FREE EDITION / Story / 5 MIN
Follow the Link, Not the Logo
Trace evidence beyond the page that wants your trust, and distinguish repetition from corroboration.
By the Oddoscope editorial desk. Times include reflection and are estimates.
A convincing doorway
The fictional page is called Civic Evidence Review. Its seal resembles a university emblem, its articles have numbered footnotes, and the About page describes a commitment to public understanding. None of those features proves that its current claim is false. None establishes the claim either. They are features of a presentation. Its headline says that a new bench design has reduced complaints in a square. You could spend ten minutes inspecting the seal. Or you could ask what the claim depends on. A footnote leads to an organisation's announcement, which leads to a two-week pilot log. The useful doorway is the evidence trail, not the decoration around it.
Leave the page for a moment
Wineburg and McGrew's original study compared how experienced fact checkers, historians and students evaluated websites. The fact checkers often left a site to investigate it elsewhere, rather than relying on its own appearance. This supports trying an outward-looking check; it does not guarantee that any search result is reliable or that a short checklist certifies truth. For our invented page, open three questions instead of three random tabs. Who operates it? What is the earliest source for this particular claim? Is there independent evidence that could contradict it? The destination of a link matters more than the number of links you collect.
Draw the source family
Imagine that four articles all cite the pilot announcement, and the announcement summarises the same two-week log. Draw arrows from each article to the announcement, then from the announcement to the log. You have a family tree with one root, not five independent measurements. Count roots for the question you are asking, not logos. The log might still be useful. Read its scope: which square, which dates, which complaints and which comparison? An announcement can accurately describe a small pilot while a headline expands it into a general claim. Independence is not a badge a whole organisation owns forever. Two reports can be independent about one observation while repeating each other about another.
Build a compact evidence ledger
Allow three minutes to make a four-line ledger for the fictional bench claim. Line one: the exact claim you would be prepared to repeat. Line two: the earliest evidence you have actually examined. Line three: its scope and the strongest relevant limitation. Line four: a next source that could change your conclusion. Try a bounded sentence: “The pilot log records fewer complaints during these two weeks; the record does not separate the bench change from other changes in the square.” This is not a finding about a real town. It is practice keeping attribution attached to scope. Finish by deleting one tab that merely repeats another source. You have not made the evidence weaker by reducing the pile. You have made its structure easier to see. The Confidence Ledger will now ask you to respond when a genuinely different piece of evidence arrives. You do not have to keep your first answer for the sake of consistency, and you do not have to change it just because another page uses a louder headline.
Sources and editorial notes
Wineburg & McGrew, Lateral Reading and the Nature of Expertise — Original research on evaluating websites; the source-tracing routine here is an editorial adaptation, not a tested assessment.
Original Oddoscope writing, fictional scenarios and code-authored optical artwork. Linked research is cited, not reproduced. Exercises are educational, not psychological diagnoses or validated ratings.