The 2015 Galway Chess Congress will take place in the Menlo Park Hotel, from Friday 2nd to Sunday 4th October. Full details of the round times, prize fund, and so on will be given later – though we don’t expect any great variation from previous years.
However, we are thinking of changing the format of the tournament by adopting an exciting new structure. This is in response to feedback last year from a few people who thought that being required to play in sections was too rigid, particularly if they were near the top of a section. Here is an explanation of our possible new structure. We would very much welcome feedback on this idea, as well as proposals for improving it or pointing out drawbacks which we hadn’t considered. Please either post your responses here (to this page) or send them privately to our email address: galwaychess at gmail dot com [please reassemble].
Possible proposal about the Structure of the Congress
Last year we got a fair number of requests to ‘float up’ by players who were well below the appropriate rating floor; when these were refused, some did not enter. It also appeared from a discussion on a blog that there were other players who did not consider entering because the ratings bands did not suit them. Being generous to those who want to float up is not the answer, because we then get complaints when players have to play opponents below the rating floor. (At one congress last year, someone in the over-1600 tournament ended up with 5 opponents out of 6 who were under-1600; he complained and said he wouldn’t be playing again – he paid his money to play in an over-1600 tournament, he said, and was upset not to get one. He had a point.)
Most people welcome the opportunity to play people stronger than themselves, and that possibility is not there for those who are stuck at the top of a section: for instance, last year one had to go down to the tenth seed in the Major, and ninth in the Minor, before getting to somebody who had played anybody higher rated than they were. It seems that sections based on ratings bands are rather rigid, and are not giving some possible entrants what they would like. So we have wondered whether we should consider adopting a system of what could be called “porous”, rather than rigid, rating bands. This would allow people to float up automatically by winning their games. We have constructed a way of doing that, which is essentially a system of accelerated pairings; the technical details are below. We have also done a hypothetical test run to check how it might work; the cross-table of that is below.
The main drawback is that the draws for rounds 2 to 4 on this new system would look odd, as most players would not be paired against somebody with the same number of points. Also, of course, people might not like the possibility of “floating down” – though they would only do so if they lose, and also, at present, some might not find it much fun to be at the bottom of a rating band, and spend the weekend being pummelled by considerably stronger players. Finally, determining the winners of the lower sections might seem a bit arbitrary, as many of the best players will have played opponents in a higher section – as happens now with grading prizes, of course. But one can’t have porous ratings bands and also retain all the features of rigid ratings bands. Part of what we have to decide is which of the various features of rigid rating bands are important, and which can be ditched in return for other advantages.
Adopting such a change will be quite a big step, but we have some sympathy with people who don’t want to be always at the top or the bottom of their section, and this would prevent that. Hence we think it is worth consideration, at least.
The details
In essence, this is a system of accelerated pairing in one large section. The field is divided into four sections. (It can be any number; four is just a suggestion, and there may be something to be said for having six sections.) In round one, everybody plays somebody in their section, as at present. In round two, the winners in the top section play other winners in that section, as at present; so do players who draw. But the losers in the top section play the winners in the second section; the losers in the second section play the winners in the third section, and so on. The losers in the bottom section play against each other.
This system carries on like that, with players who continue to win against opponents in a higher section continuing to float up to get higher-rated opponents, in rounds 3 and 4. Then, in rounds 5 and 6, players revert to playing in their sections, so that a section winner is obtained. That would allow the excitement of having a winner in the lower sections determined by a head-to-head game, rather than by results against others, as currently with grading prizes.
The technical details (for those who can be bothered)
We construct the table of entrants as if the tournament is an Open with just one section. For doing the draw, the Swiss computer programme we use allows us to divide the field up into as many sections as we wish, of whatever sizes we wish. Each section can be awarded a notional number of “bonus” points, which is used for determining the pairings but not otherwise revealed. These bonus points can be any number (though only whole numbers) with the higher groups having to have higher numbers.
