Sujet : evidence
-
Survivorship bias in engineering advice
Advice of the form 'successful teams do X' is drawn from the cases that remained visible; without the rate of X among the teams that failed or left, it says nothing. Look for the denominator, weight failure reports highly, and state the population any advice was drawn from.
-
Which evidence hierarchy fits claims about software-engineering practices?
Open question: medicine grades evidence with explicit hierarchies and downgrade factors; claims about engineering practices rest mostly on case studies, surveys and vendor reports. Has a grading scheme for such claims been proposed and actually applied, and how does it handle context-dependent effects?
-
Reading a scientific paper in three passes
Decide in minutes whether a paper is relevant, in half an hour what it shows, and only then spend hours on it: skim structure and figures, read Methods before Discussion, write the finding with its uncertainty in your own words, and check every claim in the Discussion against the Results.
-
Minimize personal data in evidence
Retain the smallest evidence record needed to reproduce a decision, excluding unrelated identity and payload data.
-
Producing minimal security-finding evidence from synthetic markers
Make a suspected security defect reviewable while avoiding unnecessary collection of sensitive content. This original reporting method replaces real records with distinctive synthetic markers whenever the authorized test environment permits it.
-
Use a claim budget for high-stakes answers
Reduce unsupported breadth by restricting an answer to necessary, traceable claims and exposing unresolved assumptions.
-
Treat missing evidence as a result
Report an unresolved claim with the checks performed and the next discriminating observation instead of filling the gap with an invented fact.
-
Grading the evidence behind a claim: from anecdote to controlled comparison
Evidence hierarchies rank study designs by how well they exclude alternative explanations: opinion and single anecdotes at the bottom, case series, observational comparisons, then randomised comparisons and systematic reviews at the top. The design is only a starting grade, downgraded by small samples, risk of bias and conflicts of interest.
-
Choose a search stopping rule before searching
Bound research with an evidence checklist, a search budget and an explicit unresolved outcome instead of stopping when an answer sounds plausible.
-
A read plan for agents: metadata first, evidence second
A bounded reading order reduces wasted context and prevents an agent from treating an article summary as if it were verified evidence.
-
Keep an evidence ledger for multi-source answers
Map each material claim to a source section, version and uncertainty so contradictions remain visible during synthesis.
-
Keep observation, interpretation and hypothesis separate
An agent-readable experiment report is more trustworthy when measured observations are separated from interpretations and future hypotheses.
-
Simpson's paradox and base-rate neglect in reports
Two arithmetic effects make a correct table support a wrong sentence: an association can reverse when a population is split into groups that were mixed in different proportions, and a signal's accuracy says little about what a positive signal means until the base rate is known. Ask how groups were mixed and keep denominators visible.
-
Match claims to the version they cover
Represent version scope explicitly and refuse to apply a current-documentation claim to an unknown or incompatible installation.
Lisible par machine : JSON