As to split decision rulings, esp, in cases such as described by US
Chess rule 20ED2. Could not the adjudication of double N’s or an N
and R be a better solution than the one advocated in the rulebook, as
an adjudicated draw for each?
(20ED2 - Unsolicited advice where such advice is the winning or game
saving move).
Thoughts please -
Rob Jones
I don’t know if there are any places in the rulebook where inconsistent results (such as both players getting a ratable win or one player getting a win and the other a draw) are specifically discussed much less recommended.
It comes down to the director’s judgement of the specifics of the case. As rule 20E2 states, “There is sometimes no good solution to the problem.”
Any ruling that may depend upon the chess skills of the TD is going to be problematical.
I agree – I think.
I don’t think these options are even mentioned, anywhere in the rulebook. And I don’t blame the rules committee for hesitating to touch this idea with a ten-foot pole. It could become entirely too commonplace, for a variety of reasons both good and evil.
By “20ED2” you apparently mean 20E2, section d, second paragraph. That’s the situation most likely to tempt a TD to make one of the rulings you suggest. And it may well be the fairest solution every so often.
So, go ahead (he said, with trepidation, hoping that such a solution will be used only rarely, but fearing an avalanche). 
Bill Smythe
And a good question then is this Mike- Why is this NOT discussed anywhere in the rule book? I do wonder how many tds even know that split decisions are possible ? But, however much desired, it is a
tool available for TDs use, and as such, should it not be discussed?
Rob Jones
It is an interesting question, perhaps some Delegate will raise it some time.
20E is one of the few rules requiring the TD to make a judgement call based on the TD’s chess ability. The TD has to decide the strength of the suggested move, the strength of the other options, and whether or not the player would have seen the move.
I’ve had spectators suggest an “only move” that was strong and/or winning with players that may not have been strong enough to see it. Looking at the scoresheet may show that the “only move” was part of a temporary sacrifice and was most likely seen before the sacrificial move was made - in which case ignoring a neutral spectator and allowing the move is fine.
I’ve heard of spectators suggest an “only move” that may well not have been seen. If said spectator has a vested interest in a draw being declared (brother of the only 5-0 with the only two remaining 4-0 players in that game) then declaring a non-rated double-win for scoring purposes may be fine while submitting the game in an extra game section as a draw for rating purposes.
I am much, much more willing to have a non-rated split result with a normal (non-split) USChess rated result (in an extra game section) - versus having a rated split result.
PS Some TDs really like rated split results. Others consider them absolute abominations. I consider them something to try really, really, really hard to avoid (I know I’ve done at least one over the last 35 years but I think the total is less than five).
I’ve only ever seen split results given at the National Scholastics and there were only a few of them. In each of these cases I’m not sure I would have done it, but neither was it a wrong decision. I can’t recall all the specifics, but one example is a situation that occurred in one of the earlier rounds (maybe 2 or 3). I was not the one making this ruling. I just dealt it with as the pairing chief.
Two players sat at the wrong board (next to each other) and swapped places. They played their respective games to completion in the wrong spot. This is normally not a big deal and you would simply just adjust the pairings to match what actually happened and move on. But, this is a National event and the stakes are higher. The effect was this (the ratings are approximate, I don’t recall what the actual ratings were):
Player A (1300) sitting at the correct board ends up playing Player C (1350), who is in the wrong spot.
Player B (650) sitting at the correct board plays Player D (600), who is in the wrong spot.
Everyone played the correct colors. Players C and D win their games. Player A finds out he played the wrong player and complains that he should have played a 650 and most likely would have won. Probably true.
What decision would you make? Obviously you can’t make all the players replay their games against the correct opponent. Do you just tell him to suck it up and move on? Player A did nothing wrong except not verify that his opponent was the correct one. He was in the right spot. I suppose there is a reasonable presumption that the person opposite you is supposed to be the right one.
The final decision was to split the result keeping the win for Player C (since he did win the game after all) and give a draw for Player A. Interestingly though, Player B never complained, so he got nothing.
Here’s a table showing how many of the N, R and S codes have been used each year:
[code]year count n r s
2004 4 4 0 0
2005 52 36 16 0
2006 18 10 6 2
2007 40 26 10 4
2008 42 24 16 2
2009 48 38 10 0
2010 56 35 21 0
2011 62 39 22 1
2012 70 43 27 0
2013 78 61 17 0
2014 136 103 31 2
2015 110 74 36 0
2016 146 109 32 5
2017 254 209 45 0
2018 186 147 37 2
2019 12 12 0 0
[/code]
That seems like an awful lot. What is a gross estimate for the total number of games rated in a year?
Alex Relyea
Looking at the data shows the following rated split results (there is no record of the non-rated split results for standings that were rated as normal results). I made an assumption there were no double-losses (there could have been as many as 
. For 2004-2018 each double-loss would reduce the DL count by two (dropping to a as few as 2), increase the WD count by two (increasing to as much as 324) and reduce the double-win count by one (dropping to as few as 317).
It looks like the number of double-wins has gone up a lot, in recent years regularly surpassing the number of win-draw results. The total number is still only 651 for the period from 2004-2018, though there were 127 in 2017. An earlier post of Nolan’s indicated we were running around 300,000 games per year just for adults (probably have to divide that by two to account for both players) and I would expect the number of kids’ games to be even more. If so then we are looking at maybe as much as 0.05% of the games having split results. Personally I agree that would be a high percentage.