So let us suppose that there are four sections, called A to D. The number of players in sections A to C would be chosen so that each section is of the same size and has an even number of players; section D would be of that size plus any extra players left over. Players in section A get a bonus of 3 points, B get 2, C get 1, and D get nothing. Hence the draw for round 1 will pair players in group A against each other, B against each other, and so on.
Round Two. Winners in group A will now have 4 (1+3) notional points (hereafter n-points) and so will be the top pairing-group. Drawers in group A will have 3½ n-points, and so will be the second pairing group, and will play against each other. Losers in group A and winners in group B will have 3 n-points and will be the third pairing group; they will play against each other. If there are the same number of losers in A and winners in B, then all the losers in A will play against winners in B, colours permitting. (If there are more winners in B than losers in A – because there are fewer draws in round 1 in Section B than Section A – then not all winners in B will be able to float up to play someone from A; but those at the top of B will certainly be able to do so assuming that at least some games in Section A are decisive.) Drawers in group B will have 2½ n-points and so will be the next pairing group. And so on.
Round Three. Players in group A with 2 points will have 5 n-points, and be the top pairing group. 1½ points in A will have 4½ n-points and be the second pairing group. 1 point in A and 2 points in B will both give 4 n-points; since there will almost certainly be more 1-pointers in A than 2-pointers in B, all the Bs will float up, whilst some of the As will play each other. 3½ n-points will be those with ½ a point in A and 1½ in B. The next group is interesting: it will contain As with 0 points, Bs with 1 point, and Cs with 2 points. A pairing of A v C is then possible, but very unlikely.
Round 4. The same principles apply. It will now be logically possible for a player from D who has 3 points to play one from A who has 0 – but in practice that is almost certain not to occur.
Rounds 5 & 6. We suggest that the accelerated pairings ends here, and the rest continues within the sections. This can be done by changing the “bonus” points to the sections from 3/2/1/0 to, say, 30/20/10/0, and then continuing to use the accelerated pairing function; this can be done easily.
Finally, there is the question of how the division into sections is done. In the example here it is determined solely by the profile of the entries: the top quarter go in section A, the second quarter in section B, and so on. There is something to be said for this. It means that players are not stuck in the same section in all tournaments, playing the same opponents (or same sort of opponents). On the other hand, it is quite a radical departure from the current system we are all used to, and so might not be welcomed by some players. Players wouldn’t know when entering which section they would be in, which might put some of them off – even though this would not be as significant as with rigid sections. And the cut-off point between sections would be an unusual number, such as 1274, which might not please those who have a fetish about the significance of zeros at the end of a number.
We could have predetermined sections, as at present, but the system works a lot better with equal-sized sections. So our present recommendation is that the divisions between the sections should be determined only by the profile of the entries. However, if our feedback indicates considerable support for pre-determined section divisions (as at present) then we will adopt that.
Simulation
To test all this, we did a simulation. 117 entries were chosen, to see how it would cope with an awkward number. The entries were fictional; to make it easier to follow, all players in section A were given a ‘name’ starting in A, B in B, etc; the top half of A starts Aa and the lower half Ab. The ratings were just made up to create an ordering. Below is the cross-table up to the pairings for round 5; after that, everything would proceed as we are used to.
Results had to be included, in order to do the draws for later rounds. To make the simulation realistic, statistics from last year’s Congress on the frequency of draws and of wins for the lower-rated player were copied. In round 1 this year there were 50 games, of which 6 were draws and 4 were wins by the lower-rated player; hence every eighth game in the simulation was made a draw and every thirteenth a win for the lower-rated. (A random device for each game should probably have been used, with those odds, but that would have been too time-consuming.) In round 2 there were 51 games; 10 were draws and 16 wins for the lower-rated, so this was followed with every fifth game being a draw and every third a lower-rated win. (This seems high, but it seemed better to just follow it.) These ratios were kept for later rounds.
