Media is often imagined as a window.
Reality happens. Media records it. People look through the representation toward the event.
But many media systems are not passive windows.
A forecast can change investment. A poll can change campaigning. A bank-rumour can change withdrawal behaviour. A review can change demand. A viral post can change the event it describes because thousands of people react to it.
Media feedback occurs when a representation changes receiver behaviour and that changed behaviour alters the reality from which future representations are made.
This article is part of the How Media Works series. How Media Distribution Works explains how tokens reach people. How Media Metrics Work explains how audience behaviour becomes new data. How Media Personalisation Works explains how those signals can alter the next media environment. Feedback joins these stages into a loop.
1. The Simple Media Model Is One-Way
reality → representation → receiver.
This model is useful for understanding recording, framing and interpretation.
It becomes incomplete when the receiver acts on what was received.
2. The Feedback Model Adds a Return Path
reality → representation → receiver → action → changed reality → new representation.
Once the loop closes, media becomes part of the causal system rather than merely an observer of it.
3. Attention Can Change the Event
Some events grow because they are covered.
A small trend can attract attention, which attracts participants, which creates more evidence of the trend, which attracts more coverage.
The later reality is partly caused by the earlier representation.
4. Feedback Can Be Positive or Negative
Positive feedback amplifies change. Negative feedback dampens change.
- A popular post receives more recommendation and becomes more popular.
- A warning causes people to avoid a danger and thereby reduces the danger.
The word “positive” does not mean good. It means reinforcing.
The word “negative” does not mean bad. It means counteracting.
5. Self-Fulfilling Predictions Are Feedback Loops
If people believe a prediction and act in ways that make it more likely, the representation participates in producing the predicted outcome.
This does not prove every prediction has such power. It means some social systems are reflexive because expectations change behaviour.
6. Self-Defeating Predictions Are the Mirror Image
A warning can prevent the outcome it predicts.
If authorities warn of severe congestion and people change routes, the final traffic jam may be smaller than forecast.
A prediction can be useful precisely because it changes the future enough to become wrong.
7. Markets Are Highly Feedback-Sensitive
Prices reflect expectations as well as present conditions.
News, forecasts, ratings and statements can change expectations, which can change buying and selling, which can change prices.
The media token can therefore become an input into the system it reports.
8. Political Polls Can Affect Political Behaviour
Polls are representations of measured opinion.
Campaigns, donors, journalists and voters may respond to those representations.
The next poll then measures a population that has already seen the previous poll.
The measurement system is no longer fully external to the measured system.
9. Reviews Change the Market They Measure
A restaurant may receive a strong review, attract more customers, become crowded, change staff and eventually provide a different experience.
Later reviewers are sampling a business partly transformed by earlier media.
This is feedback between representation, demand and operational reality.
10. Rankings Can Change the Ranked Object
Schools, universities, products, creators and destinations may change behaviour in response to ranking criteria.
Once a metric becomes influential, participants optimise toward it.
This connects directly to How Media Metrics Work.
The measurement can become part of the management system.
11. Metrics Create Behavioural Feedback
Creators see views, watch time, shares and conversion.
They adapt headlines, topics, formats and publishing schedules. The changed content produces new metrics, which drives another adaptation.
content → audience behaviour → metrics → creator adaptation → new content.
12. Recommendation Systems Close the Loop Automatically
A ranking system can observe behaviour and update future distribution without waiting for human editors.
This creates a rapid loop:
ranking → exposure → response → signal → updated ranking.
The platform does not merely describe preferences. It helps shape the media environment in which preferences are expressed.
13. Early Advantages Can Compound
If early engagement produces more exposure, an initially small difference can widen.
Popularity can partly cause future popularity.
This does not mean quality is irrelevant. It means path dependence matters.
14. Feedback Can Create Lock-In
Once a convention, platform, format or narrative gains enough participation, switching becomes costly.
More users generate more content; more content attracts more users.
The media system can stabilise around an early pathway even when alternatives exist.
15. Social Proof Is a Feedback Signal
Visible likes, views, ratings and follower counts tell receivers what other people have already done.
Those signals can influence what new receivers choose to inspect.
The audience therefore becomes part of the representation shown to the next audience.
16. Public Reaction Becomes New Media
A speech generates commentary. Commentary generates reaction videos. Those reactions become news. The speaker then responds to the reaction.
The system moves from representing the original event to representing representations of the event.
media can become the event-space in which later events occur.
17. Feedback Can Distort Sampling
How Media Sampling Works explains how visible examples become evidence.
If media coverage changes who participates, the later sample may reflect the earlier coverage.
Researchers and journalists should therefore ask whether observation itself changed the observed population.
18. Feedback Can Distort Attribution
Suppose an article predicts demand will rise. Demand rises after the article.
Was the article accurate because it detected a real trend, or did publication help cause the trend?
Both can be true.