Year / WW / WD / DL
2004-2007 / 25 / 26 / 6
2008-2010 / 26 / 45 / 2
2004-2010 / 51 / 71 / 8
2011-2014 / 76 / 94 / 3
2015-2018 / 198 / 143 / 7
2011-2018 / 274 / 237 / 10
total / 325 / 308 / 18
It looks like we rated about 820,000 games in 2018. Keep in mind that the number of results is 2X the number of games, so 254 results is only 127 games. (In 53 events for that year, FWIW.)
Usage of the N/R/S codes seems to be increasing, I’m not sure if that’s a good thing.
I’ve been directing for 11 years and I have only made 1 split result ruling so far. Hopefully, it was my last one.
Also, I did some math… 127 splitgames / 820,000 games x 100% to get… 0.0154878048% of all games played…
So, not even 2 thousandth of a percent?! This does not need to be on the delegate’s floor.
Best,
~Ybriang
Brian Yang
Unless, of course, the percentage of those games rises precipitously, now that people know it exists from reading this thread.
Bill Smythe
This isn’t the first time it has come up. Pretty much every time it has come up it has been described as something that should only be done in extreme cases.
I think if Brian’s usage increased ten-fold he would give the same advice.
This looks like it may have been a split result, but NOT a rated split result. In a case like this it may well be that the actual result went into an extra games section (win/loss) while temporarily a rated split result was retained in the original section until after the final tie-breaks were done, at which point the rated split result was changed to a non-rated split result (maybe forfeit-win/half-point-bye). Non-rated split results are much more palatable than rated split results.
Also, player B lost against a player lower-rated than originally anticipated so there was no justification to compensate for having that lower-rated opponent.
Finally, if it is a two-board difference instead of a one-board difference then it may well be a reasonable transposition of very similarly rated players (e.g. a 1300 A and a 1280 B swapped their 650 D and 620 C opponents) that could be handled by simply keeping the over-the-board results as the official results.
If such a transposition is found while the games are still going then the erroneous games could be annulled and restarted with the correct opponents if it wouldn’t otherwise be a reasonable transposition (there is strongly suggested limit as to how far a game should go before a TD changes a pairing, but there is no such limit for when a player changes a pairing).
All good points. I don’t remember offhand if we moved the game to an extra game section. If we didn’t, then we should have. I like that idea. It’s possible B was actually lower rated than D, but I don’t recall. I just remember that the ratings were similar between the opponents on both boards and A was definitely lower than C.
If such a transposition is found while the games are still going then the erroneous games could be annulled and restarted with the correct opponents if it wouldn’t otherwise be a reasonable transposition (there is strongly suggested limit as to how far a game should go before a TD changes a pairing, but there is no such limit for when a player changes a pairing).quote -
Truly the acid solution. From a rated games standpoint, it is difficult to see, indeed,
the justification for this solution.
Rob Jones
Five minutes into round one at one tournament there was a situation similar to the 650 sitting across from the 600 and the 1320 that was supposed to play the 600 having only one person left to play (the 1350 that the 650 was supposed to play against). The two 6xx players were maybe 15 or so moves into the game when the 1320 arrived slightly late. One option is to accept the way the 650 unilaterally changed the pairings for all four players. The option chosen was to deem the 600 vs 650 game as an unauthorized skittles game being played in the tournament hall and to have all four players play their correct opponent. At a big-money adult tournaments I split up a round four game that was almost an hour in (somebody sat in a wrong, and lower, section and the correct opponent made a mistake on the starting time of the morning round).
If you always simply accept how a player changes the pairings then you run the risk of a 3.5-0.5 player in the fifth and final round "accidentally" sitting at the wrong board against a weak opponent and getting an easy win. Prior to the one of the rulebook changes (back when I was only a Local-level TD assisting the chief TD) there was a scholastic player who would "accidentally" mis-report a first round win as a loss, get paired throughout the tournament with that loss, and then get it corrected after his final round game was played and before awards were given (getting the benefit of easier pairings throughout the tournament). Finally the rulebook explicitly allowed a TD the option of correcting an early round result only for the ratings report while keeping the reported result for awards. After that change a player would no longer have an incentive to "accidentally" mis-report the first round result. I see no reason to unnecessarily give players an opportunity to game the system.
That said, if the round one game had been a case were a 1310 accidentally sat in the 1320's spot and the players are a dozen or so moves into the game then the TD may (or may not) simply record the slight transposition and not interfere with the game that has already been started (readily doable in any round if the two players have the same score, are paired with the same color, and were paired against similarly rated opponents). In such cases if the 1310/1320 transposition was not discovered until after one of the games is over and the players are gone then a simple acceptance with no adjustment is often reasonable. On the other hand, in the case cited with two 13xx's playing in round one and two 6xx's also playing in round one then an (eventually) non-rated split result (win/draw) is plausible with the actual result reported in an extra game section (it would plausibly be consistent to also handle the 6xx game as a non-rated loss/draw with the actual result in the extra game section but that seems an unnecessary risk of escalating tensions). That (a non-rated split result in the section / extra games actual result) allows a TD to compensate the player adversely affected by another's error and still get the actual result rating without messing up the rating system with a (split) result that cannot be obtained over the board.
I stand corrected, and thank you for your thoughtful and reasonable explanation.
Thankyou,
Rob Jones