How does CueCrux decide which sources to trust?

When CueCrux assembles an answer from multiple sources, it needs to decide which sources to weight most heavily. This is one of the most important things a search engine does, and how it is done matters enormously for the quality and reliability of the answers you receive.

CueCrux uses a multi-signal authority scoring system. Rather than relying on a single metric like page rank or popularity, it evaluates sources across several dimensions and combines them into an overall authority assessment. Here are the key signals.

Publication authority measures the credibility of the source itself. A peer-reviewed journal article scores higher than a blog post. An official regulatory publication scores higher than a third-party commentary. A court decision from a superior court scores higher than one from a lower court. CueCrux maintains a continuously updated taxonomy of source types and their relative authority levels, calibrated by domain. What counts as authoritative differs between legal research, medical research, financial analysis, and technical documentation, and CueCrux accounts for those differences.

Recency captures how current the source is. All else being equal, a more recent source is more likely to reflect the current state of affairs. But recency interacts with authority in nuanced ways. A foundational statute from twenty years ago might be more authoritative than a recent blog post commenting on it. CueCrux's ranking model understands these interactions and does not blindly prefer new sources over established ones.

Citation and corroboration measures how widely a source's claims are supported by other independent sources. If multiple credible sources agree on a point, that convergence increases confidence. If a claim appears in only one source, it is not necessarily wrong, but the answer's confidence assessment will reflect the thinner evidence base.

Domain expertise evaluates whether the source is operating within its area of competence. A medical journal is highly authoritative for clinical questions but less so for legal questions. A financial regulator is authoritative for banking requirements but not for environmental compliance. CueCrux matches source expertise to query domain to ensure that authority scoring is contextually appropriate.

Internal consistency checks whether the source contradicts itself or makes claims that are inconsistent with its other publications. A source that is internally consistent scores higher than one that says different things in different places.

Provenance and transparency evaluates whether the source itself shows its evidence. Sources that cite their own sources, provide methodology descriptions, or link to primary data score higher than sources that make unsupported assertions. This recursive evidence checking is a distinctive feature of CueCrux's approach.

All of these signals are combined into an overall authority score for each source relative to the specific query. The scores are not absolute or permanent. The same source might score highly for one question and lower for another, depending on the topic, the domain, and the competitive landscape of available evidence.

Crucially, CueCrux's authority scoring is transparent. When you receive an answer, you can see not just which sources were cited but how they were ranked and why. The receipt for each claim includes the source's authority assessment, so you can understand why CueCrux chose to weight one source over another. If you disagree with the ranking, you can see the reasoning and form your own judgment.

This transparency distinguishes CueCrux from black-box search engines where you have no idea why certain results appear above others. Google's ranking algorithm is proprietary and opaque. ChatGPT's source selection is invisible because it does not cite sources at all. CueCrux shows you the scoring and lets you evaluate it yourself.

For enterprise customers, authority scoring can be customised. You can tell CueCrux that certain sources should be considered highly authoritative for your specific use case, or that certain sources should be deprioritised. This is useful for organisations that have preferred regulatory interpretations or trusted industry sources. Custom authority preferences are applied on top of CueCrux's baseline scoring, and they are documented in the receipt trail so that anyone reviewing the answer can see that custom preferences were in effect.

Source ranking is not a solved problem and CueCrux does not pretend otherwise. But by using multiple signals, maintaining domain awareness, and being transparent about the ranking process, CueCrux provides a more trustworthy and more useful approach than any search tool that keeps its ranking methodology hidden.