Delay vs. Increment Outlier Games

[Steve_Immitt]

I was having a discussion with Ben Finegold at the Mid-America Open recently about the advantages and disadvantages of delay vs. increment, particularly in “worst-case” scenario situations with a game which goes on and on. Ben was nice enough to give me some details of games of his and others he knew which went very long, and were using increment. He gave me permission to repost his email where he gave me the details, and so I do that here. Bill Goichberg also made a post on his website about long games using increment vs. delay, and he also said I could post the email but only that parts I agreed with, because he said he was too busy and had no time to engage in a lengthy discussion but that presumably I could respond to the various objections raised to his email. I’m not sure I agree with all the points he made, but I only post this because I think this part of the debate of increment vs. delay, the case of the “outlier” long game, may not have been considered fully.

I have a feeling, and it’s not supported by any of my personal experience, but I have a feeling that a very long game using increment has a greater chance of usurping the time for the next round than a similarly long game using delay? Admittedly, games of such inordinate length would be ridiculously rare, but they nevertheless could arise (see Ben’s email below), like at the end of the penultimate round of a large tournament, for example, when everyone is waiting for the last round to start.

I’m actually very busy right now too (too busy to even be posting anything at all, truthfully), but I think the topic is very timely and important. I’ll try to check responses on this topic, but might not get a chance to reply as often as I would like.

In a message dated 3/24/2012 4:54:37 P.M. Eastern Daylight Time, gmfinegold@gmail.com writes:
Hi Steve,

Actually, it was a RR. I was playing IM Jan van de Mortel. The other games of the next round started…and we started the next round maybe 40 minutes after everyone else. It was 2 rounds a day. I think it was 40/90, g/30… 30 sec inc throughout… but, it was years ago…so not sure. I think if the rounds are something like 10AM and 6PM, this won’t happen… but I think they were something like 10AM and 3PM! lol

In a RR years ago…I played GM Mitkov and GM Georgiev in rounds 1-2, and EACH game took 6 hours!! … 2 draws. My friend Dushyanth was going to eat lunch with me between games… he fell asleep… and left to catch a plane! lol

I played 2 LONG games in US Championships… which had inc… 139 and 153 moves (back to back years!)… against Bercys and Freidel. Both draws… one of the games went about 90 minutes longer than any other that round… and all the arbiters and press people had to eat dinner VERY late. They were not happy.

I think in FIDE tournaments, your hands are tied… increment is normal… in USCF events, that are not FIDE rated, I never see inc! so, a little odd… I think 90% of GMs in the world ALWAYS use increment, and 90% of the USCF rank and file always use delay.

Here is the vd Mortel game:

[Event “Chicago Spring”]
[Site “Chicago”]
[Date “2005.04.17”]
[Round “3”]
[White “Van de Mortel, Jan”]
[Black “Finegold, Benjamin”]
[Result “0-1”]
[ECO “D00”]
[WhiteElo “2410”]
[BlackElo “2522”]
[PlyCount “184”]
[EventDate “2005.04.16”]
[EventType “tourn”]
[EventRounds “11”]
[EventCountry “USA”]
[EventCategory “8”]
[Source “ChessBase”]
[SourceDate “2005.01.01”]

