A balanced approach to imbalances

[Lyudmil Tsvetkov]

Regarding Peter Österlund's question, I don't have WinBoard installed on this computer yet and Shredder will not accept the 'Charge of the Light Brigade' position. I don't know if SCID, for which the engine was really intended can handle it. I should check that. But for now, just in a command window. The RedundantKnightPenalty computation is not very accurate yet I think, but this position is probably still won for the Knights. Just a short search with the original 'Scid Serpent':

Rainbow Serpent Build 025 130624 by Tord Romstad, Marco Costalba and Joona Kiiski
position fen [d]nnnnknnn/2pppp2/8/8/8/8/PPPPPPPP/Q2QK2Q w - - 0 1
go infinite
info depth 1 seldepth 2 score cp -654 nodes 51 nps 17000 time 3 multipv 1 pv b2b
4
info depth 2 seldepth 2 score cp -872 nodes 229 nps 38166 time 6 multipv 1 pv b2
b4 h8g6
info depth 3 seldepth 4 score cp -525 nodes 404 nps 40400 time 10 multipv 1 pv b
2b4 h8g6 g2g4
info depth 4 seldepth 4 score cp -747 nodes 1194 nps 79600 time 15 multipv 1 pv
b2b4 h8g6 g2g4 a8b6
info depth 5 seldepth 6 score cp -501 nodes 1466 nps 69809 time 21 multipv 1 pv
b2b4 h8g6 g2g4 c8b6 h1a8 b6a8
info depth 6 seldepth 6 score cp -501 nodes 1870 nps 69259 time 27 multipv 1 pv
b2b4 h8g6 g2g4 c8b6 h1a8 b6a8
info depth 7 seldepth 10 score cp -488 nodes 2698 nps 79352 time 34 multipv 1 pv

.
.
.

info depth 28 seldepth 50 score cp -125 nodes 1382923347 nps 3009999 time 459443
multipv 1 pv b2b3 f7f6 g2g4 c7c6 f2f4 d7d6 d2d4 h8f7 c2c4 a8c7 e2e4 e7e5 f4f5 e
5d4 d1d4 c8e7 h2h4 c6c5 h4h5 c5d4
info depth 29 currmove b2b3 currmovenumber 1
stop
bestmove b2b3 ponder f7f6

Take the centipawn scores with a few grains of salt

Eelco

Hi Eelco.

This is already too much.
No one has provided sufficient evidence (there is none actually) that even 3Qs vs 7Ns with a full set of pawns is won for one of the sides, and you would claim that a handicap of another 4 pawns for one of the sides would favour the knights. Sorry, but that is already ludicrous. I want data that the standard 3Qs vs 7Ns position is won for one of the sides, and there is none.

There is already evidence that Queeny plays better such an imbalance against other strong engines, but no evidence that is won for either side. It is quite possible that engines simply do not handle the imbalance well.

[hgm]

There is already evidence that Queeny plays better such an imbalance against other strong engines, but no evidence that is won for either side. It is quite possible that engines simply do not handle the imbalance well.

Oh yes, there are thousands of games in evidence for that. They all point to the same fact: when the engine playing the Knights knows the correct strategy (avoiding Q-for-2N trades and marching the King towards the center), and is given sufficient search depth to not be tricked by the deep tactics that with 3 Queens is bound to be common, it does not really matter how much stronger the opponent that plays the Queens is, or how well he does handle it: the Knights always win.

That you are unaware of it, and in denial, or could not read it in the stars, doesn't mean that something doesn't exist...

I want data that the standard 3Qs vs 7Ns position is won for one of the sides, and there is none.

So generate it...

[Lyudmil Tsvetkov]

It takes just 10 moves to win the game for the queen side. It seems almost everybody on this forum suffers from an effect of a phenomenon that actually does not exist. I am fully convinced even the 3Qs vs 7Ns with a full set of pawns is easily won for the queens, but it will take some time to post such games.

