How to Verify a Theoretical Chess Endgame

The Theoretical Endgame Card

Use this card whenever a guide, lesson, book or diagram makes an exact endgame claim.

Identity

• Name or family:
• FEN:
• Side to move:
• Material:
• Halfmove clock: include when move-count rules can matter.

Classification

Choose one:

• exact/theoretical — the result can be verified exhaustively or by established exact theory;
• principle-driven — the position is taught mainly through transferable technique;
• hybrid — practical play is expected to transpose into an exact theoretical family.

Result and Assumptions

Record:

• win / draw / loss from a stated perspective;
• rule assumptions;
• whether tablebase coverage applies;
• service/files/version and verification date;
• WDL and DTZ when relevant.

Defining Conditions

Do not record a named position without the conditions that make the method work. Depending on the family, this may include:

• king squares;
• pawn file and rank;
• promotion-square color;
• rook checking distance;
• king cut-off;
• side to move;
• reserve tempi;
• defender access to a corner;
• available pawn breaks;
• stalemate or perpetual resources.

Human Method

Write the method as a short causal sequence, not a database dump. For example:

1. restrict the enemy king;
2. improve the king while preserving checking shelter;
3. force the defender's rook to a worse checking zone;
4. advance only when the theoretical transition is favorable.

Then record the defender's best resource and any only-move moments.

Failure Case

Every theoretical card should include at least one near-position that flips the result or invalidates the method. This is one of the strongest safeguards against cargo-cult theory. Change side to move, shift a king, move the pawn to another file or alter checking distance, then verify again.

Family-Specific Checks

K+P vs K

Record key squares, pawn file/rank, side to move, promotion-square control, rook-pawn exception and stalemate possibility.

Opposition / Zugzwang

State what would happen if the side to move could pass. If passing would not solve the problem, do not label the position zugzwang merely because all moves are unpleasant.

Pawn Race

Continue through promotion, promotion with check, immediate queen capture, skewer and underpromotion when relevant.

Wrong Bishop + Rook Pawn

Verify promotion-corner color, defending king access, extra pawns and whether the attacker can force the king away.

Lucena / Philidor / Vancura

Record pawn file/rank, king positions, rook positions, checking distance and side to move. The name is not sufficient.

Fortress

List every plausible penetration, pawn break, exchange and zugzwang attempt. A blocked snapshot is not proof.

Tablebase and Engine Use

Use tablebases as exact authority only inside their covered conditions. Engines are excellent analytical tools but are not a substitute for exhaustive proof outside tablebase coverage. When engine analysis supports a non-tablebase theoretical claim, describe it as analysis, not exact proof.

Final Publication Checklist

Before freezing an exact endgame asset, confirm:

• FEN parses correctly;
• side to move matches the text;
• the diagram and prose describe the same position;
• the claimed result has been re-verified;
• the named method's defining conditions are present;
• a near-position or failure case has been tested;
• move-count assumptions are recorded when relevant;
• the human method is understandable without the verifier.

Why Reproducibility Matters

Endgame claims are unusually sensitive to tiny details. A position can flip from win to draw because the move belongs to the other side, a king stands one square farther away, a rook has one less file of checking distance or the halfmove clock changes the legal practical result. A reproducible card lets another editor, coach or future version of the guide reconstruct exactly what was verified.

Record the Perspective Explicitly

Never write only “winning” or “drawn” in a technical record. State winning for White, winning for Black or draw with correct play, and make sure the side to move is part of the record. If the tablebase service reports results from the side-to-move perspective, translate that convention carefully into prose.

Separate Exact Result From Teaching Method

A tablebase may identify several winning moves, one of which optimizes DTZ but is awkward for humans. The published guide should distinguish:

• exact result — what perfect play proves;
• critical move set — which moves preserve the result;
• human method — the simplest reliable plan a player can reproduce;
• failure move — a natural move that throws away the win or draw.

This separation keeps technical accuracy without turning the article into machine output.

Version the Verification Context

When external computational data is used, record the source and verification date. Tablebase coverage, interfaces and supported metrics can change over time. The chess position does not change, but the tooling used to substantiate the publication claim may.

Near-Position Testing

For every named theoretical pattern, test at least one neighboring case. Good candidates are:

• side to move reversed;
• attacking king shifted one square;
• defending rook moved from side-check to rear-check geometry;
• pawn moved one file toward or away from the rook file;
• one reserve pawn move added or removed;
• halfmove clock moved close to the claim threshold.

If the explanation cannot tell the reader why that nearby position differs, the article probably contains memorized labels rather than understood theory.

Asset Freeze Rule

A final board asset should be generated only after the FEN, orientation, side to move and theoretical result are frozen together. If the diagram is later mirrored for composition, re-verify the mirrored position rather than assuming symmetry preserves every rule-state detail.

<!-- ASSET: chess-g-224-card-01 -->