1. d4 Nf6 2. Nf3 g6 3. Nc3 d5 4. Bf4 Bg7 5. e3 O-O 6. Be2 b6 7. Ne5 Bb7 8. h4
   Nbd7 9. f3 Nxe5 10. Bxe5 Ne8 11. Bxg7 Nxg7 12. f4 e5 13. fxe5 f6 14. Bf3 fxe5
2. dxe5 c6 16. Qd2 Qe8 17. O-O-O Rd8 18. h5 gxh5 19. Qd4 c5 20. Qh4 Qxe5 21.
   Rde1 Rde8 22. Bxh5 Nxh5 23. Qxh5 Qxh5 24. Rxh5 d4 25. Nd1 Rf7 26. Rg5+ Rg7 27.
   Rxg7+ Kxg7 28. Kd2 Re4 29. exd4 Rxd4+ 30. Kc1 Kf6 31. Rh1 Kg6 32. Rf1 h5 33.
   Ne3 h4 34. Rf2 Bc8 35. b3 Be6 36. Rf8 Kg5 37. a3 Rf4 38. Re8 Bf7 39. Re7 a5 40.
   Kd2 Rf6 41. Re4 Rd6+ 42. Ke2 Rd4 43. Re5+ Kf6 44. Rf5+ Kg6 45. a4 Re4 46. Rf1
   Be6 47. Rf8 Kg5 48. Kf3 Rd4 49. Ke2 c4 50. bxc4 Bxc4+ 51. Kf3 Be6 52. Rb8 Rd6
3. Ke4 Bd7 54. c4 Bxa4 55. Nd5 Bb3 56. Ke5 Rg6 57. Ne7 Rg7 58. Nf5 Rc7 59.
   Rg8+ Kh5 60. Ng7+ Kh6 61. Ne6 Bxc4 62. Rh8+ Rh7 63. Re8 Bxe6 64. Rxe6+ Kg5 65.
   Rxb6 Re7+ 66. Kd4 Ra7 67. Rb3 a4 68. Ra3 Kg4 69. Ke4 Ra5 70. Kd4 Kf4 71. Rf3+
   Kg4 72. Ra3 Rf5 73. Ke3 Rf4 74. Ke2 Rb4 75. Kf2 Kf4 76. Rf3+ Ke5 77. Re3+ Kd4
4. Re8 Kc4 79. Ke3 Kb3 80. Kd2 Kb2 81. Rc8 a3 82. Rc2+ Kb3 83. Rc3+ Ka4 84.
   Rc8 a2 85. Kc2 Rg4 86. Kb2 Rxg2+ 87. Ka1 Rg1+ 88. Kxa2 Rg2+ 89. Kb1 h3 90. Rh8
   h2 91. Kc1 Rg1+ 92. Kd2 h1=Q 0-1

Also… for kicks… here are the two VERY VERY long draws I had with inc time controls…

[Event “USA-ch GpA”]
[Site “San Diego”]
[Date “2006.03.02”]
[Round “1”]
[White “Finegold, Benjamin”]
[Black “Bercys, Salvijus”]
[Result “1/2-1/2”]
[ECO “E97”]
[WhiteElo “2563”]
[BlackElo “2358”]
[PlyCount “305”]
[EventDate “2006.03.02”]
[EventType “swiss”]
[EventRounds “9”]
[EventCountry “USA”]
[Source “ChessBase”]
[SourceDate “2006.06.07”]

