Can Platform Design Slow Myths Down?

Introduction

Platform design can slow myths down by changing the moments when people see, trust, recommend and share doubtful claims. The point is not that an interface can make everyone perfectly rational. It is that small choices — a pause before reposting, a label that adds missing context, a ranking system that does not reward outrage, or limits on repeat offenders — can change the path a false belief takes through a network. This matters because many myths and misconceptions spread less like carefully argued claims and more like convenient shortcuts: they travel when they are fast, socially rewarded, emotionally engaging and easy to pass on.

The strongest evidence supports a mixed approach. Friction can reduce impulsive sharing; labels can lower credibility or add context when they are timely and trusted; and recommendation systems can either amplify myths or limit their reach depending on what they optimise for. None of these choices is neutral. Each involves governance trade-offs between accuracy, speech, transparency, user autonomy, political trust and platform business incentives.

Why design matters more than individual good sense

A common misconception about misinformation is that false beliefs spread mainly because users are careless or gullible. Individual judgement matters, but platform architecture sets the conditions under which judgement is used. A post that is emotionally vivid, already popular and effortless to share has a structural advantage over a slower, more careful explanation. Research on misinformation belief has repeatedly found that people do not evaluate every claim as detached fact-checkers; attention, identity, familiarity, emotion and social cues all shape whether claims feel credible. [Nature]nature.comNatureThe psychological drivers of misinformation belief and its…by UKH Ecker · 2022 · Cited by 1919 — In this Review, we describe the…

That makes design choices important because they intervene before a myth has finished travelling. A correction published days later may help, but the design question is earlier: what happens at the moment of exposure, ranking, recommendation, reposting or monetisation? A platform can make the quickest action “share now”, or it can insert a question, show provenance, reduce algorithmic spread, route the user to authoritative context, or avoid rewarding the same behaviour with visibility.

This is why platform governance increasingly treats misinformation as a systems problem rather than only a content problem. The EU’s Digital Services Act, for example, focuses not just on individual illegal posts but on platform duties around systemic risks, recommender transparency, advertising transparency and accountability for very large online platforms. [Digital Strategy]digital-strategy.ec.europa.euDigital StrategyThe Digital Services Act | Shaping Europe's digital futureThe Digital Services Act helps to make the online environment s…

Friction before sharing: small delays can change what travels

Friction means adding a small obstacle before an action: a prompt, a confirmation click, a read-before-sharing nudge, or a request to think about accuracy. In consumer technology, friction was long treated as something to remove. For misinformation control, the same ease can become a problem, because myths benefit from speed.

One of the best-known examples is Twitter’s “read before you retweet” prompt, tested in 2020 for users trying to repost an article they had not opened. Twitter reported that users opened articles 40% more often after seeing the prompt, and opened articles before retweeting 33% more often. The intervention did not ban speech or decide whether an article was true; it slowed a specific behaviour at a high-risk moment. [TechCrunch]techcrunch.comTech Crunch Twitter plans to bring prompts to 'read before you retweetTech Crunch Twitter plans to bring prompts to 'read before you retweet

Accuracy prompts work on a related principle. Instead of telling users what to believe, they redirect attention towards whether a claim is accurate. A Nature study by Gordon Pennycook and colleagues found that subtly shifting attention to accuracy increased the quality of news people subsequently shared, including in a field experiment on Twitter. The mechanism is important: many users do care about accuracy, but social media often pulls attention towards humour, identity, outrage or social approval instead. [Nature]nature.comOpen source on nature.com.

Friction is not automatically good. It can annoy users, burden legitimate speech, or be interpreted as platform manipulation if applied unevenly. A useful design distinction is between targeted protective friction and blanket obstruction. A prompt before resharing a viral claim about an election, public health emergency or crisis is easier to justify than slowing every ordinary post. Recent research reviews describe friction as promising when it is overt, protective and linked to information quality, but still dependent on careful testing and context. [PMC]pmc.ncbi.nlm.nih.govSource details in endnotes.

The practical lesson is that good friction is not a punishment. It is a speed bump placed where myths gain momentum: before forwarding, reposting, joining a viral pile-on, or sharing material the user has not read.

Labels and context: warnings help, but trust decides how far they go

Labels are among the most visible platform interventions. They can say a post is false, missing context, disputed, AI-generated, state-affiliated, manipulated, or linked to a fact-check. Their appeal is obvious: they preserve access to the content while adding a corrective signal. They are also politically and psychologically delicate because the label itself becomes part of the message.

Evidence broadly suggests that warning labels can reduce perceived credibility and sharing intentions, especially when labels are clear, specific and close to the misleading content. A 2024 randomised controlled study using a mock social media environment tested misinformation warning labels and found that such soft moderation can inform users about post accuracy and reduce willingness to share. [TMU Research Repository]rshare.library.torontomu.caSource details in endnotes.

However, labels are not magic stickers. They can fail when they are vague, late, distrusted or interpreted as an attack on a community. A study reported in Harvard Kennedy School’s Misinformation Review and covered by The Guardian found that “disputed” labels on Donald Trump’s false election tweets did not persuade his supporters and, among some Trump voters, could reinforce belief in the false claims. The authors noted that the timing and distrust of Twitter during the 2020 election context may have mattered. [The Guardian]theguardian.comSource details in endnotes.

