5 second delay was to prevent the loss of trivially won or drawn games for the players while retaining the scheduling benefits of Sudden Death for the organizer/director.
Nah, let’s just make it as complex as possible to increase confusion and squabbles.
Set Desire-To-Direct-Chess-Tournaments = 0
Many players tend to use whatever time they are allowed and get into severe time pressure. With no delay (or increment) these players not only may lose easily won or drawn games, but in general will find the later stages of games unpleasant. 5 second delay addresses this problem, but not fully, as some players will still commit gross blunders or feel under great pressure. 10 second delay does a better job of what 5 second delay does, and is not more complex (though the clock settings on various clocks may take some getting used to).
If 10 seconds is better than 5, why not a still longer delay? Because with two rounds per day, there is the issue of players having time to eat or rest, and of delaying the next round. If you want rounds at 11 and 5 and use, say, 40/2, SD/30, then with 5 second delay a 60 move game may last until 4:10, and the occasional 120 move game until 4:20. Not too bad.
With 10 second delay 60 moves can last to 4:20, 120 moves to 4:40. The latter is borderline as in a large tournament, it can take about 20 minutes for the pairings to be posted and for the players to find their pairings, go to the tournament room, find their boards, and set up. And now and then a game will go more than 120 moves. But I think it’s worth putting up with these problems in order to reduce the severity of time pressure.
If the delay is 15 seconds, 120 moves can take you all the way until the announced starting time of 5:00. Adjournments should be avoided as they tend to distort the pairings, and and in this age of 2800 strength computers, they are even less desirable than they used to be. So this is where I draw the line- 10 seconds yes, 15 seconds no.
Of course, with 30 second increment (or delay), these problems are much greater and you pretty much need to speed up the time control to compensate for the increment. A 120 move game played at 40/2, SD/30, inc 30 could end at 6:00, a whole hour later than the scheduled starting time. So the organizer wanting 6 hours between rounds and 30 seconds increment will probably use 40/90, SD/30, inc 30, which will likely start the rounds on time, but now if we include the delay or increment the players are playing the first 40 moves in 110 minutes, while with 40/2, SD/30, d10 they are playing the first 40 moves in 126.7 minutes. That’s about 15% more time with the 10 second delay control for the first 40 moves, which are important in every game, not just those that are unusually long.
For the first 60 moves the 10 seconds delay control allows 10 minutes more, about 7% more time, still a meaningful benefit. Only when we get to move 90 do the two controls allow an equal amount of time- but the great majority of games are decided before move 60 and way before move 90, and for these games the 10 seconds delay control allows more time.
While very few games go past move 90, one such game can delay the next round for an entire large group. For a 120 move game the 30 second increment control allows 10 minutes more per player, but that’s not an asset. 30 second increment controls speed up the great majority of games (in which most players would rather have more time), but slow down the few games which can delay the next round. Not a good combination for events with two games per day.
Bill Goichberg
Or, we could start the morning round at 10:00.
Alex Relyea
CCA has now held 11 tournaments with a time control in all sections of 40/110, SD/30, d10, and I am very pleased with the results:
2013 Southern Open: largest turnout since 2002.
2013 Cleveland Open: largest turnout since we started running the event in 2008.
2013 Indianapolis Open: poor turnout, but hotels within many miles were sold out two weeks before the event due to a huge convention.
2013 Midwest Class: largest turnout since 2009, remarkable because we had to change the dates only 5 weeks before the event, which I thought would kill entries.
2013 Los Angeles Open: same turnout as 2012, before that we didn't run for many years.
2013 Kings Island Open: largest turnout since 2007.
2013 National Chess Congress: largest turnout since 2008.
2013 Empire City Open: largest turnout since the 1970s.
2014 Boston Chess Congress: same turnout as 2012, only other year, pretty good because Foxwoods two weeks later was expected to hurt entries.
2014 Liberty Bell Open: entries dropped from 440 to 408, decline was expected because fees were raised and lowest section removed.
2014 Golden State Open: largest turnout since event started in 2010.
While it might be that some other factor is causing all these good turnouts, it certainly seems like organizers have little to fear if they try this time control. It comfortably allows 6 hours between rounds, the first control is quite slow and surveys have found that players want a slow first control but don’t care as much about the second control, and the 10 second delay reduces severe time pressure.
Bill Goichberg
I was talking about an organizer who wants 6 hours between rounds. If you consider 7 hours between rounds OK, a similar comparison could be made to 40/2, SD/1, d10, showing that 30 second increment speeds up the great majority of games (the ones that don’t need to be speeded up), but slows down the occasional very long games (which it would be nice to speed up).
I used to run most tournaments with 7 hours between rounds, and 40/2, SD/1, d5 was preferred on player surveys for many years, but there has been a recent trend towards somewhat faster time controls. I now consider 6 hours between rounds standard, but for norm tournaments 40/110 or 40/100 is not allowed by FIDE, so I have to do 40/2, SD/30, d10, which gives back 20 minutes of the hour saved.
Surveys have also shown that most players prefer to have about one hour between the anticipated end of a round and the start of the next round on that day.
Bill Goichberg
When using a long delay such as 10s, 20s, or 30s, and a player uses all of his/her main time and is playing only on the delay, it seems to me that there is not much difference in this and the byo-yomi over time used by go players.