1. d4 Nf6 2. c4 g6 3. Nc3 Bg7 4. e4 d6 5. Be2 O-O 6. Nf3 e5 7. O-O Nc6 8. d5
   Ne7 9. b4 Nh5 10. Qc2 f5 11. Ng5 Nf4 12. Bxf4 exf4 13. Rae1 fxe4 14. Ncxe4 Nf5
2. Qd1 Nd4 16. Nf3 Nxe2+ 17. Rxe2 h6 18. Rfe1 Bg4 19. Qd3 Qd7 20. h3 Bh5 21.
   c5 Rae8 22. b5 Kh8 23. Qa3 Bxf3 24. Qxf3 Qxb5 25. cxd6 Qxd5 26. dxc7 Qc6 27.
   Qg4 f3 28. gxf3 Qxc7 29. Qxg6 Re5 30. h4 Qe7 31. Ng3 Rxe2 32. Rxe2 Qxh4 33. Re8
   Qf4 34. Re7 Qf6 35. Qxf6 Rxf6 36. Rxb7 Rxf3 37. Kg2 Rf8 38. f4 Rxf4 39. Rxg7
   Ra4 40. Rd7 Rxa2+ 41. Kf3 Kg8 42. Ne4 Kf8 43. Kf4 Ra5 44. Nf6 h5 45. Rh7 h4 46.
   Kg4 h3 47. Rxh3 Kf7 48. Ne4 Ke6 49. Rh6+ Kd7 50. Kf4 Kc7 51. Nc3 Kb7 52. Ke4
   Rc5 53. Kd4 Rc6 54. Rh5 Rc8 55. Ne4 Rc6 56. Kd5 Rc7 57. Rh6 Rd7+ 58. Kc5 Rc7+
3. Kb5 Re7 60. Nc5+ Kc7 61. Rg6 Rh7 62. Ne6+ Kb7 63. Nc5+ Kc7 64. Rc6+ Kb8 65.
   Na6+ Kb7 66. Rd6 Kc8 67. Nb4 Kc7 68. Kc5 Kb7 69. Nd5 Rg7 70. Rh6 Rd7 71. Rh1
   Rg7 72. Rh2 Kc8 73. Kd6 Rg6+ 74. Ke7 Rg7+ 75. Kf6 Rd7 76. Ke6 Rg7 77. Ne7+ Kc7
4. Rc2+ Kb6 79. Ra2 a5 80. Nf5 Rg4 81. Kd5 Rg5 82. Ke4 Kb5 83. Ne3 Rh5 84.
   Rb2+ Kc5 85. Rb8 Rh4+ 86. Kd3 Rh3 87. Rc8+ Kb5 88. Kd4 Rh4+ 89. Kc3 Rh3 90. Re8
   Rh5 91. Nc2 Rh3+ 92. Kb2 Rh5 93. Rc8 Rh2 94. Rc3 Ka4 95. Rc5 Rg2 96. Rh5 Rg4
5. Ne3 Rg3 98. Re5 Kb4 99. Nc2+ Ka4 100. Nd4 Rg8 101. Nc6 Rg5 102. Re4+ Kb5
6. Nd4+ Kb6 104. Kb3 Rg1 105. Nc2 Rg5 106. Ka4 Rg2 107. Na3 Kc5 108. Rc4+ Kb6
7. Rh4 Kc5 110. Rh5+ Kd4 111. Nb5+ Kc4 112. Kxa5 Rg6 113. Rh1 Kc5 114. Rc1+
   Kd5 115. Nc7+ Kd6 116. Kb6 Kd7+ 117. Kb7 Rd6 118. Rh1 Rd2 119. Rh7+ Kd6 120.
   Kb6 Rb2+ 121. Nb5+ Ke6 122. Kc5 Rc2+ 123. Kd4 Rd2+ 124. Ke3 Rb2 125. Nd4+ Kd5
8. Rh5+ Kc4 127. Ne6 Rb5 128. Rh4+ Kc3 129. Re4 Rb1 130. Nf4 Rb3 131. Nd5+
   Kb2+ 132. Kd4 Rb8 133. Re1 Rd8 134. Re5 Kb3 135. Re7 Kc2 136. Rb7 Rd6 137. Rb8
   Rd7 138. Rc8+ Kd2 139. Rh8 Ke2 140. Ke4 Kf2 141. Rf8+ Ke2 142. Nc3+ Kd2 143.
   Nd5 Ke2 144. Nf6 Re7+ 145. Kd4 Ra7 146. Nd5 Ra4+ 147. Ke5 Ra5 148. Kd4 Ra4+
9. Kc3 Ra3+ 150. Kb2 Ra7 151. Nc3+ Kd3 152. Rf3+ Kd2 153. Ne4+ 1/2-1/2

[Event “USA-ch”]
[Site “San Diego”]
[Date “2004.12.02”]
[Round “7”]
[White “Friedel, Joshua E”]
[Black “Finegold, Benjamin”]
[Result “1/2-1/2”]
[ECO “B88”]
[WhiteElo “2436”]
[BlackElo “2531”]
[PlyCount “278”]
[EventDate “2004.11.24”]
[EventType “swiss”]
[EventRounds “9”]
[EventCountry “USA”]
[Source “ChessBase”]
[SourceDate “2005.04.11”]

