Position
Read the position
Nyvera begins with the legal position, the opponent’s reply, and the active strategic objective.
Recorded match replays
Replay Nyvera's games,
one decision at a time.
Step through selected wins, draws, and losses. Follow the recorded plan state, tactical corrections, application record, and measured outcome—one move at a time.
Featured replayMeasured training game
GAME 20312Nyvera wins
1–0Selected measured gamesWins · draws · lossesRecorded decision tracesApplication ≠ later effect
Nyvera · decision proof
Pressure Evidence Gate Result Trace
Read each selected replay in the same order: the pressure, the evidence available then, the legal and correction gate, the measured result, and its recorded trace.
- Pressure
The legal position changes after every opponent reply.
- Evidence
Position, active objective, and recorded opponent reply.
- Gate
Legal moves first; bounded search may record a correction.
- Result
Nyvera's selected move and the measured game outcome.
- Trace
Plan state, correction, application record, and outcome.
BoundarySelected replay evidence; not a chess-strength, rating, benchmark, or later-learning claim.
The replay room
The board shows what happened.
The trace shows what was recorded.
Choose a game, press play, or move one decision at a time. The concise decision panel surfaces recorded objectives, tactical corrections, and bounded application records without turning the replay into a telemetry console or claiming a later effect.
Replay library7 of 7 games
Curated replay setSelected outcomes, not a win rate or whole-cohort summary.
GAME 20312Nyvera plays white
WIN · 1–077 pliesQd8#
These are selected measured training games against the declared opponent configuration. They show recorded outcomes, traces, and bounded application records. Those records do not establish later effect; chess strength requires separate, frozen like-for-like evidence.
The decision loop
A move is a moment.
The trace keeps the sequence.
The useful story is not a single score. It is the recurrent loop from position, to bounded choice, to observed consequence. Any recorded application is kept separate from what happens later.
Correction
Test the candidates
Bounded tactical search can agree with the first choice or record a correction before the move is executed.
Evidence
Separate application from effect
A recorded signal may be applied, rejected, or produce no value change. This replay room reports bounded event counts; whether they improve later play remains unresolved here.
ABOUT NYVERA♞
Chess intelligence built to make
decision-making visible.
Nyvera is an experimental chess research system designed to make bounded decisions under adversarial pressure while keeping its changing plan, correction path, application record, and game outcome inspectable.
It combines legal chess structure, bounded tactical search, strategic objectives, and temporal memory. Stockfish may provide opposition and chess evidence, but Nyvera’s recorded move remains its own governed selection.
This curated replay room for selected wins, true draws, and losses with recorded decision notes, without exposing the full research console or presenting the selection as a whole-cohort performance summary.
01Legal firstEvery replayed action begins inside the legal move field.
02Correction stays visibleSearch agreement, overrides, and plan changes remain distinct.
03Application is not effectThe bounded application record is visible; later improvement remains a separate question.
The broader system
Nyvera is one domain of the larger DeltaX research program.
Chess gives DeltaX a demanding, observable environment for studying recurrent decisions, bounded corrections, and temporal memory. Nyvera’s games are chess evidence; they do not transfer capability or authority to another domain.