Patch speed
How quickly problems are acknowledged and addressed, calibrated to severity.
INDEPENDENT POST-LAUNCH ACCOUNTABILITY
Will they take care of the game after you buy it?
We turn years of patches, promises, regressions, and communication into a clear answer—and show the evidence and uncertainty behind it.
96 audited
SEARCH ANY GAME · OPEN
Research does the legwork. A human owns the judgment.
Published audits and their evidence stay free. Request any game — we identify the title, assemble a cited AI-assisted evidence brief, and a human reviews it before anything publishes. Track every request in the public queue.
Identify→Gather evidence→Human review→Publish
01
Verified evaluations
Sixty-six verified scores. No automated publication, no publisher-paid placement — every judgment human-approved.
02
Complicated research. Simple answer.
The score measures the response—not whether the game was good.
Each category is worth 20 points. Launch condition is always shown separately so an exceptional recovery never erases a bad launch. Missing evidence is marked as a limitation, never silently converted into zero.
Maintenance quality
Whether fixes work, prevent regressions, and preserve platform compatibility.
Communication
Clarity, honesty, cadence, and specificity of post-launch updates.
Player responsiveness
Whether documented player problems materially influence action.
Follow-through
Whether concrete commitments are completed and owner interests are respected.
Four evidence tiers
- Tier 1: official patch notes, support pages, and first-party statements.
- Tier 2: reproducible independent technical testing and archival reporting.
- Tier 3: multiple corroborated player reports tied to a version and platform.
- Tier 4: anecdotes used as research leads, never as standalone scoring evidence.
Every judgment needs a reason
Each category is scored 0–20 against published, observable anchors — from "critical problems unaddressed for 6+ months" at the bottom to "critical issues hotfixed within days, consistently" at the top. Two researchers given the same evidence should land within ±3 points per category. Read the full methodology v1.0 →
Confidence is not praise
Confidence describes evidence completeness—not how good the studio performed. A score can be high with Medium confidence, or poor with High confidence. Sparse independent corroboration lowers confidence.
Attribute control carefully
Developer and publisher actions are separated when reliable evidence identifies who controlled the decision. When responsibility is unclear, the audit says so instead of guessing.
03
The history after launch matters.
Evidence first
Official records begin the audit. Reproducible testing and corroborated reports challenge the studio’s version of events.
Context preserved
Regressions, abandoned versions, delistings, and broken promises remain part of the permanent record.
Independence protected
Publishers cannot buy audits, placement, faster resolution, or a better score.
Scores need a paper trail.
Every future revision must preserve the previous score, new score, date, methodology version, reason, and supporting evidence.
Factual challenges are welcome.
No score will be marked Verified until a public correction and appeal channel exists. Editorial judgment remains independent from factual correction. To challenge a score or submit evidence, use the audit request form and choose "Correction".
The archive is incomplete.
The visible source ledgers are starting sets, not complete audit files. New audits publish as Provisional until they clear the verification bar; every evaluation currently live has been verified.
AI assists research. It does not publish judgment.
AI may locate, extract, organize, and flag evidence. It cannot independently publish a score, mark an audit Verified, or replace source checking and human editorial approval.