1. e4 c5 2. Nf3 Nc6 3. Nc3 d6 4. d4 cxd4 5. Nxd4 Nf6 6. Bc4 e6 7. Be3 Be7 8.
   Bb3 a6 9. f4 O-O 10. O-O Nxd4 11. Bxd4 b5 12. e5 dxe5 13. fxe5 Nd7 14. Ne4 Bb7
2. Nd6 Bxd6 16. exd6 Qg5 17. Qe2 a5 18. a4 b4 19. Bc4 Rac8 20. Bb5 Bc6 21. c3
   e5 22. Be3 Qg6 23. Rad1 bxc3 24. bxc3 Qe4 25. Bd3 Qxa4 26. Bg5 e4 27. Bc2 Qb5
3. Qxb5 Bxb5 29. Rf5 Bd3 30. Bxd3 exd3 31. Be7 Rfe8 32. Rxa5 Ra8 33. Rd5 f6
4. R1xd3 Kf7 35. Rb5 Reb8 36. c4 Ra1+ 37. Kf2 Ra2+ 38. Ke3 Rxb5 39. cxb5 Rb2
5. Rc3 Rxb5 41. Rc2 Ke6 42. h3 h5 43. Kf2 h4 44. Re2+ Ne5 45. Bf8 Kf7 46. Rd2
   Rb7 47. Rc2 Rb8 48. Be7 Ke6 49. Re2 Rb3 50. Rd2 Rb7 51. Re2 Rb1 52. Rd2 Rb8 53.
   Re2 Rh8 54. g3 hxg3+ 55. Kxg3 Ra8 56. Kg2 Kd7 57. Rc2 Nc6 58. Re2 Re8 59. Kg3
   Nxe7 60. dxe7 Rxe7 61. Ra2 Ke8 62. h4 Re4 63. Ra5 Kf7 64. Rb5 Kg6 65. Ra5 Rb4
6. Rc5 Kh6 67. Rc6 Rb3+ 68. Kg4 Kg6 69. Rc5 Rb1 70. Ra5 Rg1+ 71. Kh3 Re1 72.
   Kg4 f5+ 73. Kg3 Re3+ 74. Kf4 Re4+ 75. Kg3 Kf6 76. Ra6+ Kf7 77. Ra5 Kg6 78. Rb5
   Ra4 79. Rc5 Ra3+ 80. Kg2 Re3 81. Ra5 Kf6 82. Ra6+ Re6 83. Ra5 g6 84. Kf3 Rb6
7. Kg3 Rb3+ 86. Kg2 Re3 87. Rb5 Re5 88. Rb7 Re7 89. Rb5 Rh7 90. Kg3 Kg7 91.
   Rb8 Kh6 92. Ra8 Re7 93. Rh8+ Kg7 94. Ra8 Re3+ 95. Kg2 Rb3 96. Rc8 Ra3 97. Rb8
   Re3 98. Ra8 Kf7 99. Ra6 Re8 100. Ra5 Rh8 101. Ra7+ Kf6 102. Kg3 Rd8 103. Ra5
   Re8 104. Rb5 Re3+ 105. Kg2 Re5 106. Rb7 Re7 107. Rb5 Rf7 108. Kg3 Kg7 109. Rb8
   Rd7 110. Ra8 Rd3+ 111. Kg2 Rd5 112. Kg3 Kf7 113. Ra6 Rd3+ 114. Kg2 Rd8 115. Ra5
   Kf6 116. Kg3 Re8 117. Rb5 Re3+ 118. Kg2 Re5 119. Rb8 Rd5 120. Rb6+ Kg7 121. Rb8
   Kf7 122. Kg3 Rd3+ 123. Kg2 Ke6 124. Rb6+ Rd6 125. Rb5 Kf6 126. Ra5 Rd8 127. Kg3
   Rf8 128. Ra7 Rf7 129. Ra8 Kg7 130. Kf4 Rb7 131. Kg5 Rf7 132. Ra6 f4 133. Rxg6+
   Kh7 134. Rh6+ Kg8 135. Rg6+ Kf8 136. Ra6 f3 137. Ra1 Rg7+ 138. Kf4 f2 139. Rf1
   Rf7+ 1/2-1/2

On Sat, Mar 24, 2012 at 2:57 PM, CHESSCENTR@aol.com wrote:

Hi Ben,

You were telling me in St. Louis last weekend about a 100+ move game (some very unusual ending, I think?) you played in a tournament a while ago which was using increment, and where everyone was waiting for the game to end before the last round could be paired, and it was past the time for the start of the last round?