[pgn][Event "Computer chess game"]
[Site "OWNER-PC"]
[Date "2013.10.26"]
[Round "?"]
[White "owner"]
[Black "QN"]
[Result "*"]
[BlackElo "2000"]
[Time "23:38:25"]
[WhiteElo "2400"]
[TimeControl "300"]
[SetUp "1"]
[FEN "nnnnknnn/2pppp2/8/8/8/8/PPPPPPPP/Q2QK2Q w - - 0 1"]
[Termination "unterminated"]
[PlyCount "30"]
[WhiteType "human"]
[BlackType "program"]

1. a4 Nhg6 {(h8g6 h2h4 b8c6 h4h5 g6f4 g2g4 e7e5 e2e3 f4e6 a4a5 g8f6 b2b4)
-1.75/14 5} 2. b4 d5 {(d7d5 h2h4 b8d7 h4h5 g6e5 f2f4 e5c6 b4b5 c6a5 d2d4
g8f6 g2g4 a8b6) -1.64/14 10} 3. h4 Nbd7 {(b8d7 a4a5 g8f6 g2g4 c7c6 g4g5
f6e4 c2c4 a8c7 c4d5 c6d5 d2d4) -1.76/14 4} 4. h5 Nge5 {(g6e5 f2f4 e5c4 g2g4
c7c6 a4a5 a8c7 h1f3 e7e6 d2d3 c4d6 e2e4 g8f6) -1.73/16 7} 5. Qh2 f5 {(f7f5
a4a5 g8f6 d2d4 e5c4 c2c3 c8d6 f2f3 d8e6 e1f2 f5f4 g2g4 f4g3 f2g3) -0.83/14
8} 6. Qhxe5 Nxe5 {(d7e5 a1e5 e7e6 e5g7 g8e7 h5h6 d8f7 h6h7 e7g6 a4a5 c7c6
c2c3 a8c7 d2d3 c8d6 d3d4) -1.20/17 6} 7. Qxe5 e6 {(e7e6 e5g7 g8e7 h5h6 d8f7
e2e4 c8d6 h6h7 e7g6 e4d5 e6d5 a4a5 d6e4 d2d3 e4g5 d3d4) -1.94/18 11} 8. Qg7
Nge7 {(g8e7 h5h6 d8f7 e2e4 d5e4 a4a5 f8g6 h6h7 f7h8 d1a1 e6e5 a1e5 e8d8
e5f6 c8d6 d2d4) -2.07/18 7} 9. h6 Nf7 {(d8f7 h6h7 e7g6 e2e4 c8e7 d1h5 d5e4)
-2.53/16 5} 10. e3 c6 {(c7c6 h6h7 f8h7 g7h7 a8c7 a4a5 c8d6 d2d4) -2.93/15
6} 11. Qh5 Neg6 {(e7g6 h6h7 c8e7 a4a5 a8c7 g7f7 e8f7 h7h8 c7a6 h8d4 a6c7
d2d3 f7g8 d4b6 c7b5 e3e4) -3.26/18 11} 12. Qhxg6 Nxg6 {(f8g6 g7g6 c8e7 g6e6
a8c7 e6f6 c7a6 c2c3 a6b8 h6h7 b8d7 h7h8 f7h8 f6h8 e8f7 b4b5 c6b5 a4b5 d7f6
d2d4 f6e4) -4.97/20 6} 13. Qxg6 Ne7 {(c8e7 g6e6 a8b6 a4a5 b6d7 h6h7 d7f8
e6f7 e8f7 h7h8 f8g6 h8h7 f7e6 d2d4 g6f8 h7h6 f8g6) -4.89/19 4} 14. Qg7 Nxh6
{(f7h6 g7h6 e8d7 e1e2 a8c7 d2d4 d7d6 e2d3 c7a6 c2c3 a6c7 a4a5 c7b5 h6g7
g8e7 f2f4) -4.19/19 4} 15. Qxh6 Kd7 {(e8d7 f2f3 a8c7 d2d4 c7e8 a4a5 e8d6
g2g4 f5g4 f3g4 d6f7 h6g7 f7d6 e1e2 d6e4 g4g5 d7d6 e2d3) -4.49/20 4} *
[/pgn]