Here is the cross table showing the pairings, and also who the prize winners would be, given these hypothetical results (GP = grading prize).
Cross Table
Galway 2015 Simulation - Round 6
Cross Table
No Name Loc Total 1 2 3 4 5 6 Prizes
1. Aaa, 2400 6 16:W 10:W 12:W 2:W 5:W 14:W 1st
2. Aab, 2390 5 17:W 9:W 7:W 1:L 15:W 12:W 2nd
3. Aac, 2380 3 18:W 12:L 24:L 37:L 30:W 22:W
4. Aad, 2370 3.5 19:W 11:W 15:L 21:L 29:W 23:D
5. Aae, 2360 3.5 20:W 14:D 8:W 15:W 1:L 18:L
6. Aaf, 2350 3 21:W 15:L 25:W 23:W 14:L 9:L
7. Aag, 2340 4 22:W 28:W 2:L 24:D 18:D 16:W
8. Aah, 2330 2.5 23:D 13:W 5:L 29:L 20:L 27:W
9. Aai, 2320 4 24:W 2:L 28:W 18:L 22:W 6:W
10. Aaj, 2310 2 25:W 1:L 21:L 42:W 23:L 13:L
11. Aak, 2300 3.5 26:W 4:L 33:D 31:D 25:D 21:W
12. Aal, 2290 4 27:W 3:W 1:L 44:W 21:W 2:L
13. Aam, 2280 3 28:L 8:L 51:L 63:W 26:W 10:W
14. Aan, 2270 4 29:W 5:D 23:D 27:W 6:W 1:L
15. Aao, 2260 4 30:W 6:W 4:W 5:L 2:L 24:W
16. Aba, 2250 3 1:L 23:L 48:W 43:W 27:W 7:L
17. Abb, 2240 2.5 2:L 32:D 45:W 33:W 24:L 25:L
18. Abc, 2230 4.5 3:L 34:W 42:W 9:W 7:D 5:W 3rd
19. Abd, 2220 1.5 4:L 33:L 53:D 47:L 28:D 26:D
20. Abe, 2210 3 5:L 36:W 44:L 50:D 8:W 29:D
21. Abf, 2200 3 6:L 35:W 10:W 4:W 12:L 11:L
22. Abg, 2190 2 7:L 38:D 54:D 58:W 9:L 3:L
23. Abh, 2180 3.5 8:D 16:W 14:D 6:L 10:W 4:D GP
24. Abi, 2170 3.5 9:L 40:W 3:W 7:D 17:W 15:L GP
25. Abj, 2160 3.5 10:L 37:W 6:L 51:W 11:D 17:W GP
26. Abk, 2150 1.5 11:L 42:L 50:L 67:W 13:L 19:D
27. Abl, 2140 2 12:L 43:W 32:W 14:L 16:L 8:L
28. Abm, 2130 2.5 13:W 7:L 9:L 54:L 19:D 30:W
29. Abn, 2120 3 14:L 45:D 39:W 8:W 4:L 20:D
30. Abo, 2110 1.5 15:L 44:L 56:W 32:D 3:L 28:L
31. Baa, 2100 5 46:D 56:W 38:W 11:D 44:W 34:W 1st
32. Bab, 2090 2.5 47:W 17:D 27:L 30:D 48:L 41:D
33. Bac, 2080 3.5 48:W 19:W 11:D 17:L 34:L 54:W
34. Bad, 2070 4 49:W 18:L 55:W 74:W 33:W 31:L 3rd=
35. Bae, 2060 3 50:W 21:L 58:L 72:W 56:W 46:L
36. Baf, 2050 4.5 51:W 20:L 68:D 53:W 47:W 37:W 2nd
37. Bag, 2040 4 52:W 25:L 63:W 3:W 54:W 36:L 3rd=
38. Bah, 2030 4 53:W 22:D 31:L 68:W 50:D 48:W 3rd=
39. Bai, 2020 2.5 54:D 46:W 29:L 61:L 59:L 60:W
40. Baj, 2010 3 55:W 24:L 74:L 64:L 53:W 49:W
