Use of Technology at the British Championships

[Brian Towers]

An interesting item from Alex's report appended to the FIDE delegates report, which seems to imply that manual pairings will not be permitted in the British Championships 2017 and beyond.

The implication of which is that if the ECF are still deeply in love with ECF pairing rules (median float being the major difference, I think) then they have just over a year to get a programmer to write a new pairing program / modify an existing open source pairing program and get it FIDE ratified.

I suspect the ECF are not that emotionally attached to manual pairing as such. It is more their pairing rules.

[Roger de Coverly]

An interesting item from Alex's report appended to the FIDE delegates report, which seems to imply that manual pairings will not be permitted in the British Championships 2017 and beyond.

I'm guessing that even as Home Director, Alex feels he lacks the authority to insist that arbiters at the British Championships use the now well established programs widely used in the rest of the world.

[Stewart Reuben]

Roger >I'm guessing that even as Home Director, Alex feels he lacks the authority to insist that arbiters at the British Championships use the now well established programs widely used in the rest of the world.<

It is more likely that Alex did not want to impose his will not very experienced arbiters who are working without a fee.

The results on the lower scores in the British are important. Two championships were awarded to people who scored 7 and 6.5. Thus the pairings matter, even if they don't materially affect the Championship itelf. The well-established programs to which you refer are rather poor for this purpose. Doing the pairings manually has taken too long and led to errors. Thus the switch.

In Abu Dhabi I was told that to write a new Swiss Pairings program would be a month's work. Then simulations would have to be done, perhaps 1000 examples? And the results compared. But I pointed out that, before setting out on that route, you must decide what your objectives are.

[Roger de Coverly]

But I pointed out that, before setting out on that route, you must decide what your objectives are.

From a player's viewpoint:-

(a) the pairings are the same regardless of which arbiter or program is doing them
(b) the pairings can be available within minutes of the last reported result from the previous round.

In the same way that certain pairings can be overruled, a bias that competitors for minor titles should meet could also be incorporated.

[Stewart Reuben]

Roger >
(a) the pairings are the same regardless of which arbiter or program is doing them<

I'm not sure we are talking on the same wavelength. I believe there are several Swiss Pairing Systems which are quite different from the ubiquitous Seeded Swiss Pairing.
e.g. Lottery; Accelerated; Dubov; Floating median to median; Burstein; Welch idea of 6 groups then pairing 1v3 2v5 and 4v6 in round 1 (never tried); Reuben fantasy (never tried). Say 64 players 1v33 in the usual way. If 1 wins he keeps his place. If it is a draw, they both become 17 with the original 1 higher. If 33 wins he becomes 1. And so on throughout the tournament. this is more like what happens in a lottery.
But I agree absolutely that each system should be computerised and quickly available.
I am pretty sure, without any evidence, that a lottery system would give the 'best' overall result if repeated 1000 times. But each individual event would throw up absurdities. Of course it would fail (a).

Avoiding certain pairings is a slippery slope. Was not the British messed up this year because they avoided the Pert v Pert pairing and then had to do it much later anyway? Of course there is the unwritten FIDE Law that Israelis cannot be paired with opponents from certain Arabic countries.

[Roger de Coverly]

I'm not sure we are talking on the same wavelength.

A rule is defined, whatever it is, and the pairings follow that rule. It should be possible to computerise a rule that incorporated arbiter's discretion, that the program presents a range of potential pairings and the arbiter selects the ones considered most appropriate.

With the persistent rumours of tournaments that didn't take place or where results were fixed, FIDE could be a little more relaxed about pairings altered to optimise Norm chances.

[Stewart Reuben]

Roger >FIDE could be a little more relaxed about pairings altered to optimise Norm chances.<

I spent about 25 years getting the FIDE Title Rules more user-friendly. It would now be very seldom substantially advantageous to amend Swiss Pairings to optimise Norm chances. When doing so, I assumed that the rules that suited British chess best would also be optimal for the rest of the world.
One example no longer allowed. Two players in a Swiss require a draw in the last round to get a norm. So they are paired together and draw. There are no losers in this scenario.

[NickFaulks]

One example no longer allowed. Two players in a Swiss require a draw in the last round to get a norm. So they are paired together and draw. There are no losers in this scenario.

I can't tell whether you applaud this or lament it.