I wonder if you wouldn't mind recalling  the details for me, because I wanted to see, when I have a little more time to think about this, how  the increment compared to a 40/2/ SD/1 + 5-second delay in these kind of "worst-case" scenarios? 

Do you remember the tournament, and what the increment time control was you used?

Any other details would be helpful as well.

Thanks very much!        

    Steve

See below for a post I made regarding 30 seconds increment.

Bill

30 seconds increment controls are unsound
by Chesstour » Sat Mar 26, 2011 6:41 pm

The following was written by Nick Faulks of Bermuda in 2008.

STANDARDISATION OF TIME CONTROLS

I was puzzled by the conclusion of this debate. Essentially, the only permitted increments will be zero or 30", even though each of these has drawbacks. Furthermore, it will not be permitted to introduce an increment only for the final section of the game - I had always been told that the top players favoured this but the clocks could not handle it, but now that the new DGT clocks have this feature it has been banned.

The problems with a zero increment are well known – it can lead to the nonsense of K+B chasing K+N around the board until someone’s flag falls. Note that this behaviour would become entirely rational if the trendy 3-1-0 scoring were to catch on. Still, if FIDE and the ACP are united in seeing this as the way forward for chess, I suppose it’s not my place to stand in the way of progress.

The biggest problem with the 30" increment is that, despite having been forced to play on this basis for many years, many players still don’t seem to like it. This may be partly because, at the end of a long game, the issue of bladder control can become crucial. For these reasons I have always argued that it should be routine for the sudden death phase of any game to be played with a 2" increment, but that will now be prohibited.

---

I agree with Nick, though even better than a 2 seconds increment is the 5 seconds delay that most US tournaments use. 5 seconds increment would also be OK, and maybe a little more. But 30 seconds is an unsound system for various reasons. There will often be games in which one or both players has no time left other than the increment- which may sound OK, but what if the game lasts another 50 or 100 moves from that point? That may be far too long to wait, so you are virtually forced to blitz off a lot of moves in order to accumulate time, but in many positions, being required to blitz off a long series of moves will result in almost certain disaster. I won’t even get into what happens when you survive the blitzing, make a mad dash to the nearest rest room, and find all the facilities occupied.

Another major weakness of 30 seconds increment: it’s a step backwards in the direction of unlimited time controls, when the Director had little idea when the round would end. At least in those days you could adjourn in the event of emergency, but that is no longer a fair option because computers would usually decide the game.

Because of the problems that a long game could cause, events with 30 seconds increment usually use a fast first time control. Most players dislike this and would rather use a slower control, a major reason why on US player surveys, 40/2, SD/1 with 5 seconds delay annihilates 40/90, SD/1 with 30 seconds increment. Here is how the two controls compare based on game length (using the abbreviations “5d” for 5 seconds delay and “30i” for 30 seconds increment.

Time for each player (hours, minutes, seconds):

30 moves: 5d 2:02:30, 30i 1:45:00. 5d allows each player 17:30 more time.
40 moves: 5d 2:03:20, 30i 1:50:00. 5d allows each player 13:20 more time.
50 moves: 5d 3:04:10, 30i 2:55:00. 5d allows each player 9:10 more time.
60 moves: 5d 3:05:00, 30i 3:00:00. 5d allows each player 5:00 more time.
70 moves: 5d 3:05:50, 30i 3:05:00. 5d allows each player 0:50 (50 seconds) more time.

