All Position Reviews
Position Review · Welsh correspondence chess

Clough vs Coleman: Defusing a King Hunt

Three tense choices from a prize-winning correspondence game: reject a poisoned pawn, answer mate before taking material, and keep the king under pressure.

8 min readBCCA William Mason Championship 2009Fred Clough - George Coleman
By

Reproducible analysis run complete

6 saved models on NVIDIA GeForce RTX 5080 · Stockfish 18 · 250,000 nodes per checked line.

Download evidence JSON

Why this game?

A public thank-you to Welsh correspondence chess

Austin Lockwood kindly invited Spider Chess to study a game from the Welsh Correspondence Chess Federation website. This review thanks WCCF by linking the exact Fred Clough page and turning one of its complete correspondence games into an interactive lesson about checks, threats, and king safety.

White
Fred Clough
Black
George Coleman
Event
BCCA William Mason Championship
Result
½-½

Interactive comparison

Choice one: counterattack without cashing in

Measured run

Black to move after 15.Nd4. White has pressure in the centre, while Black can choose between an immediate queen capture and a developing counterattack.

Position FEN
r3kb1r/p2bn1pp/1p2ppq1/8/3NP3/1Q1BB3/P4PPP/1N3RK1 b kq - 3 15
Try This Position

Model candidates

Choose a viewpoint

6 profiles

V6 Active: Qxg2+

Measured GPU choice. The check looks forcing, but White can capture the queen and keep the advantage.

Model top three: Qxg2+ · Qxe4 · Qh6

Stockfish 18 safety gate

Complete

Stockfish prefers 15...Rc8 at about +1.22 for White. The played 15...e5 is 143 cp behind that snapshot, while 15...Qxg2+ falls 535 cp behind and loses the queen after Kxg2.

Engine move
Rc8
Depth
14
Node budget
250,228
Evaluation
-1.22

Verification line

15...Rc8 16.Bb5

Position lesson

The visible pawn on g2 is poisoned because the king can take the queen. The played 15...e5 drives the knight and brings the queen toward h3. Stockfish prefers 15...Rc8 in this fixed-node run, but both ideas are far stronger than the model capture.

Interactive comparison

Choice two: answer the threat first

Measured run

White to move after 21...Kb7. Black threatens Qxg2 mate, so every attractive capture must be checked against that immediate danger.

Position FEN
r4b1r/pk4pp/1pR2pq1/1B2p3/4P3/3QB2b/P4PPP/1N4K1 w - - 1 22
Try This Position

Model candidates

Choose a viewpoint

6 profiles

V6 Active: Rxb6+

Measured GPU choice. The check avoids mate, but it lets Black force a draw by repetition.

Model top three: Rxb6+ · Bxb6 · Rxf6

Stockfish 18 safety gate

Complete

Stockfish keeps 22.g3 at about +1.49 for White. 22.Rxb6+ falls to equality, while 22.Rxf6 and 22.Bxb6 allow mate in one.

Engine move
g3
Depth
18
Node budget
250,483
Evaluation
+1.49

Verification line

22.g3 Rc8 23.Qd5 Rxc6 24.Bxc6+ Kc7 25.Nc3 Be7 26.Nb5+ Kb8 27.Ba8 Bc8

Position lesson

22.g3 is the only measured choice that answers the mate and keeps White ahead. Four models choose a rook capture that gives up most of the advantage, while two model choices allow checkmate on the next move.

Interactive comparison

Choice three: keep the checks coming

Measured run

White to move after 25...Be7. The source notes a missed win here, and the fixed-node engine also starts with a forcing knight check.

Position FEN
7r/p1k1b1pp/1pB2pq1/3Qp3/4P3/2N1B1Pb/P4P1P/6K1 w - - 3 26
Try This Position

Model candidates

Choose a viewpoint

6 profiles

V6 Active: Qxe5+

Measured GPU choice. Black can answer with fxe5 and the attack collapses.

Model top three: Qxe5+ · Bxb6+ · Bd7

Stockfish 18 safety gate

Complete

Stockfish prefers 26.Nb5+ at about +2.75 for White. The played 26.Ba8 also remains winning in this fixed-node run, while every model choice fails the 100 cp safety gate.

Engine move
Nb5+
Depth
15
Node budget
250,326
Evaluation
+2.75

Verification line

26.Nb5+

Position lesson

26.Nb5+ keeps Black's king under direct pressure. The played 26.Ba8 still scores as winning in this measured snapshot, but the model choices either lose the advantage or hand it to Black.

Published finish

The forcing line

15...e516.Nb5Bh317.Nd6+Kd718.Bb5+Kxd619.Qd3+Kc720.Rc1+Nc621.Rxc6+Kb722.g3

The published game turns into a race between checks against Black's exposed king and the direct Qxg2 mate threat. White finally spends a move on 22.g3 before continuing the attack.

Review method

Source, models, then verification

01

Start with the WCCF game

Every position comes from the complete Fred Clough versus George Coleman score published by WCCF.

02

Measure six saved evaluators

Each saved checkpoint ranks every legal move on the NVIDIA GPU with the same 30% hybrid positional weight.

03

Apply the Stockfish safety gate

Stockfish checks the engine best move, every distinct model choice, and the move played in the game at a fixed node budget.

What this review can and cannot show

Clear boundaries keep a useful experiment from becoming a bigger claim than the evidence supports.

  • The WCCF source identifies the event as the 2009-10 BCCA William Mason Championship but does not give an exact game date.
  • The six neural checkpoints are scalar evaluators, not move-policy, search, or explanation models.
  • Their inputs encode piece placement only; move history, repetition, clocks, castling rights, and en-passant state are not model inputs.
  • Profile labels describe intended training behavior, not proven human-like personalities or internal reasoning.
  • Stockfish used 250,000 nodes per checked line, two threads, and 256 MB hash. This is a reproducible snapshot, not a claim of perfect play.
  • Two-thread search can produce small score, depth, node-count, and principal-variation differences on another run.
  • The prose describes visible board features and measured move consequences. It does not claim to reveal a model's internal reasoning.

Continue learning

Put the position on your own board

Start from the exact FEN and test whether you can find and convert the idea against a human-style bot.

Try This Position