[d]n1nnknn1/2p3Q1/4p3/3p1p1P/PP6/8/2PPPPP1/3QK3 b - - 1 8
8 moves from the start of the game

[d]n7/3kn3/2p1p2Q/3p1p2/PP6/4P3/2PP1PP1/4K3 w - - 1 16
And 15 moves from the start of the game

[hgm]

Note that I would not claim that with 4 extra Pawns for the Queens it would be won; especially if some of these Pawns are protected passers that can march to near promotion before the Knights even have a chance to deploy. That obviously is an even far bigger advantage than having 8 vs 8 Pawns all spread out. Which also makes it ore difficult for the Knights than with just 4 centralized Pawns each, but not enough to turn the game in favor of the Queens.

The claim is that 7 Knights against 3 Queens with symmetric Pawn chains is a win for the Knights. Nothing more, nothing less. No cheating by giving the Queens some Pawn advantage is allowed.

Adding 4 Pawns to a position and showing that you can then win it proves absolutely nothing about the original position!

[Lyudmil Tsvetkov]

Still have not won in the 3Qs vs 7Ns with full set of pawns as it takes some practice, however here is a game with 2 pawns handicap. I did not win it, and my time elapsed (I should have set an increment, but I did not), but it is evident the queen side has an advantage. So that I think it is obvious that the 7Ns can not give even a single pawn handicap to the queens. The fact that Komodo loses against a modified version of Texel does not mean that the position is lost, it obviously is not.

[pgn][Event "Computer chess game"]
[Site "OWNER-PC"]
[Date "2013.10.27"]
[Round "?"]
[White "owner"]
[Black "QN"]
[Result "0-1"]
[BlackElo "2000"]
[Time "00:11:24"]
[WhiteElo "2400"]
[TimeControl "300"]
[SetUp "1"]
[FEN "nnnnknnn/1pppppp1/8/8/8/8/PPPPPPPP/Q2QK2Q w - - 0 1"]
[Termination "time forfeit"]
[PlyCount "52"]
[WhiteType "human"]
[BlackType "program"]

