From “Arbiter’s Notebook” on Chess Cafe
FEN “1q6/8/8/8/7R/k7/1q4QQ/K7”
In this position with White to move, White’s flag falls. It’s a draw because there is no legal sequence of moves where Black can win (the only legal moves force a mate in 2 for White)!
I guess it’s consistent, but…
Looks like 14D4 should be copied (with a minor change) to 14E4. Also, it looks like 14D4 should probably have been written with “either player” rather than “the player”.
Current
14D. Insufficient material to continue.
…
14D4. No legal moves leading to checkmate by opponent.
There are no legal moves that could lead to the player being checkmated by the opponent.
14E. Insufficient material to win on time.
Add
14E4. No legal moves leading to checkmate of opponent.
There are no legal moves that could lead to the player checkmating the opponent.
Here’s a contest along these lines.
Think up a position where a draw is impossible (other than by agreement), but a win for either white or black is possible.
Ideally, think up a position, white on move, where white has only two legal moves, one of which immediately checkmates black, and the other of which leaves black with only one legal move which immediately checkmates white.
Then think of various necessary improvements in 14D and 14E, which would be compatible in some reasonable way with the existence of such positions.
Bill Smythe
I did it with black on move.
White
Kg1, Qf2, Pg4, Bh5, Pg6
Black
Be3, Pg2, Rh2, Kg3, Rh4, Pg5, Pg7
Last two moves were … Bxe3+, Qf2+
Black is in check and the two legal moves are Kh3 and Bxf2#
After Kh3 white only has Qxe3#
Both players can lose on time since each has the capability to lose. After …Kh3 only black can lose on time since there is no legal set of moves leading to a win by black. A draw by agreement is always possible.
P.S. After …Kh3 white can also resign.
P.P.S A possible win by black after …Kh3 is Qe2 (illegal) Bc1, Qf1 Be3+, Qf2 Rh1 # (the mating move is legal). Black cannot cite that for claiming a win on time since it is not a legal continuation. TDs can stop the illegal Qe2 if they are not constrained to only be witnesses to illegal moves (witness-only is a common variation used when the number of players greatly outnumbers the number of TDs).
Nice. Congratulations!
Hmm, somehow I thought it might have been possible with fewer pieces (perhaps about 3 on each side). But oh well.
And now for the sticky rule question created by this example.
Define a position as purely winnable by white in case there exists a sequence of legal moves leading to white checkmating black.
Ditto for purely winnable by black, substituting “black” for “white” and vice versa in the above.
The word “purely” is designed to exclude various inconvenient and quasi-legal methods of winning, such as time expiration, resignation, agreement, penalty imposed by TD, etc. In other words, we’re talking pure chess.
Now, what is a reasonable definition of purely drawable? I think the following will do it:
A position is purely drawable in case either there exists a sequence of legal moves leading to stalemate, or there exists a sequence of legal moves leading to a repeated position.
This seems to cover all the “pure” ways of drawing, including triple occurrence, 50-move rule, etc. Of course it excludes the quasi-legal stuff, such as agreement, time expiration, penalty, etc.
FIDE is very good about making sure that if, for example, the position is not purely winnable by white, then white cannot win by other means either, such as time expiration, opponent’s cell phone ringing, or TD penalty.
In view of the diagrammed position, shouldn’t FIDE also make sure that, if the position is not purely drawable, then a draw cannot occur by other means either? For example, the players should not be permitted to agree to a draw. They must play on until there is a decisive result (including checkmate, time expiration, resignation, penalty, blah blah blah).
FIDE is also very good about making sure that, if a position is neither purely winnable by white nor purely winnable by black, then the game is an immediate draw (dead position, comparable to USCF 14D4).
I think FIDE should handle the other two dual impossibilities in the exact same way. If, for example, a position is neither purely winnable by white nor purely drawable, then the game should be an immediate win for black, regardless of any subsequent resignations, time expirations, cell phones, etc.
Bill Smythe
So would this be an automatic black win? (black’s last move was a capture on g7)
and this an automatic white win?
and this an automatic draw?
Nothing more to add but
Yes, yes, and yes. Why not?
The third one is already a draw under present FIDE rules (dead position) and USCF rules (14D4).
Bill Smythe
Although my chess playing skills are very weak, I’m not sure I understand how the third position (with two white queens and the black king on the 7th rank) could have arisen by any sequence of legal moves. As the black king is in check, the last move must have been White’s. At least one of the white queens must already have been on the seventh rank already, so the black king would have been in check before White’s last move.
The queen captured a piece which was blocking the check.
Thanks. 
Personally I’d prefer to see “either player” instead of “the player”
Cf. ADM 15-44. 
I actually managed to sneak this ADM in a little past the deadline, motivated by your mention of this earlier in the thread. I figured my opus magnum ADM 15-41 would keep the people putting the Delegates’ Call together busy enough that I could slip another one in … (Poor Cheryle Bruce must be out for my hide. 
)
Yes, 14D needs some work.
I think FIDE could improve its version, too. In addition to a “dead position” (a position in which there is no sequence of legal moves leading to checkmate by either player) there should be a notion of a “half dead” position. For example, a position is “dead for white” in case there is no sequence of legal moves leading to white checkmating black. Adding this definition might enable several other FIDE rules to have their wording simplified.
As for the USCF version, I think 14D1, 14D2, and 14D3 should be removed. They are redundant as they are all special cases of 14D4. The positions in 14D1, 14D2, 14D3 could better be used as examples following the main rule, or they could be included as TD Tips.
Bill Smythe
I once set up a position where the only legal move was
pawn takes pawn en passant, checkmate.
I see a lot of positions written out in notation but do they
distinguish whether a pawn can capture en passant or not?
Just looking at the chessboard you wouldn’t see what
the previous move was or, for that matter, whether the
king and rook, presently in their starting positions, are still
eligible for castling.
FEN allows the inclusion of that information, as well as move number and player to move.
Alex Relyea
In the problem-solving world, I believe it is the convention that castling should be assumed legal unless you can prove it is illegal, but that en passant should be assumed illegal unless you can prove it is legal.
Once upon a time in the TD room at the National Open, one of the TDs presented a mate-in-3 problem (or something like that). The rest of us tried various moves for white and black (which, incidentally, is considered not the professional method of trying to solve a mate-in-N problem). At one point, when one of us tried O-O-O for black, the presenter of the problem immediately said “Illegal.” The problem then became one of proving that, indeed, castling was illegal. We figured it out fairly quickly after that.
Bill Smythe