Loptr Lab · Experiment review · October 2026

Veiled Dominion: Duet

The Google Cloud trial ended. Local Classic play is available below while we analyze the earlier experiment and continue development without a required AI service.

LOCAL PLAY RECOVERY AVAILABLE · CLOUD GAMEPLAY SERVICES OFFLINE

The Google Cloud trial has ended. Gameplay services are offline while we analyze the experiment’s results and plan continued development without a required AI service. The separate browser recovery supports two people on one device or a local computer opponent. Remote matches, accounts and cloud saves remain offline.

Play DUET · local recovery

01 · Why the server is offline

The experiment needs an honest evaluation before another deployment.

The shutdown is a pause for evidence and engineering—not a claim that the idea failed or that nobody played. Cloud Run request activity alone cannot tell us how many people played, finished matches, returned, or encountered broken rules.

Rules and game behavior

Duet is a chess-derived variant, not standard chess. Its custom rules must be tested against the intended design, including piece movement, Veil and Sanctuary, Rebirth/Death interactions, promotion, and consistent behavior between local, remote, and AI play.

Reliable, accessible play

Remote matches, reconnection, persistence, keyboard operation, and screen-reader feedback require real-client acceptance tests. Source code or a passing unit test alone is not proof that the full experience works.

Infrastructure and rebuild

The Google Cloud trial ended. The next version is being assessed for reliable play without a required AI service. Hosting and gameplay acceptance remain separate work.

Evidence before investment

We need to recover available usage records and distinguish genuine play from automated traffic, then use those findings to define focused fixes and continued development without an AI service dependency.

02 · What we are evaluating

What did the experiment actually demonstrate?

Reach: how many genuine visitors or players can be established from available evidence.
Activation: whether a visitor started a game or made meaningful moves.
Completion: whether matches reached a recorded outcome.
Return interest: evidence of repeat sessions, returning players, or unsolicited feedback.
Playability: rule correctness, remote reliability, and keyboard/screen-reader usability.
Feasibility: what can be repaired in the existing design versus what needs redesign.

We will report only what the surviving logs, match records, tests, and reviews support. Missing telemetry will be described as unknown—not interpreted as proof of no interest.

03 · Open review

Please review and comment on the rebuild.

We invite players, accessibility testers, chess-variant designers, engineers, and researchers to review the evidence and proposed rebuild. Focused, reproducible observations are especially useful: the rule or interaction, what you expected, what happened, and the steps needed to reproduce it.

Review pull requests Report a finding or comment Read release status Read playtest protocol

Useful review areas: intended variant rules versus implemented rules; accessibility and recovery; evidence quality; and the smallest credible next experiment. Please identify assumptions as assumptions and distinguish confirmed findings from hypotheses.

Project boundaries

A research artifact, with human-led decisions.

Duet is an experimental game and does not automatically establish canon for the wider Veiled Dominion setting. PIXIE is bounded accessibility support for orientation, continuity, and executive-function assistance—not an autonomous player or decision-maker. Humans retain authority over rules, canon, releases, and whether this experiment proceeds.