1. a4 e5 {(e7e5 b2b4 h8g6 h2h4 d7d6 e2e3 b8c6 b4b5 c6b4 d2d4) +2.54/14 8}
2. h4 d5 {(d7d5 b2b4 b8d7 f2f4 h8g6 g2g3 g6e7 b4b5 f7f6 d2d4 d8f7 a4a5)
+2.90/14 9} 3. g4 Nce7 {(c8e7 d2d3 b8c6 a4a5 d8e6 e2e4 d5d4 g4g5 h8g6 a5a6
b7a6 a1a6) +2.94/14 11} 4. b4 Nbd7 {(b8d7 b4b5 f8e6 a4a5 g8f6 a5a6 b7a6
g4g5 f6e4 c2c4 a8b6 c4d5 e6c5) +2.98/14 6} 5. d4 exd4 {(e5d4 e2e3 a8b6 e3d4
f8e6 a4a5 b6c4 b4b5 c7c6 b5b6 g8f6 g4g5) +3.27/14 6} 6. Qdxd4 Nfe6 {(f8e6
d4d1 f7f6 a4a5 c7c6 e2e3 a8c7 f2f4 f6f5 g4f5 e7f5 e3e4 d5e4 h1e4) +3.39/15
8} 7. Qa7 Nab6 {(a8b6 a4a5 b6c4 b4b5 h8g6 g4g5 g6e5 a7a8 f7f5 a5a6 b7a6
b5a6 d7b6 a8a7) +3.96/15 8} 8. a5 Nc4 {(b6c4 c2c3 h8g6 a5a6 b7a6 e2e4 g8f6
h4h5 g6f4 e4d5 e7d5 a7a6) +4.07/14 4} 9. a6 Nec6 {(e7c6 a6b7 c6a7 a1a7 g8f6
g4g5 f6e4 f2f3 e4c3 b7b8 d7b8 a7b8 c4e3 e1d2) +4.11/15 7} 10. axb7 Nxa7
{(c6a7 a1a7 g8f6 g4g5 f6e4 f2f3 e4c3 b7b8 d7b8 a7b8 h8g6 e2e4 g6f4 e4d5
f4d5) +4.23/15 7} 11. Qxa7 Ngf6 {(g8f6 b7b8 d7b8 a7b8 c4b6 g4g5 f6d7 b8a7
d8c6 a7a3 e6d4 a3e3 e8f8 e3c3 h8g6 e2e4) +4.28/15 5} 12. b8=Q Nxb8 {(d7b8
a7b8 c4b6 g4g5 f6d7 b8a7 d8c6 a7a3 d7e5 e2e4 e6d4 e1d1) +4.31/16 6} 13.
Qxb8 Ng6 {(h8g6 e2e4 f6d7 b8a7 d5d4 h4h5 g6e5 h1f1 e5f3 e1e2 d7e5 b4b5)
+4.24/16 8} 14. g5 Nd7 {(f6d7 b8b5 g6e7 b5a4 c4b6 a4b3 d8c6 e2e3 d7e5 f2f4
e5c4 h4h5 e7f5) +4.33/16 5} 15. Qa8 Ndb6 {(d7b6 a8a6 g6e7 e2e4 e6f4 h4h5
d8c6 h5h6 g7h6 e4d5 e7d5 b4b5 c6b4 h1e4) +4.14/15 6} 16. Qa1 Nc6 {(d8c6
h4h5 g6f8 a1c3 e6g5 c3g7 f8e6 g7h8 e8e7 c2c3 f7f5 f2f3 e6f4 b4b5) +3.87/15
5} 17. h5 Ngf8 {(g6f8 h5h6 g7h6 g5h6 c6b4 a1c3 e6d4 h6h7 b4c2 e1d1 f8g6
e2e3 b6a4 h7h8 g6h8 h1h8) +3.14/15 3} 18. h6 gxh6 {(g7h6 g5h6 e6d4 a1c1
f8h7 h1g2 d4e6 g2g8 e6f8 c1f4 c6e5 b4b5 e8e7 e2e4 h7f6) +2.96/16 3} 19.
gxh6 Nh7 {(f8h7 a1h8 e6f8 c2c3 c4d6 f2f4 c6e7 h1h5 d6e4 h8g7) +2.70/15 4}
20. Qg7 Nxg7 {(e6g7 h6g7 h7f6 h1h6 e8e7 g7g8 f6g8 h6c6 e7d8 b4b5 f7f5 c6c5
g8e7) +4.49/18 4} 21. hxg7 Nf6 {(h7f6 h1h6 e8e7 h6h4 e7e6 h4h3 e6e5 h3h2
e5f5 h2c7 c6b4 c7f7 c4e5 g7g8 e5f7 g8f7 b4c2 e1d2 c2d4) +4.54/19 5} 22.
Qh8+ Kd7 {(e8d7 h8h3 d7e7 h3h4) +4.60/18 3} 23. g8=Q Nxg8 {(f6g8 h8g8 f7f5
g8g7 c6e7 e1d1 c7c6 e2e3 b6c8 d1e2 c8d6) +4.74/19 4} 24. Qxg8 f5 {(f7f5
g8g7 c6e7 e2e3 c7c6 e1e2 b6a8 c2c3 a8c7 f2f4 c7b5 g7h8 d7d6 e2d3) +4.69/16
4} 25. b5 Ne7 {(c6e7 g8h8 c4a3 h8b2 b6c4 b2b3 d7e6 e1d1 f5f4 b3b4 e7f5 b4c5
f5d6 c5c6) +5.01/18 3} 26. Qf7 Na3 {(c4a3 f7f6 a3c2 e1f1 d5d4 f1g2 b6d5
f6h6 f5f4 h6h7 c2b4 h7g7 d4d3 e2d3 b4d3 g2f3) +6.06/18 4} 27. ... {White
forfeits on time} 0-1
[/pgn]

[d]8/2pknQ2/1n6/1P1p1p2/8/n7/2P1PP2/4K3 w - - 3 27

[Lyudmil Tsvetkov]

Note that I would not claim that with 4 extra Pawns for the Queens it would be won; especially if some of these Pawns are protected passers that can march to near promotion before the Knights even have a chance to deploy. That obviously is an even far bigger advantage than having 8 vs 8 Pawns all spread out. Which also makes it ore difficult for the Knights than with just 4 centralized Pawns each, but not enough to turn the game in favor of the Queens.

