Why we'd rather our coach say nothing than guess

Ask a general chatbot about a chess position and it will answer confidently. It will also, fairly often, invent an evaluation, describe a threat that isn’t there, or suggest a move that isn’t legal. It sounds like a coach. It is not checking anything against the actual board.

For most questions that’s a minor annoyance. In coaching, it’s disqualifying. If you can’t trust that the evaluation is real, you can’t trust the lesson built on top of it. A confident wrong answer is worse than no answer, because it teaches you the wrong thing and feels like it taught you the right one.

The rule is simple: checked, or not said

Every concrete claim Socratic Chess makes about a position, an evaluation, a best move, a line, comes from actually running that position through a chess engine first. Nothing is recalled from training or pattern-matched from similar-looking games. If a claim can’t be checked against the position on the board, the coach stays quiet rather than guess.

★A coach that’s fluent but wrong is worse than one that says nothing.

The same principle applies to tone, not just facts. A coach that praises every move to keep you engaged isn’t coaching, it’s flattering, and flattery that isn’t backed by the position is its own kind of dishonesty. Praise here never outruns what the position actually shows. A move that only holds the balance doesn’t get called brilliant. A losing position doesn’t get softened into “some pressure.”

What this rules out

It rules out inventing a number when the true one isn’t known yet. It rules out guessing at a threat that hasn’t been verified. It rules out telling you a move was fine because that’s the easier thing to say. None of that is generous. It’s just dishonest, and it wastes the one thing you actually came for: the truth about your game.

This is checkable, not just claimed. Bring a game over and ask a hard question about it, see what comes back.