players should alternate colors round by round, i.e. WBW or BWB.
  4. If possible, all players should alternate colors between rounds 1 and 2, i.e. no player should start with WW or BB.
  5. If possible, all players should alternate colors between rounds 2 and 3, i.e. no player should end with WW or BB.

Comments:

  • (a) is a no-brainer.
  • The two intra-trig pairings are the most important for each player, hence (b).
  • In view of (b), it is impossible to achieve (c).
  • (d) is more important than (e). If (d) is violated, then some players will be out of balance by 2 after two rounds.

In the organizer’s paper, he suggests the following colors:

round 1: 1-4 and 3-2 and 5-6
round 2: 2-5 and 3-1 and 4-6
round 3: 6-3 and 1-2 and 5-4

I would prefer the following:

round 1: 1-4 and 2-3 and 6-5
round 2: 5-2 and 3-1 and 4-6
round 3: 6-3 and 1-2 and 5-4

Notes:

The organizer’s version violates criterion (b).
Both versions satisfy criterion (c) for 4 of the 6 players.
The organizer’s version violates (d) but satisfies (e).
My version satisfies (d) but violates (e).
If you disagree that (d) is more important than (e), use my version but switch colors in the 5-2 pairing.

Of course, all this depends on the coin toss. If the coin comes up tails, simply reverse all colors in the above discussion.

Bill Smythe

He has also proposed a new method of pairing players that is supposedly an improvement over the Swiss System

mukilteochess.pbworks.com/w/page … ournaments

What do you think?

What do you think?

It really sucks.

Actually, no. Saying it really sucks is offensive to a lot of perfectly fine things that really suck.

Adding rating calculations to the pairing algorithm makes it easier for the TD to err and harder for players to check the pairings.

At one time in the relative dark ages of US Chess (pre-Fischer boom), it was standard practice to recalculate a player’s rating after each round in order to pair the next round. This practice was abandoned for very good reasons that haven’t changed with time.

I asked you first

If your friend wants to discuss his ideas, how about he post them? If you don’t think it has merit, why are you asking? If you think it has merit, then God help the people in your area.

I think it’s obvious that the person coming up with all these really bad ideas is a very inexperienced club TD.

Why are there so many bad ideas coming out of the PNW in the past few years? I can’t blame PNW for 10 second delay, but I’m not at all surprised it’s been adopted there.

Alex Relyea

Some people in this thread who are more open minded have said the “rotating triangle blocks” isn’t a bad idea so maybe these ideas aren’t as bad as you think.

Also, CCA’s idea to use 10 second delay was an excellent idea. At the G/45 Quad event I run, it has helped mitigate time pressure without expanding the round times much at all.

Does CCA use 10 second delay in that short of a time control? A good idea in the right setting can be a horrible idea in the wrong setting.

Since Micah is steadfastly refusing to endorse the horrible rating based pairing idea, the logical inference is that he too fails to see any redeeming value in the idea. If that’s true, this would be a rare case where I agree with him.

As long as the subject of unusual ideas from the NW is on the table, Clint Ballard’s BAP scoring system fits into the discussion. That was an OK idea for a fun event, but it certainly isn’t something many people would like to see become part of our standard fare.

You have heard the saying about blind squirrels, haven’t you?

I hope your friend got some college credit for writing the program, as that was otherwise a truly phenomenal waste of time otherwise. It’s nice for people to think “outside the box”, but it does help to have some idea what the “box” is and why the “box” is the way it is. A very inexperienced club TD doesn’t have that knowledge. The basic Swiss rules are actually relatively simple—it’s the tough cases that can come about when you’re later in the tournament and the score groups get smaller that give rise to all the special rules. To pitch what’s basically the prime rule governing the Swiss (players as much as possible play people with the same score) overboard because it’s hard to write a program to handle small score groups (yes, it is!) is using rather backwards logic.

Yes. For example, the fastest schedules of the World Open this year used G/35;d10.

The author of the above paper seems to be an adult (16 or over) USCF member, so should be capable of writing posts in this forum.

Travis, how about posting some of your ideas directly in this forum? You have some interesting thoughts.

Before I read the paper, and relying only on comments on this forum, my first thought was “Oh gawd, another idea that relies on single-event performance ratings. What do you do about those who have scored all wins or all losses?” But upon reading the paper (i.e. going to the original source) I realized that this question had already been anticipated, and solved by grouping the 100% and 0% groups by score, and all other groups by performance rating.

Overall I’d say the idea has some merit.

But like all new ideas, it doesn’t gain wide acceptance right away. Chess players are notable stick-in-the-muds when it comes to any deviation from convention. We’ve already seen the usual curmudgeon response, right here in this thread, just within the past 24 hours since the idea was introduced.

So new ideas need to be tried first in events with a low entry fee, and with no prizes or minimal prizes. That’s unfortunate, but good luck with your innovative ideas! :slight_smile:

Bill Smythe

There are a number of potential problems that can be listed for the performance rating 1vs2 pairings and prizes.

Here’s one that is applicable to any 1vs2 pairings, which is a reason why 1vs2 is used only in rare cases where it happens to make sense (i.e. normal Swiss pairings make a heck of a lot more sense than 1vs2 in any decent sized tournament with a prize fund - I do occasionally use 1vs2 pairings for club side events, and the idea behind quads has some similarity to the idea behind 1vs2 pairings).

Take a four round tournament with 20 players, two of them being B-players and the others under 1200. The percentage of draws is generally much lower at the lower rating levels, so if the round one game between B-players is a draw that gives a very high chance of the two B-players finishing 3.5-0.5 while the winner of the tournament is a 4-0 that never had to face them (fairly likely result is 1 @ 4-0, 2 @ 3.5-0.5, 3 @ 3-1, 8 @ 2-2, 5 @ 1-3, 1 @ 0-4).
Take the US Open with more than 300 players and only nine rounds. There will be a higher percentage of draws every round and a 6-3 results may well tie for first place. The most critical games may be in the first round while the final round can see huge mismatches on the top boards.

Actually, not really. If one wants a “maximize competitive games, score’s no object” pairing system, 1-2, 3-4 (which he’s using for the perfects and zeros) is simpler and does a better job.

Actually, he’s using 1-vs-2 pairings for the middle group, too – except that he arranges that group by performance rating so far, rather than by score and pre-event rating.

A somewhat similar idea would be to use 1-vs-2 pairings in round 1, then in subsequent rounds add 100 points to each player’s rating for each win, and subtract 100 for each loss, and continue to pair (the entire field) 1-vs-2 with these adjustments.

Bill Smythe

He could also do 1 vs 2 pairings arranged by height, which would make about as much sense.

The PR isn’t an especially useful measure conditioned on score. It’s a miserable piece of junk when done unconditionally. A system that would by construction pair someone with a minus score paired up every round against a player with a plus score paired down every round does not have merit (and does so for no other reason other than being easier to program).

Every now and then you find people that are absolutely certain they’ve discovered the greatest thing since sliced bread. Sometimes you just have to let them try it (very preferably in a non-rated event) so they can see if it really is improved sliced bread, and also see whether or not they need to do the equivalent of promoting that sliced bread to people on the Atkins meal plan.

To echo Tom’s point:
There are a lot of different criteria we could use to value various pairing methods but ease of programming would not make my top 100.

Mike Regan

Going back to the question of the time control I am not in favor of it. However, do recall in the past 1 or more players stating a preference for 2 time periods over a straight sudden death time control.

Larry S. Cohen