41. Bak, 2000 2.5 56:L 54:L 72:L 93:W 52:W 32:D
42. Bal, 1990 3.5 57:W 26:W 18:L 10:L 49:W 50:D
43. Bam, 1980 2 58:W 27:L 73:W 16:L 51:L 56:L
44. Ban, 1970 3 59:W 30:W 20:W 12:L 31:L 51:L
45. Bao, 1960 3.5 60:W 29:D 17:L 46:D 58:D 59:W
46. Bba, 1950 4 31:D 39:L 62:W 45:D 60:W 35:W 3rd=
47. Bbb, 1940 2.5 32:L 61:D 75:W 19:W 36:L 58:L
48. Bbc, 1930 3 33:L 64:W 16:L 86:W 32:W 38:L
49. Bbd, 1920 2 34:L 63:L 78:W 73:W 42:L 40:L
50. Bbe, 1910 3.5 35:L 66:W 26:W 20:D 38:D 42:D GP
51. Bbf, 1900 4 36:L 65:W 13:W 25:L 43:W 44:W 3rd=
52. Bbg, 1890 1 37:L 68:L 80:D 77:D 41:L 55:L
53. Bbh, 1870 2.5 38:L 67:W 19:D 36:L 40:L 57:W
54. Bbi, 1860 3 39:D 41:W 22:D 28:W 37:L 33:L
55. Bbj, 1850 2 40:L 71:W 34:L 81:L 57:L 52:W
56. Bbk, 1840 3 41:W 31:L 30:L 59:W 35:L 43:W
57. Bbl, 1830 1 42:L 73:L 81:L 94:L 55:W 53:L
58. Bbm, 1820 3.5 43:L 72:W 35:W 22:L 45:D 47:W GP
59. Bbn, 1810 2.5 44:L 75:D 77:W 56:L 39:W 45:L
60. Bbo, 1800 2 45:L 74:L 84:W 103:W 46:L 39:L
61. Caa, 1790 5.5 76:W 47:D 70:W 39:W 64:W 74:W 1st
62. Cab, 1780 2.5 77:D 85:W 46:L 89:L 83:W 79:L
63. Cac, 1770 2.5 78:W 49:W 37:L 13:L 79:L 77:D
64. Cad, 1760 3 79:W 48:L 83:W 40:W 61:L 81:L
65. Cae, 1750 3.5 80:W 51:L 86:L 88:W 70:D 78:W
66. Caf, 1740 3.5 81:W 50:L 110:D 85:W 80:L 73:W
67. Cag, 1730 4 82:W 53:L 91:W 26:L 76:W 70:W 3rd=
68. Cah, 1720 4.5 83:W 52:W 36:D 38:L 89:W 80:W 2nd
69. Cai, 1710 1 84:L 70:L 99:L 105:L 90:D 88:D
70. Caj, 1700 3 85:D 69:W 61:L 92:W 65:D 67:L
71. Cak, 1690 3.5 86:W 55:L 103:L 91:W 77:W 82:D
72. Cal, 1680 3 87:W 58:L 41:W 35:L 82:L 84:W
73. Cam, 1670 2.5 88:W 57:W 43:L 49:L 86:D 66:L
74. Can, 1660 4 89:W 60:W 40:W 34:L 81:W 61:L 3rd=
75. Cao, 1650 2.5 90:W 59:D 47:L 80:L 78:L 87:W
76. Cba, 1640 2 61:L 92:D 98:W 110:D 67:L 83:L
77. Cbb, 1630 2.5 62:D 84:W 59:L 52:D 71:L 63:D
78. Cbc, 1620 2.5 63:L 94:W 49:L 95:D 75:W 65:L
79. Cbd, 1610 4 64:L 91:L 100:W 99:W 63:W 62:W 3rd=
80. Cbe, 1600 3.5 65:L 96:W 52:D 75:W 66:W 68:L GP
81. Cbf, 1590 4 66:L 95:W 57:W 55:W 74:L 64:W 3rd=
82. Cbg, 1580 3.5 67:L 98:D 93:D 106:W 72:W 71:D GP
83. Cbh, 1570 3 68:L 99:W 64:L 116:W 62:L 76:W
