Engage with the evidence

An observation and an account of how it came to be are related, but distinct. A useful explanation must do more than accommodate an observation after the fact.

What counts as evidence

Evidence, in the sense that matters for comparing explanations, is a fact that bears on the relative credibility of competing accounts. A fact every rival explanation accommodates equally well has no power to discriminate between them, whatever its intrinsic interest. Evidence that does useful work is a fact some explanations expect and others do not.

Observation and explanatory story

A measurement, a formation's composition, the position of a fossil, is not the same thing as the sequence of past events proposed to account for it. The measurement is comparatively secure. The sequence connecting it to a cause is an inference, and an inference carries assumptions a bare measurement does not. Keeping these apart is not skepticism about measurement; it is precision about which part of a claim was observed and which part is interpretation. Methodology develops this distinction further.

Predicted in advance, or explained after the fact

An explanation that predicts a result before it is observed has taken a real risk, because the prediction could have failed. An explanation constructed or revised to accommodate a result already known has taken no equivalent risk, however well it now appears to fit. Both have a place in serious inquiry, but they do not carry the same evidential weight, and a summary that treats them as equivalent is overclaiming. For any explanatory claim on this site, readers should ask whether it anticipated the evidence or was shaped to fit it after the fact.

What makes evidence discriminate

Evidence discriminates between rival explanations only where the explanations make different claims about what should be observed. Where two frameworks converge on the same expectation, the evidence that follows cannot be used to prefer one over the other, however strongly it is emphasized on either side. Evidence-domain pages on this site will state what each compared framework expects, not only what one framework predicts correctly in isolation.

A reading checklist

1. Identify the measurement or observation and its primary source.
2. State the assumptions used to interpret it.
3. Compare what different explanations predict.
4. Ask which result would count against the explanation.
5. Record the unresolved burden as well as the proposed response.

Evidence-domain summaries are being developed for Earth history, cosmology, and biology. For technical claims and current status, consult the canonical research programmes.