How can I tell whether an online post mixes established facts with speculation?

Start by splitting the post into separate claims. A single sentence may contain a checkable detail, an explanation of what it means, and a guess about what happens next. Label each claim on its own rather than assigning one label to the whole post.

Use four labels:

1. Supported assertion: A checkable statement for which the post gives identifiable evidence. This means evidence is offered—not that the claim has been independently established.
2. Interpretation: An explanation, conclusion, or judgment drawn from details. It may be reasonable, but it is not the same as the details themselves.
3. Prediction: A statement about what may or will happen. Treat it as a forecast, even when phrased confidently.
4. Open question: Something the post asks or leaves unresolved.

If a claim has no evidence in the post, don’t promote it to “supported.” Mark it “assertion—evidence not shown” or leave its support status unresolved. That keeps your notes honest about what you have and haven’t checked.

Look for a traceable basis for the exact claim: for example, a quoted document, a named record, or a link that leads to material relevant to the statement. Then ask whether that material actually says what the post says it says. A link by itself is not enough; it might be unrelated, incomplete, or about a narrower point.

Separate these questions in your notes:

• What does the post claim? Write the claim in plain, neutral wording.
• What support does it show? Record the source or say “none provided.”
• What does that support cover? Note limits, missing context, or whether the source addresses only part of the claim.

If you haven’t opened or checked the supporting material, say so. You can label the post’s claim as “evidence cited, not checked” rather than implying you verified it. This method helps classify the post; it does not establish the truth of a claim on its own.

Here’s a fictional example, not a report about a real event: “The community notice says the library will close at 6 p.m. Friday. That proves the new schedule is working. The library will probably extend evening hours next month. Why hasn’t the city announced the full schedule?”

You could annotate it like this:

• Supported assertion—evidence cited, not checked: The notice reportedly says the library will close at 6 p.m. Friday. The post points to a notice, but this wording alone doesn’t establish whether the notice exists or whether it is current.
• Interpretation: “That proves the new schedule is working.” The closing time alone does not show what “working” means or demonstrate that conclusion.
• Prediction: “The library will probably extend evening hours next month.” “Probably” signals a forecast, not an established event.
• Open question: “Why hasn’t the city announced the full schedule?” The post asks something without answering it.

A careful summary could be: “The post says a notice lists a Friday 6 p.m. closing time, interprets that as evidence the schedule is working, predicts possible longer hours next month, and asks about a full schedule. The notice and prediction have not been verified here.” That preserves the distinction without endorsing the post.

Copy this template and fill one row for each distinct claim:

Claim: [neutral wording] Label: [supported assertion / interpretation / prediction / open question] Evidence shown: [specific material, or none provided] What it supports: [scope or limitation; “not checked” if applicable] Careful summary: [wording that keeps the label and uncertainty visible]

Before sharing a summary, compare each sentence with your notes. Keep forecasts in forecast language (“the post predicts…”), identify interpretations as interpretations (“the author argues…”), and describe unanswered questions as unanswered. If the post gives no evidence for a checkable statement, don’t silently present it as established. When you cannot tell which label fits, mark it “unclear” and explain what information is missing rather than guessing.