Context Stripping & Provenance Decay

A statement can remain word-for-word unchanged and still become misleading when the information needed to interpret it disappears.

A screenshot loses its URL. A cropped chart loses its denominator. A quotation loses the sentence before it. An old photograph loses its date and location. A study result becomes a social-media card with no link to the paper.

Context stripping is the removal of verification-relevant context. Provenance decay is Claim by Claim’s editorial term for the broader process by which repeated copying, cropping and reposting moves information farther from its traceable origin.

The content may survive. The evidence trail does not.

Provenance is different from the claim itself

Provenance answers questions such as:

• Where did this item come from?
• Who originally published it?
• When was it created or posted?
• Has it been edited, cropped, translated or reframed?
• Is the version being shared the same as the original?
• Was there a later correction, retraction or update?

Those questions do not by themselves determine whether the underlying claim is true. They determine whether the reader can reconstruct what the material actually is before using it as evidence.

An authentic document can support a false interpretation. A fake document can contain a true sentence. Provenance and truth are related but distinct problems.

Source memory is not the same as content memory

Research on source monitoring shows that people can remember information while becoming uncertain or mistaken about where it came from.

That distinction matters in high-volume feeds. A reader may remember the statistic, the quotation or the image after forgetting whether it came from a peer-reviewed paper, a news report, a parody account or an unsourced repost.

Experimental work on social-media presentation has also examined source blindness: conditions in which attention to the content becomes separated from attention to the source.

This is one reason repeated exposure can create the impression of independent confirmation. Several posts may feel like several sources even when they all descend from the same original item.

Screenshots are unusually good at travelling without provenance

Screenshots are portable, visually concrete and easy to repost. They can also remove exactly the information needed for verification:

• the URL or domain;
• publication date;
• author or account identity;
• surrounding thread or article;
• correction notice;
• interactive chart labels;
• links to underlying data;
• clues showing whether the page is authentic.

A screenshot is therefore not weak evidence because it is a screenshot. It is weak when the screenshot is the end of the trail rather than a pointer back to a verifiable original.

Research on misinformation videos has documented how screenshots can be recontextualized and presented as visual evidence, acquiring authority from interface elements and the appearance of documentation even when the surrounding source relationship is unclear.

Common forms of provenance decay

The mechanism appears in several recurring forms:

Screenshot chains. A screenshot of a post becomes a screenshot inside a video, which becomes another screenshot in a new post.

Quote stripping. A sentence is separated from the paragraph, interview question or qualification that determines its meaning.

Date and location loss. Authentic images from an earlier event or another place are recirculated as evidence of a current event.

Document cropping. A table, chart, email or official document is cropped so that title, methodology, units or surrounding text disappear.

Reposting without links. Content is copied into a new platform where the original author, URL and correction history are no longer attached.

Translation and reframing. A claim crosses languages or communities while its wording, certainty or attribution changes.

None of these operations automatically creates misinformation. Each one increases the distance between the audience and the information required to check the item.

Context stripping can create evidence-shaped objects

A cropped artifact often looks more decisive because uncertainty has been removed along with context.

For example:

• a single line from a study can look like the study’s conclusion;
• a cropped graph can hide the baseline or comparison period;
• a regulatory phrase can be presented without the definition that limits its meaning;
• a screenshot of a headline can be mistaken for the underlying reporting;
• a quotation can appear to endorse a proposition that the full speaker was actually qualifying.

This overlaps with cherry-picking when the removed context selectively favors one interpretation. The distinction is useful:

• Reasoning Error: evidence is selected or interpreted one-sidedly.
• Misinformation Mechanic: the artifact itself has been packaged so the omitted context is difficult for the next viewer to recover.

How to reconstruct the trail

Verification should move backward through the chain:

1. Identify the earliest version you can find.
2. Search distinctive text rather than relying only on the caption attached to the repost.
3. Replace screenshots with the original page, document, dataset, post or recording where possible.
4. Check dates, locations and version history.
5. Look for corrections, retractions, updated captions or later statements.
6. Recover the material immediately before and after the quoted fragment.
7. If the item is translated, compare the original wording and degree of certainty.
8. Record the original URL or document identifier so the next reader does not have to rebuild the chain again.

Archives can help when original pages have changed or disappeared, but an archived copy should still be interpreted in its original date and context.

What provenance reconstruction does not prove

Finding the original source does not automatically validate the substantive claim.

A primary source can be mistaken. A study can be weak. An authentic expert statement can be outside the expert’s field. An official report can later be corrected.

The correct sequence is:

identify the artifact → reconstruct provenance → recover context → then evaluate evidential quality.

This separation is important in Claim by Claim because a future structured Claim Provenance / Origin Reference model will describe where a claim originated or circulated, while the Evidence/Source system describes what supports or challenges the claim. Those relationships should not be conflated.

Preserve context before it decays

The easiest provenance trail to reconstruct is the one that was never destroyed.

When sharing evidence:

• link to the original source;
• include the date and author where material;
• preserve units, axes and captions on charts;
• distinguish a quotation from your interpretation of it;
• identify translations;
• avoid turning an entire evidentiary chain into an unattributed screenshot.

Information becomes easier to transmit when context is removed. Verification becomes easier when context travels with it.