10

As some background, in the pre-engine era time-controls were often slower and chess players could adjourn their games, taking a break till the next day and resuming from the same position.

In all turn-based boardgames where this is or was an option, as far as I can see (chess, Go, Shogi, ...), adjournment was accompanied by 'move-sealing': one player writes down the move they will make, the other player does not see it, and the next day play begins with this move.

This was obviously a measure to avoid the inequity of discontinuing the game in a position where one player knows the exact gamestate while preparing their move. (For example, if I play my move openly rather than sealing it, and the game then adjourns, my opponent has an advantage because they can prepare a response to that exact move.) However, it is not obvious to me that this avoids inequity: there is a clear advantage for the sealing player, as they will know the starting gamestate the next day and the opponent will not.

What is the justification for this second advantage being less than the first?

2
  • 7
    Not sealing the move would be the equivalent of sealing it and then being allowed to have a takeback if your overnight analysis reveals there was a better option. While there's still an advantage, it's been greatly reduced
    – David
    Oct 24 at 18:16
  • 10
    "as they will know the starting gamestate the next day". Sure. And then the opponent makes their move, and they don't.
    – Polygnome
    Oct 25 at 17:41
32

However, it is not obvious to me that this avoids inequity: there is a clear advantage for the sealing player, as they will know the starting gamestate the next day and the opponent will not.

But the opponent has the move!

So the sealing player knows their own move, but doesn't know what the opponent will reply to it. And the opponent knows the position before the sealed move, but doesn't know what move was sealed.

It's probably not completely equal, but I don't see a way to get closer than this.

4
  • 17
    For a completely symmetric description: both players get to know the game state 1 ply before their next move. Oct 25 at 6:53
  • 2
    a potentially better option is if both players get unlimited amount of conditional premoves in their sealed moves, but in reality no one would want to potentially commit the game into a weakly analyzed position of a single line from a very deep tree analysis... Oct 25 at 16:40
  • @FedericoPoloni how can that be implemented? Oct 25 at 23:10
  • 2
    @theonlygusti That sentence was a description of what happens with the normal adjourning procedure; not a proposal for a new alternative procedure. To expand: if the game gets adjourned after ply N, with ply N+1 in the sealed envelope, then the non-sealing player knows the board state after ply N, and will move again after ply N+1; the sealing player knows the board state after ply N+1, and will move again after ply N+2. Oct 26 at 6:47
14

Sealing can provide an advantage to the sealing player - but it's not that clear. The point is that after you seal a move, it is opponent's turn, so neither player knows with certainty what the position will be when it's their turn to move. If both players have many many possible, reasonable moves then both players have to prepare for all of them. However, if the sealed move forces a specific response then the sealing player has the advantage.

You can see some discussion on this from the two videos by Nakamura and Carlsen on a game from the Queen's Gambit. At 1:30 in Carlsen's video, he specifically says Borgov makes a mistake by adjourning the game when he only has one decent move, which obviously makes it easy for Harmon to prepare. Nakamura analyzes the position as well around 3:30 in his video. To quote, "Borgov should have played 36...Qg6, and then after 37. Ne6 he should adjourn." So because Borgov chose to adjourn at a bad time, his opponent gains the advantage.

There's one more point: as you can see from the Wikipedia article on adjournment, both players can choose to adjourn the game at any moment after the allotted time has passed. However, choosing to adjourn comes with the drawback of forfeiting all your remaining time. Your opponent keeps their time, which can be an advantage if complications arise after resumption.

3
  • But equally if the move just before the sealed move forces a specific response, then the non-sealing player has the advantage. Oct 25 at 12:39
  • 1
    In practice, how was the decision made about when the adjournment would occur, i.e. which move would be the sealed move? Oct 25 at 12:41
  • @JamesMartin en.wikipedia.org/wiki/Adjournment_(games)#Chess "Once the time control has passed, either player has the option of adjourning, and may do so on their move." There's an added drawback of choosing to adjourn as well in point 2, which I will edit into this answer.
    – Allure
    Oct 25 at 12:44
7

I'm not sure about the exact quote or source, but it's something like

Chess is won by the player who makes the second-to-last mistake.