The claim is that 7 Knights against 3 Queens with symmetric Pawn chains is a win for the Knights. Nothing more, nothing less. No cheating by giving the Queens some Pawn advantage is allowed.

Adding 4 Pawns to a position and showing that you can then win it proves absolutely nothing about the original position!

I understand, I will win that too, but it takes some time, as the position is very unusual. So far I do not have any problems with 2 pawns less, that is obviously won for the queens, however, claims have been made that it was the other way round.
Please note, that while playing, I am missing in the hurry an enormous amount of tactics for the queens (the knights do not have so many tactics), which means that the queen positions are even more favourable. Objectively, I think that in the initial position with 3Qs vs 7Ns and all pawns the queens have at least some 75 centipawns advantage, which should suffice for a win. It is another thing if the knights are able to organize quickly and defend well, but they do not have that time. Elephantiasis is all about extremely good defence - you can not attack objects that are extremely well defended, and therefore the queen side should play as tactical as possible.

Another game with 2 pawns handicap, obviously favouring the queens, but 5 minutes again are not enough time in a complex position.

[pgn][Event "Computer chess game"]
[Site "OWNER-PC"]
[Date "2013.10.27"]
[Round "?"]
[White "owner"]
[Black "QN"]
[Result "0-1"]
[BlackElo "2000"]
[Time "01:19:55"]
[WhiteElo "2400"]
[TimeControl "300"]
[SetUp "1"]
[FEN "nnnnknnn/1pppppp1/8/8/8/8/PPPPPPPP/Q2QK2Q w - - 0 1"]
[Termination "time forfeit"]
[PlyCount "44"]
[WhiteType "human"]
[BlackType "program"]