I see all the above as advantages for 40/2, SD/1 with 5 seconds delay. For the most common game lengths, the players have more time, and most players prefer slow time controls, so long as they are not so slow as to be inconvenient or have games going very late at night.

80 moves: 5d 3:06:40, 30i 3:10:00. 30i allows each player 3:20 more time.
90 moves: 5d 3:07:30, 30i 3:15:00. 30i allows each player 7:30 more time.
100 moves: 5d 3:08:20, 30i 3:20:00. 30i allows each player 11:40 more time.
120 moves: 5d 3:10:00, 30i 3:30:00. 30i allows each player 20:00 more time.
140 moves: 5d 3:11:40, 30i 3:40:00. 30i allows each player 28:20 more time.
160 moves: 5d 3:13:20, 30i 3:50:00. 30i allows each player 36:40 more time.

These controls allow more time at 40/90, SD/1 with 30 seconds increment, but is that an advantage? No! For the majority of players, who like slow controls, more time for 80 move and up games hardly compensates for less time for games of under 70 moves, because the great majority of games are less than 70 moves. However, the extra time for unusually long games is a clear disadvantage to organizers, TDs, and players who want to start rounds on time. There will be occasional games of about 150 moves, and they can take an hour longer to play at 40/90, SD/1 +30 than 40/2, SD/1 +5. That means either rounds start late or the starting time is announced with unnecessary extra time between rounds.

A time control that speeds up most games but slows down the few that need to be speeded up is fundamentally unsound.
A time control that on occasion will result in a reckless dash for the nearest bathroom is also unsound.

Time controls with 30 seconds increment are a mistake, and are used only because FIDE has, for whatever reason, been pushing them, but they are unlikely to stand the test of time.

Bill Goichberg

[Sevan_Muradian]

The time control in Ben’s game vs Jan van de Mortel did not use increment. It used delay. It was 40/2 SD/1 with a 5-sec delay.

[hmb]

It’s been quite a few decades since Bill G. thought about these things from the perspective of the active chess master he once was.

Tournaments survived for many decades dealing with the occasional very long game. Adjournments made managing such things possible.

In the days when it was feared that computers would be the death of chess, and in the days in which it was argued that sudden death would lead to chess on network television, we did away with such things.

Neither the fear, nor the hope, materialized.

Perhaps the assumptions of the last decade or two have not been borne out. As adult membership numbers decline, maybe these things should be revisited.

[Smythe_Dakota]

Having “delay vs increment” in the title of this thread is a bit misleading.

Most of the comparisons here involve a 5-second delay vs a 30-second increment. It should be obvious that the difference between 5 and 30 is a far bigger factor, in calculating the theoretical length of a game, than the difference between delay and increment.

One occasionally sees estimates like “a 5-second delay is equivalent to about a 4-second increment” in its effect on the length of a game, but such comparisons seem based on little more than comfortable guessing.

Even if one compares apples to apples (e.g. a 5-second delay vs a 5-second increment), one still encounters various claims as to why one or the other is more likely to end a game quickly. One such argument runs along the following lines. Suppose, in a deadlocked position, black is down to his last few seconds of main time. With a 5-second delay, white is likely to play on, hoping that black’s state of perpetual time trouble will eventually do him in. But with a 5-second increment, black can accumulate time and soon establish a comfort zone, at which time white will see the light and agree to the draw.

But I am digressing a bit. The phrase “5 seconds vs 30 seconds” might be preferable to “delay vs increment” in the title of this thread.

I am wondering whether all the worry over long games is a serious problem in practice, or only in theory. By now I’ve played in several 30-second increment tournaments, none of which have run into serious problems because of long games. On the other hand, these tournaments have been smallish (fewer than 200 players). Perhaps, in the larger, stronger events run by Bill Goichberg, the problem could become very real.

Maybe we’ll never know. Some organizers will continue with 30-second increment and never run into a problem. Another will continue with 5-second delay in order to avoid even the possibility of trouble, and then we may never have a real test. The question will remain an unanswered mystery.

Bill Smythe