The sealed move, the last one of the day, perhaps made a few minutes after an intense time-trouble battle, may be the last mistake. The opponent has a lot of time to 'guess' the sealed move, and could concentrate their analysis not only on the best move in the position, but also on potential mistakes which could quickly be refuted.

This is likely not enough of an advantage to compensate for knowing the starting position next day, but at least it's better than the alternative (with both players knowing it).

5

However, it is not obvious to me that this avoids inequity: there is a clear advantage for the sealing player, as they will know the starting gamestate the next day and the opponent will not.

Not necessarily. For quite a few board states, there's only one non-blunder move, and sometimes only one legal move. And there's pretty much never more than a dozen feasible moves (and I could probably go a lot smaller than "dozen" and still have that statement be correct, I'm just being cautious).

Suppose the players' time averages to 10 min per move, and the board state at adjournment is such that there are 10 moves that aren't obviously stupid (for both of these numbers, I picked numbers that I think are on the upper end). If the non-adjourning player spends 20 min on each of those moves, that's 200 min total. And then, no matter what the sealed move ends up being, the non-adjourning player has spent double the normal time on analysis, with no time being charged to their clock. And in the days of adjournment, players were often allowed to have a second, or even a team, that they could consult with. So if there's two plausible moves, the player can assume that one of them will be made, and spend all night deciding on a response, and the second could assume that the other would be moved, and spend all night deciding on a response to that.

Now, of course, the adjourning player can do something similar: they can take all the possible responses, and do an analysis for each of them. So it comes down to which player's move is less predictable; if a player has more viable moves, then the other player's analysis has to be split between more moves, and so the amount of time per move is smaller. On average, each player is equally likely to be advantaged by the adjournment. Adjourning at the right time can be a large advantage, but the advantage comes from picking a point where the complexity tree is most in your favor, not in knowing what the next half-move is. Basically, you want to stop at a point where you're the next player to add uncertainty to the game.

What is the justification for this second advantage being less than the first?

In any board state, there's a finite number of legal moves. There no rule saying that the non-sealing player can't write down every legal move, and their response to it, before leaving the playing area. Now they know exactly what the board state will be in the morning! Committing to a move is a steep price to pay for knowing what the board state is.

With sealed moves, one player has the advantage of knowing the board state, and the other has the advantage of their next move being free. Without sealed moves, one player would have both advantages. It's not a case, as you put it, of choosing between giving one sort of advantage, or giving another sort of advantage. It's a case of choosing between giving two advantages, or giving only one.

Suppose you're White, it's Black's turn, and you have a choice between:

(A) Black makes their move secretly, it's sealed, and then there is an adjournment.

(B) Black makes their move publicly, you make a move publicly, and then there is an adjournment.

Which would you prefer? In both cases, Black knows the board state, but Black's advantage in (A) is nowhere near Black's advantage in (B). I, and I expect the vast majority of chess players, would much prefer (A).

1
  • Back in the day, a typical time control for championship games was two and a half hours for the first 40 moves, and 16 moves per hour thereafter. So 3 minutes 45 seconds per move. And I would say the number of reasonable moves in a game that is already 40 moves old would be more like three or four than ten. (But I like your answer.)
    – TonyK
    Oct 25 at 10:03
4

It could occasionally be advantageous, and even strongly so, either way. I can recall being in a pawn-up Rook endgame well ahead on time, but unsure of the correct plan. I waited without moving until I could seal a noncommittal move. Alternatively a situation could have arise where I might have been obliged to seal a committal move. I do not think there was much of a systematic advantage either way

Incidentally, the sealed move itself was almost always the first move after the time control, and so was never made in time pressure unless for some exceptional reason. Arbiters accepted to remain in the playing hall as long as they needed to while the player made a decision

2
  • 1
    Ah, interesting. So in practice, the sealed move was almost always made by the white player? Oct 25 at 12:43
  • 1
    Ah no, I was not quite accurate. Just after the time control had passed, the TD would go round all the unfinished games placing an envelope before every player on move. If you wanted to seal you could just wait until they came round but you would be sacrificng your remaining time. You could not refuse to seal.
    – Philip Roe
    Oct 25 at 20:14
1

It's not the idea of advantage to the sealing player 'cause he knows the move that will resume play. That's not an advantage to him anyway, and there's a far, far more important issue that arises if he doesn't seal a move.