84. Cbi, 1560 2 69:W 77:L 60:L 108:L 85:W 72:L
85. Cbj, 1550 2.5 70:D 62:L 88:W 66:L 84:L 90:W
86. Cbk, 1540 2.5 71:L 100:W 65:W 48:L 73:D 89:L
87. Cbl, 1530 2 72:L 103:L 102:D 101:D 88:W 75:L
88. Cbm, 1520 1.5 73:L 102:W 85:L 65:L 87:L 69:D
89. Cbn, 1510 3.5 74:L 117:D 101:W 62:W 68:L 86:W GP
90. Cbo, 1500 1 75:L 110:L 108:L 107:D 69:D 85:L
91. Daa, 1490 2.5 104:W 79:W 67:L 71:L 101:D 92:L
92. Dab, 1480 3.5 105:W 76:D 117:W 70:L 102:L 91:W
93. Dac, 1470 3 106:D 114:W 82:D 41:L 116:L 113:W
94. Dad, 1460 5 107:W 78:L 105:W 57:W 108:W 103:W 1st
95. Dae, 1450 4.5 108:W 81:L 109:W 78:D 113:W 102:W 2nd
96. Daf, 1440 3 109:W 80:L 116:L 117:W 105:W 108:L
97. Dag, 1430 2 110:L 109:L 114:L 111:W 112:L 98:W
98. Dah, 1420 1.5 111:W 82:D 76:L 113:L 100:L 97:L
99. Dai, 1410 4 112:W 83:L 69:W 79:L 115:W 110:W 3rd=
100. Daj, 1400 3.5 113:W 86:L 79:L 109:W 98:W 116:D
101. Dak, 1390 3.5 114:D 106:W 89:L 87:D 91:D 107:W
102. Dal, 1380 3.5 115:W 88:L 87:D 114:W 92:W 95:L
103. Dam, 1370 4 116:W 87:W 71:W 60:L 110:W 94:L 3rd=
104. Dba, 1360 2 91:L 111:D 113:D 115:L 109:W 114:L
105. Dbb, 1350 2.5 92:L 112:W 94:L 69:W 96:L 115:D
106. Dbc, 1340 3 93:D 101:L 111:W 82:L 114:D 112:W
107. Dbd, 1330 2.5 94:L 113:L 112:W 90:D 117:W 101:L
108. Dbe, 1320 4 95:L 115:W 90:W 84:W 94:L 96:W 3rd=
109. Dbf, 1310 2 96:L 97:W 95:L 100:L 104:L :W
110. Dbg, 1300 3 97:W 90:W 66:D 76:D 103:L 99:L
111. Dbh, 1290 1.5 98:L 104:D 106:L 97:L :W 117:L
112. Dbi, 1280 2 99:L 105:L 107:L :W 97:W 106:L
113. Dbj, 1270 2.5 100:L 107:W 104:D 98:W 95:L 93:L
114. Dbk, 1260 3 101:D 93:L 97:W 102:L 106:D 104:W
115. Dbl, 1250 2.5 102:L 108:L :W 104:W 99:L 105:D
116. Dbm, 1240 3.5 103:L :W 96:W 83:L 93:W 100:D GP
117. Dbn, 1230 2.5 :W 89:D 92:L 96:L 107:L 111:W
[…] in part a result of this tension around how to set rating bands. Chief organiser Pete Morris is now considering a radical solution and is looking for feedback from intending competitors – I reckon chessplayers are by and […]
I got an email from galway chess club asking for my feedback on this new exciting plan for the next congress,so I shall give it .
