Short comment about accelerated pairings

[Frank Dixon]

Re: Short comment about accelerated pairings

Yes, thanks, Steve; I got my numbers wrong. 512 is right, as 2 raised to the ninth power. :)

[Jonathan Berry]

Re: Short comment about accelerated pairings

Yes, thanks, Steve; I got my numbers wrong. 512 is right, as 2 raised to the ninth power. :)

Actually, a straight Swiss will handle 512 players in 9 rounds without producing multiple perfect-score winners. Ever. We knew that. But a Haley Accelerated Pairing will not necessarily do the same. After round 1, you have 256 max winners. In round 2, the third quartile winners meet the second quartile losers. You expect this to result in a few more than 64 2-0 scores, but if every game is an upset you could have 192 2-0 scores, and there goes the neighbourhood, Martha.

Frank understands about pairings, yet he made a boo-boo about their consequences. We have also seen people who understand pairings far less than Frank does, make blanket statements: you can't do this, you can't do that, this won't work, etc etc. There might be a kind of Parkinson's (i.e., inverse) Law at work. Pairing systems deserve respect.

Phil Haley had little interest in the binary power law, as he invented HAP before Bobby met Boris, when tournaments were not huge. Haley Accelerated Pairings worked very well at the 1970 Canadian Open, with 55 players (56 registered, but one never showed up) and 10 rounds. There was a complete RR among the top five finishers, winner Bent Larsen played nobody who finished below 13th. We had our cake (one section) and we ate it too (good chess). 2 to the 10th is 1024, my favourite binary power to remember, as it is very binary (it is one-zero, one-zero and three powers of two) and helps you with 32 (squared is 1024) (fifth power is the most common length of chess tournament).

I skimmed Bryan's thoughtful comments about TvB (Top versus Bottom) against TvM (Top versus Middle), in hyper-accelerated pairings. First, it would be nice to have a pointer that describes the system used. Second, before I give those comments the full attention they deserve, one question: was TvB the full extent of what happened differently with the round 3 pairings? It kind of went by in a blur, but I saw mention of ghost point totals being changed and ....

Here's a half-formed idea. We see a lot of those in ChessTalk, but not many that take up a huge chunk of the screen: I thought of a pairing system (or maybe you'd call it a dispensation) that might be do-able using conventional software. For each round, you set a Rating Difference Limit. For the first round it might be 250 points. If your top guy is 2700, then everybody rated below 2450 is considered already to have played him. If #200 is 1450, then everybody above 1700 or below 1200 is considered to have played him. The first round pairing is whatever the computer would spit out under those conditions. AFAIR, SwissSys has a Club Team mode that might be used for this, though maybe it doesn't allow for a player to belong to multiple clubs. Anyway, where there's a will, there's a way. For subsequent rounds you might want a different RDL. For example, RDL (2) = 225 and RDL (3) = 200. I envisage decreasing RDLs until the late middle of the tournament, then backing away from them (their job, one hopes, already done) by the end. Would take a bunch of testing and prototyping before use in combat, but it would sure stop the worst effects of yo-yo in their tracks, eh? I name thee the RDL pairing system. Sieve long and chess more.

But please don't use the above without testing. Hmm, let's foresee ... caravans of low-rated players with perfect scores, untouchable by the top players because of RDL and equally untouchable by the players 100 points above them because of point group. Onwards to the next oasis and Victory.
Maybe the RDL needs to be based upon TPR rather than Rating, after the first round. Hmm....

[Gary Ruben]

Re: Short comment about accelerated pairings

Jonathan, sometimes I read these kind of posts and find myself wondering if the pairing systems are designed to serve the chess players needs OR if the chess players are pressed to accept non standard chess pairing systems for whatever obscure purpose.

[Emil Smilovici]

Re: Short comment about accelerated pairings

... sometimes I read these kind of posts and find myself wondering if the pairing systems are designed to serve the chess players needs OR if the chess players are pressed to accept non standard chess pairing systems for whatever obscure purpose.

You're petty, selfish and reg-re ... reg ... egg - damn! I can't spell this.

...and a curmudgeon too!

Best regards,
JB

[Jonathan Berry]

Re: Short comment about accelerated pairings

Jonathan, sometimes I read these kind of posts and find myself wondering if the pairing systems are designed to serve the chess players needs OR if the chess players are pressed to accept non standard chess pairing systems for whatever obscure purpose.

I'm glad you asked that question, Gary. Sure, pairing systems can have obscure purposes. For example, the FIDE pairing system is difficult for humans to understand, but it is algorithmic, so computer programs can be written that do it exactly. The same is not true of old-fashioned North American pairings. So, you have one step forward, one step back, but the net result is designed to serve the chess players: computer pairings are faster and much more technology-friendly than hand pairings.

If you're talking about the avowedly half-baked idea of RDL pairings, well, it was just an idea that hit me, and I thought I'd put it out there. If it ever sees service, it may be the first computerized pairing system that eliminates the single-section yo-yo effect. All players will appreciate that.

