A citation is a lead, not a proof

A cited answer feels verified. It isn't — it's an answer with a lead attached. The citation tells you where to look, and the looking is the part that produces confidence.

The specific failure is subtler than a fabricated link. The page is real, the page is relevant, the page's title matches the claim — and the page doesn't actually say the thing. A claim about 24-hour study access cites the opening-hours page, which gives term-time hours only. Nothing is fabricated; the support just isn't there.

So build a claim ledger rather than reading a narrative: one row per claim, each with the exact supporting excerpt and the URL. A table makes a weak claim visible, because the excerpt column is either quoting something that settles it or it isn't.

| Claim              | Supporting excerpt        | URL + date |
| Open until 10pm    | "Mon–Thu 09:00–22:00      | .../hours  |
| on weekdays        |  during term"             | 20 Sep     |
| Has 40 study desks | (none found — page says   | .../study  |
|                    |  "study spaces available")| 20 Sep     |

That second row is the point of the exercise. "Study spaces available" doesn't support "40 desks", and in a paragraph the two would have read as the same fact. Correct the claim to what the source supports, or mark it unverified.

Keep genuine gaps as gaps. When one library's page simply doesn't mention power sockets, the honest cell is unknown, plus a note on how a person could find out. A comparison table with no empty cells is more often a sign of guessing than of thorough research.

And check what each statement applies to. An archived announcement about reduced hours and a current information page can both be accurate about different periods. Before using either, establish which dates and which branch each one covers.

Loading your workspace…