1. a4 e5 {(e7e5 b2b4 h8g6 e2e3 a8b6 f2f4 b6c4 f4f5 g6e7 e3e4) +2.46/13 5}
2. h4 d5 {(d7d5 b2b4 b8d7 f2f4 h8g6 g2g3 g6e7 f4e5 d8c6 d2d4 c6b4 e2e4)
+2.84/14 9} 3. b4 f6 {(f7f6 a4a5 h8f7 d2d4 e5d4 a5a6 b7a6 a1d4 a8b6 d4g4
d8e6 e2e4) +3.01/14 9} 4. g4 c6 {(c7c6 g4g5 h8f7 e2e4 c8e7 d2d4 e5d4 d1d4
a8c7 g5f6 g8f6 e4e5 d8e6) +3.20/14 9} 5. e4 dxe4 {(d5e4 h1e4 d8e6 d2d4 e5d4
d1d4 h8f7 c2c4 g8e7 d4d1 a8c7 c4c5) +3.27/14 5} 6. Qxe4 Nde6 {(d8e6 d2d4
e5d4 f2f4 c8d6 e4f3 h8f7 f4f5 e6c7 d1d4 b8d7) +3.07/14 8} 7. d4 exd4 {(e5d4
f2f4 c8d6 e4h1 b8d7 f4f5 e6c7 b4b5 g8e7 b5c6 e7c6 c2c3 d4d3 d1d3) +3.15/14
4} 8. Qaxd4 Nf7 {(h8f7 f2f4 g8e7 d4d2 a8b6 f4f5 e6c7 a4a5 b6d7 c2c4 c8d6
d2d6 f7d6 d1d6) +3.27/14 7} 9. g5 Nge7 {(g8e7 d4b2 c8d6 e4e2 b8d7 b4b5 a8c7
c2c4 e6f4 e2c2 f8e6 b5b6) +4.06/14 7} 10. gxf6 Nxd4 {(e6d4 f6e7 f8e6 c2c3
c8d6 e4g6 d4f5 g6e6 a8c7 e6e2 f5e7 d1d4 e8f8 a4a5 b8d7) +4.65/16 4} 11.
fxe7 Nfe6 {(f8e6 c2c3 c8d6 e4g6 d4f5 g6e6 a8c7 e6e2 f5e7 c3c4 b8d7 b4b5
c6c5 f2f4 d6f5) +4.61/16 4} 12. c3 Ncd6 {(c8d6 e4g6 d4f5 g6e6 a8c7 e6e2
c7d5 d1d2 d5e7 c3c4 f5h4 b4b5 c6c5 a4a5 b8d7 a5a6) +4.39/16 6} 13. Qg6 N4f5
{(d4f5 g6e6 a8c7 e6g6 b8d7 c3c4 d7e5 g6h7 f5e7 h7g7 d6c4 f2f4 e7f5 g7f6)
+4.09/16 8} 14. Qxe6 Nc7 {(a8c7 e6g6 c7d5 c3c4 d5e7 g6e6 f7d8 d1h5 d6f7
e6f5 e7f5 h5f5 b8a6 b4b5 a6b4 a4a5) +3.92/16 4} 15. Qg6 Nd7 {(b8d7 c3c4
c6c5 d1e2 c5b4 f2f3 f5d4 e2e3 d7e5 g6d6 e5f3 e3f3 f7d6) +4.34/15 4} 16. Qd3
Kxe7 {(e8e7 g6f5 d6f5 d3f5 c7e6 f2f4 d7f6 b4b5 c6c5 h4h5 f7d6 f5g6)
+4.21/15 7} 17. Qgxf5 Nxf5 {(d6f5 d3f5 c7e6 f2f4 d7f6 f5g6 e6f4 g6g7 f4d5
e1d2 f6e4 d2d3 d5c3 g7c3 e4c3 d3c3 f7e5 h4h5) +4.11/17 5} 18. Qxf5 Ne6
{(c7e6 f2f4 d7f6 f5a5 f6d5 f4f5 e6f4 c3c4 f4d3 e1d2 d5b4 a5c7 e7f6 c7b7
f7e5 d2c3) +4.09/17 3} 19. a5 Nd6 {(f7d6 f5d3 d7e5 d3g3 e5c4 h4h5 e7d7 h5h6
g7h6 g3g6 e6f4 g6h6) +4.06/17 6} 20. Qg6 Ne5 {(d7e5 g6g3 e5c4 h4h5 e7d7
h5h6 g7h6 a5a6 b7a6 g3g6 e6g5) +3.91/17 3} 21. Qg3 Kf6 {(e7f6 f2f4 e5g6
g3d3 d6f5 d3e4 g6f4 a5a6 b7a6 e4c6 a6a5 b4a5 f5h4 e1d2) +3.86/17 2} 22. f4
Ng6 {(e5g6 g3d3 d6f7 d3e4 g6f4 a5a6 b7a6 e4c6 a6a5 b4a5 f7e5 c6b7 f4d3)
+3.56/18 4} 23. ... {White forfeits on time} 0-1
[/pgn]

[d]8/1p4p1/2pnnkn1/P7/1P3P1P/2P3Q1/8/4K3 b - - 0 23
Here f5 should win for white.

[Uri Blass]

I understand, I will win that too, but it takes some time, as the position is very unusual. So far I do not have any problems with 2 pawns less, that is obviously won for the queens, however, claims have been made that it was the other way round.

Nothing is obvious and being able to win against some weak engine is no evidence.

If you can win some modified version of stockfish that is stronger than QN in this type of position then it can be more convincing and I guess it should be easy to make stockfish significantly stronger than QN in this position by some simple modification of the code.

[Lyudmil Tsvetkov]

I understand, I will win that too, but it takes some time, as the position is very unusual. So far I do not have any problems with 2 pawns less, that is obviously won for the queens, however, claims have been made that it was the other way round.