You can’t please everyone in a tournament and don’t know if it is a great idea to change entry rules just to satisfy a few individuals ,some of whom, whose egos far exceeds their talent ,But I think the proposal on offer could scare away players in top section as they don’t know what they are getting in . Here is my idea for an exciting alternative !, create up to maximum 3 wild entry cards to the top section , players who want to play up apply for it and pay entry fee and play against each other just before the start of main tournament ( depending on applicant numbers ) one or two quick rapid games ( 15 minutes each ) and the top 3 go through into masters and rest play in their own section , so good players win wild card entry places .
I also think playing 1 or 2 unrated rapid games in open tournaments before start of main competition and using the points gained for pairing for the main competition would speed things up and encourage more high rated players to take part .
Hi Shahram,
Thanks for your response. Yours is an interesting idea, but unfortunately there is no time to have games before the start of the first round: most people who travel only arrive with a few minutes to spare as it is, if that. Presumably the thing to do if this plan was adopted is to have the rapid games instead of the first round, with the tournament proper starting on the Saturday. This is done in several elite tournaments now (such as the one in Zurich which just finished a few minutes ago), but they are all-play-alls; I’m not aware of it in a large Swiss.
I suspect that reducing the Galway tournament to 5 rounds would not be popular, though our new proposal is somewhat similar to yours, in that people get to play stronger opponents by winning games, rather than by pleading to the tournament organizers. So we’re thinking along similar lines.
Hi Pete
Thanks for the reply ,If arrival time is an issue , then playing up candidates can battle out among themselves for wild card places in rapids on friday night instead of round 1 and can player longer & more rounds as a result and whether successful or not ,get 1/2 point bye for round 1 and start their tournament proper on saturday morning , every one else in their appropriate section will play 6 rounds as before and not effected .I think this idea will give deserving individuals ( many underrated juniors ) a chance to play up and encourage more people to attend for a chance of winning wild card entry places and it’s simple with minimum disruption to the tournament.The idea of players beaming in and out of sections as if on board star trek is confusing, having said that it is probably fine as well . I also want to say ,I did enjoy playing in galway congress last year and it was a nice venue and well organised and hope to be able play again later this year.
[…] from the section above theirs. A full explanation of how this would work can be found here. The organisers very much welcome feedback on this idea, as well as proposals for improving it or […]
Hi, Pete
With regard to you new ideas regarding a more “porous” draw two items come to mind: firstly as you cannot use Swiss Master/Swiss Perfect to do your draws and you probably will have only about an hour between rounds, time will tight to do the draw manually.
The second item is that at the moment the ICU rating system only accepts results in Swiss Master/Swiss Perfect format so how will you submit you results for rating.
I understand where you are coming from as I was on the receiving end as top seed in the Ennis U1400 last year and my rating took a hiding.
Regards,
Ted.
Hi Ted,
Thanks for your comments. We can do this in SwissPerfect: the simulation here was done in SwissPerfect, and the method is explained in the text. You’re right: I wouldn’t want to be doing the draw by hand – though when I tried to explain this to Stewart Reuben he thought that any competent organizer (of his generation!) could do a draw for a tournament of 500 by hand in 20 minutes! But I won’t be putting that to the test.
Hope you come to Galway and win back some of the rating points you left in Ennis!
Hi Pete,
You may get a lot of flak from people who don’t understand in its entirity how the system will eventually work. However trying something new should always be applauded (providing it’s not too exotic for the tastes of the proletariat) because it is only through trial and error that we can make the correct decisions on the future of tournaments. If it doesn’t work it doesn’t work but someone has to be the pathfinder so good luck
Your porous ‘Snakes & Ladders’ system sounds complicated at first read but the amount of work you have put in to develop and test it seems impressive – it has to be worth a go in a real event. And if it’s a success then you can write the mechanics & results up for New In Chess!
Thanks Nick. But would it attract you to play this year? That’s far more important than achieving fame in New in Chess!