Source evaluation
Not all sources deserve equal trust, and the credible-looking ones are
often the trap. Source evaluation is the habit of asking, before you rely
on anything, who is behind it, what backs their claim, what they want you
to believe, and whether it is current, so your conclusions rest on solid
ground.
Method
- Identify the author and their authority. Who produced this, and are
they in a position to know? Expertise, credentials, and a track record
in the specific domain matter; a confident stranger and a domain expert
are not equal. Anonymous or unattributed claims start at low trust (see
web-research's lateral reading to check them).
- Check the evidence behind the claim. Does the source show its work
(data, citations, methodology, primary references) or just assert? A
claim with checkable evidence beats an eloquent one without. Follow the
citations: a source citing strong primary evidence is stronger than one
citing nothing or citing other opinions (see reading-papers,
writing-with-evidence).
- Assess bias and motivation. What does the source want, and how might
that shape what they say? A vendor on their own product, an advocate for
their cause, an outlet with a known slant: bias does not make a source
wrong, but it tells you what to scrutinize and where to seek a
counter-source. Follow the incentives (see mental-models' incentives).
- Weigh source type by the claim. Peer-reviewed research and official
primary data for facts; reputable journalism for events; practitioner
accounts for real-world experience; forums and social media for leads,
not conclusions. Match the source's authority to the weight of the
claim you are hanging on it.
- Check recency and currency. Is it current enough for a claim that
changes over time (technology, prices, best practices, contested
science)? An authoritative source from years ago can be authoritatively
outdated. Note the date and whether newer work supersedes it.
- Prefer the primary and corroborated. Trace claims to their origin
rather than trusting the retelling (see web-research), and for important
or contested claims, require independent corroboration (see
fact-checking). A single source, however credible, is a single point of
failure.
Boundaries
- Evaluation manages risk; it does not make any source certain. Even good
sources err, so keep your confidence proportional to the evidence and
flag uncertainty (see the honesty ethic throughout).
- Credibility is domain-specific: a source authoritative on one topic is a
layperson on another. Evaluate per claim, not by the source's general
reputation.
- Speed-versus-rigor scales with stakes: a low-stakes fact needs a quick
check; a decision-shaping or public claim needs full evaluation and
corroboration (see fact-checking).
1---2name: source-evaluation3description: Judge whether a source is credible and relevant: authority, evidence, bias, and recency. Use when deciding whether to trust and cite a source, in any research.4---56# Source evaluation78Not all sources deserve equal trust, and the credible-looking ones are9often the trap. Source evaluation is the habit of asking, before you rely10on anything, who is behind it, what backs their claim, what they want you11to believe, and whether it is current, so your conclusions rest on solid12ground.1314## Method15161. **Identify the author and their authority.** Who produced this, and are17 they in a position to know? Expertise, credentials, and a track record18 in the specific domain matter; a confident stranger and a domain expert19 are not equal. Anonymous or unattributed claims start at low trust (see20 web-research's lateral reading to check them).212. **Check the evidence behind the claim.** Does the source show its work22 (data, citations, methodology, primary references) or just assert? A23 claim with checkable evidence beats an eloquent one without. Follow the24 citations: a source citing strong primary evidence is stronger than one25 citing nothing or citing other opinions (see reading-papers,26 writing-with-evidence).273. **Assess bias and motivation.** What does the source want, and how might28 that shape what they say? A vendor on their own product, an advocate for29 their cause, an outlet with a known slant: bias does not make a source30 wrong, but it tells you what to scrutinize and where to seek a31 counter-source. Follow the incentives (see mental-models' incentives).324. **Weigh source type by the claim.** Peer-reviewed research and official33 primary data for facts; reputable journalism for events; practitioner34 accounts for real-world experience; forums and social media for leads,35 not conclusions. Match the source's authority to the weight of the36 claim you are hanging on it.375. **Check recency and currency.** Is it current enough for a claim that38 changes over time (technology, prices, best practices, contested39 science)? An authoritative source from years ago can be authoritatively40 outdated. Note the date and whether newer work supersedes it.416. **Prefer the primary and corroborated.** Trace claims to their origin42 rather than trusting the retelling (see web-research), and for important43 or contested claims, require independent corroboration (see44 fact-checking). A single source, however credible, is a single point of45 failure.4647## Boundaries4849- Evaluation manages risk; it does not make any source certain. Even good50 sources err, so keep your confidence proportional to the evidence and51 flag uncertainty (see the honesty ethic throughout).52- Credibility is domain-specific: a source authoritative on one topic is a53 layperson on another. Evaluate per claim, not by the source's general54 reputation.55- Speed-versus-rigor scales with stakes: a low-stakes fact needs a quick56 check; a decision-shaping or public claim needs full evaluation and57 corroboration (see fact-checking).
Amey-Thakur