Response: ignore, verify, report, correct, or prebunk?
What do I do about it: ignore, verify, report, rebut, or prebunk?
This chapter turns the findings of the earlier stages into a decision. It sets out a ladder of responses, explains when a rebuttal spreads the thing it rebuts, gives the structure of a correction that works, and holds firmly to attribution discipline. For organisations it covers the response plan, the holding statement and the question of who decides.
All stages
The decision ladder
Five responses are available and they form a ladder: ignore, verify, report, correct, prebunk. Verification comes before every other rung, including the decision to ignore. After that, choose the lowest rung that deals with the problem.
Ignoring is a decision, not a failure. Material that has stayed inside its starting community, in the lowest categories of the Breakout Scale, is usually best left alone and kept on file. Reporting sends the evidence to those who can act on it: the platform, the moderators of the space, the organisation being impersonated, or an established research group. Correcting is for a false or misleading claim that has reached real people, or one where silence risks concrete harm, such as wrong information about how to vote or how to stay safe. Prebunking is for what you can see coming.
When rebuttal amplifies
A correction reaches people who never saw the original, and it repeats the claim to them. Whitney Phillips describes the same bind for reporters covering manipulators: coverage is oxygen. Before correcting, ask whether the claim has already reached the audience you would be addressing. If it has not, a public rebuttal mostly spreads it.
This is a concern about reach, not about corrections failing. The fear that corrections routinely backfire and strengthen belief has not held up. Swire-Thompson, DeGutis and Lazer reviewed the evidence and found backfire effects to be rare and hard to reproduce, and the Debunking Handbook takes the same view. When the threshold is met, correct without hesitation and do it well.
How to correct
The Debunking Handbook's structure is fact, myth, fallacy, fact. Lead with the truth in a clear and memorable form. Warn that a myth is coming and state it once. Explain how it misleads, which matters because a bare denial leaves a gap that the myth goes on filling. End by restating the fact. The truth sandwich is the same idea in three layers: truth, the falsehood briefly, truth again.
Put the fact, not the myth, in the headline and the first line, since many readers go no further. Concede any fair point. Address the claim and leave out its author. A false claim is false whoever made it.
Attribution discipline
Rid and Buchanan describe attribution as a matter of degree that depends on the evidence, time and resources available, and the attribution guide on this site shows how little of that evidence is public. Content and behaviour can support a description of a pattern. They cannot support a name. So never name an account or an actor on behavioural evidence alone. If you are wrong, a real person pays for it. If you are right, you have told the operator what to change.
Stopping where the evidence stops is a complete result. Graphika published a detailed account of Secondary Infektion without identifying the body responsible. Use the write-up template: what was observed, the alternative explanations considered, a confidence level for each judgment, and an explicit list of what is not claimed. When you repeat an attribution made by a platform or a government, carry its confidence level and its caveats with it.
For organisations
Decide before the incident who decides. A response plan names one person with authority to choose a rung and a deputy for when they are away, sets thresholds for moving up the ladder, and lists contacts at platforms and, where relevant, legal advisers. The UK Government Communication Service's RESIST toolkit is a public model of this kind of plan.
Keep a holding statement ready: two or three sentences that say what is known, what is not yet known, and where the authentic information can be found, with no repetition of the claim in its opening and no guess at who is responsible. Keep dated records, look after any staff who are being targeted, and expect persistence. Doppelganger was first exposed in 2022 and was still drawing enforcement action two years later. Removal of accounts is maintenance.
Techniques you will meet here, and the tell
The defender's own error at this stage: naming a responsible party before the evidence can carry it.
The tell: A name appears in the first statement, supported by narrative alignment or posting behaviour and nothing else.
Critics or opponents are dismissed as fake so that their arguments need no answer. Defenders slide into this when they respond to who is speaking.
The tell: The reply to an argument is a claim about the account that made it, with no response to what was said.
Every account is sorted into real or fake, when the useful question is whether a set of accounts is coordinating and concealing it.
The tell: Individual accounts are being judged one by one on handles, grammar or posting volume.
A hedged or unsupported attribution hardens into fact as outlets and officials cite one another.
The tell: Each report cites another report, the confidence language has dropped out, and no one links to the original finding.
A genuine criticism or leak is answered only by pointing at a hostile source, leaving the substance untouched.
The tell: The statement says who is behind the material and never says whether the material is accurate.
A corrected claim can go on shaping what people think, especially when the correction leaves a gap in the story.
The tell: People accept that the claim was wrong and still reason as though it were true, because nothing has replaced it.
The belief that corrections make things worse is used as a reason never to correct. The evidence suggests such effects are rare.
The tell: A decision not to correct a widely believed falsehood is justified only by a general fear of backfire.
Worked example
Priya has findings from every earlier stage of the Alder Bay footbridge affair. The vote is over, a further council paper is due, and the look-alike site is still publishing. She has to decide what the Courier does, item by item.