[Brian Towers]

One example no longer allowed. Two players in a Swiss require a draw in the last round to get a norm. So they are paired together and draw. There are no losers in this scenario.

I can't tell whether you applaud this or lament it.

I think the phrase " There are no losers in this scenario" gives a big hint, Nick.

[Brian Towers]

The problem with this FIDE initiative to end manual pairing is that the information on their own web site, apart from being slightly confusing due to the usual non-native English speaker authorship, is out of date - C.04.4 The endorsement procedure and the officially endorsed programs

Specifically this part:

Below is the list of endorsed programs (supported by Microsoft Windows), the pairings system they support and the date where their endorsement is going to be withdrawn unless it is confirmed by the Tromso Congress

after which follows a list showing that endorsement for several programs including the ubiquitous Swiss Manager and Swiss Master are not valid after 2014.

Interestingly Vega, which I believe Gerry Jepps used for the Frome congress, is one of only two programs which are endorsed without time limit. Time for English arbiters to read Gerry's review of Vega - http://www.chessarbitersassociation.co.uk/Vega.pdf.

[Stewart Reuben]

Nick > One example no longer allowed. Two players in a Swiss require a draw in the last round to get a norm. So they are paired together and draw. There are no losers in this scenario.
I can't tell whether you applaud this or lament it.<

I neither applaud, nor lament it. I simply reported a fact. People sometimes disagree with me about how things should be done. Provided their opinion is based in commonsense, why should I object? Of course I might continue to hold my opinion.
I was simply giving an example of how 'FIDE could be a little more relaxed about pairings altered to optimise Norm chances' as recommended by Roger.
I chose that one because: it is in the last round and cannot affect future pairings. In a Swiss, if two players draw in the last round, it is usually impossible for them both to gain financially. If SB or Bucholz is used for tiebreaks, then that draw will affect the tiebreaks of the other players. But I am sure my views on these tiebreak systems re well-known.

[Ian Thompson]

One example no longer allowed. Two players in a Swiss require a draw in the last round to get a norm. So they are paired together and draw. There are no losers in this scenario.

I chose that one because: it is in the last round and cannot affect future pairings. In a Swiss, if two players draw in the last round, it is usually impossible for them both to gain financially.

There clearly can be losers in this scenario - other players in the tournament:

1. Everyone who gets a different pairing from the one they should have got and ends up playing a stronger opponent than they should have done, or gets Black when they were due a White.

2. Anyone who gets less prize money, or misses out on a prize, as a result of the draw compared to what they would have got with a decisive result.

[Stewart Reuben]

Ian, only two other people got different pairings and they can be consulted about the change, as are the two players being offered the opportunity. Nobody gets a Black when due White. The change wouldn't otherwise be made. The two norm seekers only get paired together when they have the same score.
The way prize money is determined usually means that, two people agreeing a draw in the last round, results in their getting less prize money - if any prizes were involved. Thus the rest of the players get more.
That said, it is the reason this, and other amendments, are no longer allowed in the way Roger would like. 'FIDE could be a little more relaxed about pairings altered to optimise Norm chances'.

[Ian Thompson]

The way prize money is determined usually means that, two people agreeing a draw in the last round, results in their getting less prize money - if any prizes were involved. Thus the rest of the players get more.

That may be true, but I would think that most people are not interested in total amounts paid, but what they individually get. When you're behind the leaders, it's often the case that you want decisive results from the players above you to catch up as many losers as possible.

Here's an example:

Prizes: £1000, £800, £600, £400, £200

last round pairings:
Bd. 1: 6 v 6 points
Bd. 2: 6 v 6 points
Bd. 3: 5 v 5 points

A player on board 3 wins, reaching 6 points. The prize he wins is dependent on the results on boards 1 and 2:

1. Two draws - £200
2. Win and draw - £300
3. Two wins - £400

[Stewart Reuben]

Ian, True. But you have chosen a very shallow gradient at £1000, 800, 600, 400, 200. More common in British chess is a steeper one. £1200, 800, 500, 300, 200. Then it is
1. Two draws - £200
2. Win and draw - £250
3. Two wins - £333. So the effect you describe is less pronounced. Of course in many other countries they don't simply split the prizes where two or more people tie for first place. The most likely prize, for our player with 5 points who wins, is going then to be £200.