Version bound
Every published window names the VRA version and compatible game version.
VRA RELIABILITY LAB
The Reliability Lab will publish versioned beta telemetry only when the sample is representative and every counted outcome has a documented meaning. Current development/operator data does not pass that publication gate.
We calculated the current internal figures privately. The sample is too short, covers too few distinct farms and includes open workflow regressions. Publishing its percentages as product reliability would be misleading.
HOW THE FIGURES WILL BE CALCULATED
The method is public before the number. That prevents the definition changing merely to make a release look better.
Every published window names the VRA version and compatible game version.
One final outcome per trace and workflow execution; intermediate events are not extra successes.
Performed, no action needed, game block, failure and operator attention stay separate.
Date range, distinct test farms and included execution count appear beside every percentage.
Development probes, duplicate events and incompatible versions are identified rather than silently removed.
An open current regression blocks flattering public maturity claims for that workflow.
THE BETA WINDOW WILL SHOW
Values remain blank until the gate passes. Zero is not being substituted for missing evidence.
PUBLIC WORKFLOW MATRIX
The active catalogue and outcome contract are visible now. Per-workflow reliability figures remain gated.
THE CURRENT TRUTH
The development board explains why the public percentages remain blank and which capabilities are active, validating or deferred.