Nothing is obvious and being able to win against some weak engine is no evidence.

If you can win some modified version of stockfish that is stronger than QN in this type of position then it can be more convincing and I guess it should be easy to make stockfish significantly stronger than QN in this position by some simple modification of the code.

But I thought QN=Queeny is the world champion in the discipline, and results against other engines actually showed it performs better than the top with these imbalances. How much stronger than the top you want to make Stockfish?

Anyway, I will be waiting for your modified version of Stockfish. In the meantime, I will also prepare some modifications.

[Lyudmil Tsvetkov]

I understand, I will win that too, but it takes some time, as the position is very unusual. So far I do not have any problems with 2 pawns less, that is obviously won for the queens, however, claims have been made that it was the other way round.

Nothing is obvious and being able to win against some weak engine is no evidence.

If you can win some modified version of stockfish that is stronger than QN in this type of position then it can be more convincing and I guess it should be easy to make stockfish significantly stronger than QN in this position by some simple modification of the code.

And when you make Stockfish-Elephantiasis, please do not forget also to create another version of Stockfish-Antielephantiasis, with the following features:

- encourage trading of queen for 2 knights at the earliest opportunity
- encourage play on the wings, rather than in the center (playing with pawns on the wings), as it is easier to break through on the wings where the knights can not concentrate in the same amount as in the center
- discourage backward pawns for the queen side, as the knights will land there
- discourage allowing passers for the knight side, as those are usually more dangerous than passers for the queen side
- keep king for the queens as far away from the knights as possible; always keep at least one pawn with the king to shelter it
- discourage blocked positions, as well as symmetrical pawn structures
- allow tactics as deep as possible

When you are ready with both versions, you might conduct a significant test and report the results. I bet Stockfish-Antielephantiasis will win, simply because the position favours it. Please note that engines losing against Queeny and derivatives do nothing of the above, most importantly, they will be waiting for quite some time before trading queen for 2 knights, and after couple of moves the knights already occupy the center, where, in distinction to other places, they defend each other not one or 2 times, but couple of times each other, as well as the friendly pawns, and attacking them becomes practically impossible, while the queens are easily attackable. In this way, engines without special knowledge lose game after game, regardless of their mammoth strength.
The queen side does not have a minute to waste, it should break through before the knights come to the center. Objectively, in the starting position it has plenty of time for this, but not after couple of aimless moves, when the position would already be objectively lost for the queens.

[hgm]

And when you make Stockfish-Elephantiasis, please do not forget also to create another version of Stockfish-Antielephantiasis, with the following features:

- encourage trading of queen for 2 knights at the earliest opportunity
- encourage play on the wings, rather than in the center (playing with pawns on the wings), as it is easier to break through on the wings where the knights can not concentrate in the same amount as in the center
- discourage backward pawns for the queen side, as the knights will land there
- discourage allowing passers for the knight side, as those are usually more dangerous than passers for the queen side
- keep king for the queens as far away from the knights as possible; always keep at least one pawn with the king to shelter it
- discourage blocked positions, as well as symmetrical pawn structures
- allow tactics as deep as possible

This is very useful comment, and when I have time I will certainly try to put it into an improved engine for the N vs Q game. (Too busy preparing my engine for the UEC Cup that will take place on Nov. 24 now, however.)

You should realise that elephantiasis and anti-elephantiasis are one and the same: building in knowledge like you suggest will automatically improve the play of the engine for both the Queen and the Knight side. Knowing what a good attack plan is for the Queens will automatically make the engine try to prevent the execution of that plan when playing the Knights, while otherwise it would allow the opponent to creep up on it, and see the danger tactically only when it is too late. In their analysis engines always play both sides.

QueeNy now only implements your first point, and it tremendously benefits from it (so much that it can beat opponents that are 600 Elo stronger in normal games), both when playing Knights and Queens.

Still, my earlier statement that this is an admission on your part that the Queen value in this position is suppressed to below that of two Knights. Which was really all that elephantiasis was about. So if you agree on that, there isn't really anything left to prove.