Causal interpretation must therefore consider reflexivity.
19. Feedback Makes Forecast Evaluation Harder
A forecast can change behaviour enough to alter the outcome.
This means a forecast should not always be judged only by whether the predicted event occurred exactly.
The relevant question may be whether the model identified a risk that prompted successful mitigation.
This links feedback to How Media Uncertainty Works.
20. Corrections Are Feedback From Reality to Media
Not all feedback comes from audience behaviour.
New evidence can return from the world and force the representation to change.
How Media Corrections Work is therefore the epistemic feedback loop of the media system.
representation → test against reality → discrepancy → repair → new representation.
21. Moderation Is Behavioural Feedback
Rules and enforcement change user behaviour.
Users adapt, moderators observe the adaptation, and policies change again.
This is the loop described in How Media Moderation Works.
22. Economic Feedback Shapes Production
Revenue tells publishers which media appears financially sustainable.
Advertising, subscriptions, sponsorships and creator payments feed back into what gets commissioned next.
How Media Economics Works explains the incentive layer behind this loop.
23. Feedback Can Reward the Wrong Proxy
If a system optimises clicks as a proxy for value, creators may produce more clickable material rather than more useful material.
The feedback loop is working mechanically while failing the human objective.
A feedback loop is only as wise as the signal it treats as success.
24. Feedback Needs Multiple Sensors
Views alone can reward clickbait. Complaints alone can over-represent dissatisfied users. Revenue alone can underweight public value.
Strong systems combine signals: attention, satisfaction, correction rates, long-term retention, trust, harm indicators and task outcomes.
25. Delays Can Destabilise Feedback
If consequences appear long after decisions, the system may over-correct before it sees the effect of earlier changes.
Fast media environments can therefore create oscillation: one pattern is rewarded, copied, saturated, rejected and replaced before long-term quality is known.
26. Feedback Can Create Herding
When people use others’ visible behaviour as evidence, independent judgment can shrink.
Everyone may appear to agree partly because everyone observes everyone else’s apparent agreement.
This prepares the ground for How Media Creates Common Knowledge.
27. Feedback Can Fragment Audiences
Personalised systems learn from individual behaviour and return more tailored media.
The loop can increase fit while reducing shared exposure.
Different groups can therefore evolve different media environments through different feedback histories.
28. Feedback Can Repair Misalignment
Feedback is not inherently distorting.
User reports can reveal inaccessible interfaces. Reader questions can reveal unclear explanations. Corrections can repair facts. Long-term learning outcomes can reveal whether educational media actually works.
The return path is how the system learns.
29. The Receiver Is Also a Sensor
Receivers do not merely consume media. They notice gaps, challenge claims, ask questions, report errors and provide evidence from contexts the publisher cannot directly observe.
A mature media architecture treats the audience as a distributed sensing layer without assuming every signal is correct.
30. Feedback Needs Provenance
Before changing a system because “users are unhappy,” ask which users, how the signal was collected and whether it is representative.
Feedback becomes evidence only when its origin and scope are visible.
This links feedback to attribution and sampling.
31. AI Accelerates Media Feedback
AI can analyse audience response, cluster complaints, predict engagement and adapt content rapidly.
It can also generate the next media token automatically.
generation → distribution → response → model update or prompt adaptation → new generation.
The faster the loop, the more important it becomes to choose the right outcome signals.
32. AI Can Create Synthetic Feedback
Bots and generated accounts can produce likes, comments, clicks and apparent consensus.
If the system treats those signals as human response, synthetic behaviour can alter future ranking and production.
Feedback quality therefore depends on receiver authenticity.
33. Feedback Needs a Stop Rule
Not every signal requires immediate adaptation.
Constantly changing a system in response to noise can make it unstable.
Strong governance distinguishes signal from variance, urgent harm from ordinary fluctuation, and durable trend from temporary reaction.
34. Wintour V1.0 Feedback Test
- What representation entered the world?
- Who received it?
- What behaviour changed because of it?
- Did that behaviour alter the underlying reality?
- What new media signals were generated?
- Are those signals human, synthetic or mixed?
- What metric is being optimised?
- Does that metric represent the real objective?
- What time delay separates action from consequence?
- What evidence would justify changing the loop?
35. The Feedback Equation
media feedback = representation effect × receiver response × behavioural transmission × world sensitivity × measurement return × adaptation rule.
This is not literal mathematics. It maps the points where a representation can become part of the reality it describes.
36. Final Thesis
Media does not stand outside civilisation.
People see representations and act. Institutions respond. Markets move. creators adapt. Platforms rerank. policies change. New evidence returns.
The media system then records the reality that its earlier representations helped shape.
The mature media reader asks not only “What does this representation describe?” but also “What will happen because people see it, and how will those reactions alter the next version of reality that media comes back to describe?”
Return to the canonical hub: How Media Works | Reality, Representation, Memory and the Human Interface.