I once had the idea to apply Graph Theory (I didn't know what it was called at the time, I just knew the method I wanted the computer to use) and computer guru (chess player and Air Miles maven) Jan Pajak said he was interested in the idea. But then he died. Dangerous idea, eh?

For 2007 I tried to work on it. I posted on sci.math, and you can see the result here: Problem on an nxn grid. Warning: math is involved. Real math genii were discussing this problem, while I, a former math undergrad from a time when this type of math was not taught to undergrads, looked on in near amazement.

So it was all very obscure, but the underlying purpose was to come up with a program which would install any pairing system on a computer, with undoubted eventual service to all chess players. Those are two incomplete systems designed to have benefit to players. The actual completed pairing systems, scoring systems, and methods of prize division, really do have benefits for players.

Rotary telephones worked great, but now not many people use them. That's progress.

[Jonathan Berry]

Re: Short comment about accelerated pairings

Emil, your yo-yo is showing.

[Steve Douglas]

Re: Short comment about accelerated pairings

... but if every game is an upset you could have 192 2-0 scores, and there goes the neighbourhood, Martha.

Martha? I think you mean (beautiful) Doreen....

"With every wish, there comes a curse."

Steve

[Jonathan Berry]

Re: Short comment about accelerated pairings

Martha? I think you mean (beautiful) Doreen....

"With every wish, there comes a curse."

Steve

This beautiful Doreen? Sorry, I wasn't thinking of anybody. Martha was the first old-fashioned female name that knocked.

[Emil Smilovici]

Re: Short comment about accelerated pairings

Emil, your yo-yo is showing.

that brings bad memories: my in-laws live very close to the Japanese Embassy in Lima.

anyways, I thought I was on your ignore list ...

[Gary Ruben]

Re: Short comment about accelerated pairings

You're petty, selfish and reg-re ... reg ... egg - damn! I can't spell this.

...and a curmudgeon too!

Best regards,
JB

Those are some of my better qualities. :)

[Gary Ruben]

Re: Short comment about accelerated pairings

I think we're talking the difference between a regular Swiss paired event and the accelerated and hyper accelerated pairings.

Wasn't the CO in Alberta last year done with regular Swiss pairings? If so, I don't think it correct to suggest it as obsolete as a rotary telephone.

If 4 sections were desirable, probably advertising the event in that manner would have done the trick. Surely it would be easier to run 4 sections with a prize fund for each one. The problem is the total entry might have been lower. Accelerated pairings were advertised and I would think tournament players would understand what that entails.

I notice someone upset over the result of a game which ended a draw. While I consider the draw to be perfectly regular, it's not the entire reply. Frankly, if the TD has the right to decide the pairings, the players have the right to decide the result. "Looking after business" is nothing new in chess.

[Egidijus Zeromskis]

Re: Short comment about accelerated pairings

Wasn't the CO in Alberta last year done with regular Swiss pairings?

I think so.

There are normalized points distributions at CO'10 (accelerated^2) and CO'09 (normal pairings).




Don't make any far going conclusions, as the plot was not peer-reviewed :D

[David Ottosen]

Re: Short comment about accelerated pairings

if organizers want international norms to be possible in the Canadian Open in the future, with many titled international players on hand, many who have been secured using money coming from sponsors and players' entry fees, then USE SECTIONS.

I agree with this. Which is why I think that what is necessary is to re-think this:

the Canadian Open is supposed to be our premiere event, one would think, and it should be those players' best opportunity for scoring norms.

It seems very strange to me to take the Canadian Open, and try to turn a 200 person tournament into an event that is based heavily around the wishes of 5-10 people (of which likely only 1-2 might achieve the desired goal). In the first two rounds, you might see 10-20% of the field with a shot to play someone way beyond their weight class, and that's something I think every one of those players remembers and enjoys the memory of (I sure do), much moreso than the 10 players who had a shot at a norm pre-tournament and busted it halfway through with a two game losing streak or whatever.

My personal feeling is that the Canadian open should be the annual get-together of Canadian chess players, and having one big section (norms be damned) fosters that much more. Serious norm hunting tournaments and the Canadian open are, to me, two very different beasts.

[Emil Smilovici]

Re: Short comment about accelerated pairings

Those are some of my better qualities. :)

Actually I almost fell from my chair when you directed this towards someone else.
But these days I am staying away from THAT thread.

In fact I might be forced out of all the threads as my children threaten to put me in the chesstalk.com rehab

[Kerry Liles]

Re: Short comment about accelerated pairings

My personal feeling is that the Canadian open should be the annual get-together of Canadian chess players, and having one big section (norms be damned) fosters that much more. Serious norm hunting tournaments and the Canadian open are, to me, two very different beasts.

Isn't the idea of an "Open" exactly the possibility for a duffer to play a master?
Sort of like a Pro-Am in other sports?

The GMs (invited, subsidized, whatever) get a chance to win some money; the duffers get a chance to play a world-class GM that they wouldn't normally ever play.

Olympic team members can use the Open as a tune-up to get rid of the rust...

Everyone is a winner! ;)