- 1She runs each item up the ladder. The unverified voice message and the wrecked-bridge photograph reached few people: she ignores them publicly and keeps them on file. The cost claim was repeated by real residents in letters and at the meeting: correct. The cut clip spread widely and led to abuse of a named person: correct, and show the full recording. The look-alike site: report, and tell readers how to recognise the real paper.
- 2She reports. The platform receives the impersonation and the coordinated posting pattern with dated examples. The company hosting the look-alike address receives a complaint from the Courier as the organisation being imitated. She takes advice from the paper's solicitor about the masthead. She notes that outcomes are out of her hands and may be slow.
- 3She writes the correction. The headline states the fact: the footbridge will cost 1.6 million to build, and 4.1 million is the thirty-year total. The piece warns that a misleading figure is circulating, states it once, explains that a whole-life cost was labelled as a build cost, concedes that the council buried the figure in a long paper with no summary, and ends on the fact with a link to the page.
- 4She prints a short reader notice giving the Courier's exact address and how to check it. It says that a site imitating the paper exists and that the Courier does not know who runs it.
- 5She agrees a three-sentence holding statement for enquiries and settles who decides: the editor, and the deputy editor when she is away, with thresholds for when each rung applies.
- 6She completes the write-up. Observed: a site imitating the Courier, and a set of posts sharing text, timing and links. Alternatives considered: a local opponent, a click-revenue site, a prank, a political rival, an outside actor, and the part that proved to be open organising by the traders. Confidence: high that the site imitates the paper, moderate that part of the posting was coordinated, none as to who. Not claimed: that any account is fake, that any sponsor is known, or that the vote was affected.
Outcome. A month on, the platform has removed some accounts, the look-alike site is still online, and the letters page is arguing about loading bays. The reader guide on whole-life costs ran before the new council paper. Priya still does not know who was behind the campaign, and the Courier's coverage says exactly that.
Assessment write-up and response decision
This follows the write-up template in the attribution guide. Fill it in about the activity you have actually examined, then record the response you have chosen. Saved in this browser only.
Say what you looked at, over what dates, gathered how, and what you did not look at.
Describe the activity in neutral terms with counts and examples, with no adjectives about intent and no actor names.
Say separately what you know about each, and leave actors blank if public evidence gives you nothing verifiable.
List at least three, including genuine grassroots activity and an artefact of your own collection, and say what you would expect to see for each.
Give low, moderate or high separately for each judgment (that coordination exists, that it is inauthentic, that any party is responsible) and name the evidence each rests on.
Name the specific evidence that would raise or lower your confidence.
Tick each statement that is true of your write-up; all four should normally be ticked.
Tick the rungs you are using, having verified first, and prefer the lowest that deals with the problem.
Decision check: is this the right response?
0/8If any of these is not true: Step down a rung. If you have not verified, verify. If the claim has not spread, hold the correction and keep the file. If your draft names an account or an actor, remove the name and describe the pattern. If nobody has authority to decide, settle that before issuing anything, and use the holding statement in the meantime.
Common mistakes
- ×Rebutting everything. A public answer to a claim nobody has seen introduces it to a new audience.
- ×Leading with the myth. A headline that repeats the false claim, even to deny it, is what most readers will remember.
- ×Over-attribution: naming accounts or actors on behavioural evidence, or treating every critic as part of the operation. This harms real people and discredits the rest of your findings.
- ×Expecting exposure to end the matter. Documented operations have continued for years after being named; plan for maintenance.
- ×Improvising under pressure. Without a plan, a holding statement and a named decision-maker, the first response is usually the loudest one available.
What do you owe the audience now, and what do you change next time?
Related
References
- Lewandowsky, S., Cook, J., Ecker, U. K. H., et al. (2020). The Debunking Handbook 2020The fact, myth, fallacy, fact structure for corrections, and the assessment that backfire effects are uncommon.
- Rid, T., & Buchanan, B. (2015). Attributing Cyber Attacks. Journal of Strategic StudiesAttribution as a matter of degree that depends on evidence, time and resources.
- Phillips, W. (2018). The Oxygen of Amplification: Better Practices for Reporting on Extremists, Antagonists, and Manipulators Online. Data & SocietyCoverage and rebuttal can extend the reach of the material they address.
- Swire-Thompson, B., DeGutis, J., & Lazer, D. (2020). Searching for the Backfire Effect: Measurement and Design Considerations. Journal of Applied Research in Memory and CognitionBackfire effects from corrections are rare and difficult to reproduce.
- Roozenbeek, J., & van der Linden, S. (2019). Fake news game confers psychological resistance against online misinformation. Palgrave CommunicationsPrebunking as a response available before a predictable wave of misleading content.
- UK Government Communication Service (2021). RESIST 2: Counter-disinformation toolkitA public model of an organisational response plan, from early warning through strategic communication.