If he does not seal a move, and it is, oh, 17 hours until play resumes, he gains 17 hours... 1020 minutes of time to consider his next move and be ready to play it. So, if each player ended up using two hours of on the board time, the sealing player had 19 hours to consider his moves while the other player had 2 hours to consider his moves.

To kill off the niggling quibble here and to point out why it was never much of a "knowing the next move advantage" consider the ways you can look at it. I am referring only to time used to consider real, my move is next, what shall I play, oh, and I know everything there is to know about the opponent's moves since they've all been made, kinds of moves. Yeah, sure, I can think "on my opponent's time" but he could play lots of moves and if he's thinking about things while I think too, then clearly at least two of them intrigue him. What are those two? Maybe that's easy to predict, maybe it's not. I have quandaries and some amount, maybe a lot of it, of the time I think "on my opponent's time" is pointless as the opponent makes some other move altogether, or since he can only make one move, the time I spent considering his other options was rendered "wasted."

If I seal a move, then BOTH of us analyze from that point with one opponent's move "in between" the position now and our next move. In his case, my unknown sealed move is in between now and his next move. He must waste some time pondering paths forward that won't happen due to my sealing a move that makes other paths happen. But... I may know the next move that will happen, but the move he must make, the one I just mentioned, is in between the move I sealed and whatever next move he makes. Both of us have plenty of time to think about everything but both of us have one opponent's move in between now and our next move and that opponent's move could be many things.

Sealing is and was simply a method to maintain that fact and apply fairness to all.

Remember, if I don't seal a move, there is no opponent's move in between now and my next move. EVERYTHING I spend time on is then worthwhile, nothing wasted, for me. A huge advantage over my opponent.

I'd also point out that after whatever time control after which adjournment was allowed, EITHER player could choose to seal his move. Not just, say, White. That had a minimum of two important effects.

First, why would either player seal a move in a situation in which his next move was dead obvious (due, for example, to hanging a queen if he doesn't seal a particular move... might as well be clear about how advantage could be had there)? Do so, and you might are adjourning the game in a situation in which the opponent is essentially thinking "OK, if he doesn't make "the move" I take the queen and win. That was four seconds. So, let's spend the other 16:59:56 doing real analysis assuming he makes "the move." Yay, this is soooo sweet!" So almost by definition, a sealed move will be one in which the position has some amount of complexity.

Second, a player could actually choose to aim for "very complicated" so as to minimize his opponent's benefit from the extra thinking time. The opponent would have to consider all the reasonable moves while he knows the move chosen and might enter his analysis looking at a much simpler future, thereby gaining an advantage. I'd say two mitigating things though about that advantage. One is that most complex situations seem to arise over a few moves, not instantly after a single move, and as a practical issue, why would you want to give him the, say, 17 hours, to think about the "sitch" instead of making him burn clock time? So moderately complex situations, not clock burners per se, but not 2) ... Kc6 either, would seem the target and most folks playing at a level in which sealing occurs ought to be able to handle things without the opponent gaining large advantage. The other is that either player can adjourn, so your opponent sees you're aiming at something and adjourns a move before you intended to. Now YOU have any disadvantage that arises.

So basically, it is/was just meant to make adjournment not the source of possibly incredible advantage to one player. Everything else is details and small plus/minus effects. EACH player is in the position of analyzing all he cares to, but always while not knowing his opponent's next move. Not just the non-sealer. Just keeping the playing field as level as the situation can permit.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.