The second time control with Game 30, d10 is too short and very stressful. There is no chance to take even a little break after the first time control to get something to eat and go to the bathroom or even let your heart rate drop a little. In non-trivial endgames, every minute and second count. This speeding up of the game to please organizer/TDs and those with ADHD who are antsy to play the next round is not a positive for the game of chess. In Grand Prix games, the second time control should reflect the possibility of playing at the same rate per move as the first time control.
Now I’m curious. Would it be legal to run a rated USCF tournament using byo-yomi time, such as 60 minutes plus 10 byo-yomi periods of 1 minute each? How about hour glass timing, where each player starts with N seconds, and on your move your clock goes down and your opponent’s goes up, and you flag if your clock reaches zero?
If these non-standard time controls are allowed, how would you determine what rating category they fall under?
Another thing I’ve wondered. Suppose a tournament is 40/120, SD/60, no delay or increment, and suppose a player bought a DGT Easy Game Timer. That clock only supports three timer modes: move timer (e.g., each player has N seconds per move), game timer (each player has N minutes for the game), and count up (each player starts at 0:00 and counts up).
In game timer mode, when you run out of time, a flag and a “+” appear in the display, and your clock switches to count up mode.
One obvious way to use the Easy Game Timer would be to start in game timer mode with 120 minutes. When black completes move 40, the clock could then be stopped, 60 minutes added to both sides, and white’s clock started. I’m pretty sure that this would be a legal way to use this clock, by analog with analog clocks when an adjustment is needed for flag timing.
Another possibility would be to use count up mode. Each player starts at 0:00, and they have to make 40 moves by time their clock reaches 2:00, and they have to finish he game by 3:00. This has the advantage that there is no need to stop the clock and adjust the times between time periods. However, there is no flag.
Would using a count up clock this way be legal? Or would the lack of a flag preclude it? On an analog clock, the flag was necessary to provide an objective way to decide when the minute hand passed the hour mark. On a digital count up clock, there is no question when the clock reaches and passes 2:00 and 3:00, so one can make a good case that a flag is not necessary.
For many clocks, one problem with using count-up mode is that the seconds are only displayed if the clock is under 10 or 20 minutes. If both players are sitting at 1:59 (or 2:59) you may not know which player is really in deeper time trouble (one could be at 2:59:02 while the other is at 2:59:57).
I do not see how time controls that do not comply with rule 5 of the Official Rules of Chess would be ratable. A byo-yomi time control does not comply with rule 5A.
I have noticed that a local state championship is advertising a ten second delay. Does that mean that Mr. Goichberg’s misguided idea is catching on, or that there is actually something to it. I have found his arguments in this thread unpersuasive to say the least.
Alex Relyea
The fact that I direct a lot of CCA tournaments and am also the chief organizer of the 83rd Massachusetts Open might have something to do with it. 
I agree with Bill Goichberg’s reasoning: a lot of players prefer a 5-hour game instead of a 6-hour game, but 30/90, SD/60 d5 isn’t FIDE-ratable (first time period must be 40 moves), so instead we use 40/120, SD/30, but to reduce the time crunch in the sudden death period we use d10 instead of d5.
I know you like increment but my impression is that most New England players prefer delay. With increment most games finish in a reasonable amount of time but a few games can run late, so there has to be more time between rounds.
I think the main difference (for the purpose of getting the round finished on time) is not so much the difference between delay and increment, but rather the difference between 5 seconds and 30 seconds.
Bill Smythe
Did the time control for the Mass. Game/60 include a ten second delay? It seems to me that all of Mr. Goichberg’s arguments apply equally.
Alex Relyea
No, I used 5 second delay for the Mass. G/60 Championship. I see that as being a faster tournament so there is less concern about using d10 to give players more time in the endgame. If I get feedback from players asking for d10 I’ll consider making it G/60 d10 next year. No one said anything about the delay time this year.
Did any players in the Mass. Open last year request a 10 second delay? I really think that this is a solution in search of a problem.
Alex Relyea
By the way, many tournaments in Oklahoma are G/50; +15 for the express purpose of getting high quality games (to the extent that any of us class players is capable of playing high quality games) and having them be dual rated. Even in their “maximum length” games, that is G/90; +30, players are encouraged to hand off their scoresheets to a spectator when they get below five minutes so that the organizer has a complete score.
Alex Relyea
Last year the Mass. Open had a six hour time control: 40/120, SD/60 d5. One player, who didn’t enter the tournament, said this was too slow and suggested a four hour time control. Other players also said that six hour games were too long.
Since CCA has switched to five hour games I decided to switch to five hour games for the Mass. Open, and it was simplest to copy the time control CCA is using for its biggest tournaments: 40/120, SD/30 d10.
We are likely going to experiment with d/10 at the Quad tournament I run. Currently, the time control is G/45;d5 or G/45;d0 for clocks without delay. If we use G/45;d/10, should clocks without delay still use G/45;d0?
As far as I know that’s the rule, unless otherwise announced in all advance publicity, etc. For a club or local quad event, I would imagine announcing a variation at the site would suffice, in practice. I would not grant a time adjustment for analog clocks until the delay or increment reached at least 15 seconds, personally.