This points to a central design choice: labels should add usable context, not merely signal institutional disapproval. Better labels usually answer questions a reader actually has: What is wrong? Who checked it? What evidence is missing? Is the issue false content, manipulated media, satire, old footage, or a misleading framing? Partnership on AI’s principles for labelling manipulated media emphasise that labels should be understandable, proportionate, tested with users and adapted to different media contexts. [Partnership on AI]partnershiponai.orgPartnership on AIIt matters how platforms label manipulated media. Here arePartnership on AIIt matters how platforms label manipulated media. Here are

Context labels can also work when they disclose source relationships rather than adjudicating every claim. Research on state-media labels found that warnings about state control can mitigate the influence of election misinformation from outlets such as RT. That kind of label does not say “everything here is false”; it gives readers relevant provenance information they might not otherwise have. [Misinformation Review]misinforeview.hks.harvard.eduSource details in endnotes.

The hardest label problem is coverage. A label shown after a myth has already reached millions of people may reduce further spread but cannot undo all exposure. A label system that catches only the most viral falsehoods may still be valuable, but it should not be mistaken for a complete misinformation strategy.

Community notes: useful context, uneven coverage

Community-based fact-checking systems, such as X’s Community Notes, try to solve a trust problem by replacing top-down platform judgement with notes written and rated by users. X describes Community Notes as a system that lets contributors collaboratively add helpful context to posts that might be misleading. Notes are displayed only after ratings from contributors with differing viewpoints meet the system’s helpfulness threshold. [X (formerly Twitter]nature.comSource details in endnotes.

The strongest case for community notes is that peer-supplied context may feel less like institutional censorship and more like a public correction. Studies have found that community-based fact-checking can increase trust in fact-checking compared with simple misinformation flags, and recent large-scale research on X found evidence that displayed notes reduced the spread of misleading posts. [PMC]pmc.ncbi.nlm.nih.govSource details in endnotes.

The weakness is speed and coverage. Community notes often require enough contributors, ratings and cross-viewpoint agreement before a note appears. In fast-moving myths — a crisis, election rumour, war image, public health scare or celebrity hoax — the false claim may travel before consensus catches up. Reporting on a Center for Countering Digital Hate analysis of US election misinformation found that many misleading posts in its sample lacked visible Community Notes, and that original misleading posts had far more views than their corrections. [Reuters]reuters.comMusk's X ineffective against surge of US election misinformation, report saysMusk's X ineffective against surge of US election misinformation, report says

This does not mean community notes are useless. It means they are better understood as one layer in a design stack, not a replacement for all professional fact-checking, ranking changes, crisis response and enforcement. They can be particularly useful for adding context to ambiguous or misleading claims, but they struggle when myths move faster than the note approval process.

Recommendation incentives: the hidden engine of myth spread

The most important design choice may be the one users rarely see: what the recommendation system rewards. If a platform ranks content mainly by predicted engagement — clicks, comments, shares, watch time or reactions — it may give a boost to claims that provoke anger, fear, surprise or identity defence. Myths and misconceptions are often designed, accidentally or deliberately, to be engaging.

A UK parliamentary report on social media, misinformation and harmful algorithms warned that engagement-maximising design can amplify false or harmful content regardless of accuracy, because harmful and false material is often built to attract attention. [UK Parliament]publications.parliament.ukSource details in endnotes. Academic work on engagement-based ranking has also described a trade-off: giving more weight to social interactions such as likes and shares can increase engagement while also increasing misinformation and polarisation. [ifo Institut]ifo.deSource details in endnotes.

Platforms can respond by changing what counts as success. YouTube’s “Four Rs” approach — remove violative content, raise authoritative sources, reduce borderline content and reward trusted creators — is an example of recommendation governance rather than simple takedown. In 2019, YouTube said it had made more than 30 changes to reduce recommendations of borderline content and harmful misinformation, reporting a 70% average drop in US watch time for such content from non-subscribed recommendations. [blog.youtube]blog.youtubeThe Four Rs of Responsibility, Part 2: Raising authoritativeThe Four Rs of Responsibility, Part 2: Raising authoritative

Facebook has used related tools, including demotion for content rated false by fact-checkers and reduced distribution for repeat offenders. Harvard Kennedy School’s Misinformation Review examined Facebook’s downranking interventions against groups and websites that repeatedly shared misinformation and found a significant reduction in engagement per post or article after repeated false links. [Misinformation Review]misinforeview.hks.harvard.eduSource details in endnotes. Meta’s own transparency materials describe fact-checking, labelling and reduced distribution as part of its approach outside the United States, while its 2025 US policy shift away from third-party fact-checking shows how contested these systems remain. [Transparency Center]transparency.meta.comTransparency Center Fact-Checked MisinformationTransparency Center Fact-Checked Misinformation

The key governance issue is incentives. If a myth earns watch time, comments, creator revenue or follower growth, design has to decide whether those signals should be treated as popularity, risk, or both. Recommendation reform is therefore less visible than a warning label but often more consequential, because it determines whether a misconception remains a fringe claim or becomes a recurring suggestion.

Sharing limits and repeat-offender rules: reducing reach without deleting everything

Not every intervention needs to remove content. Platforms can reduce forwarding limits, restrict reshares, require group admin approval, demonetise misinformation, lower the ranking of repeat offenders, or prevent known false content from being recommended. These measures are especially relevant when a myth is not merely a single post but a repeated tactic.

During the COVID-19 pandemic, Meta said it removed some false claims likely to contribute to imminent physical harm, while other claims could be fact-checked, labelled or demoted. It also said groups, pages and accounts that repeatedly shared debunked claims could face removal, and some group admins could be required temporarily to approve posts before publication. [About Facebook]about.fb.comAbout Facebook More Speech and Fewer MistakesAbout Facebook More Speech and Fewer Mistakes

This illustrates a useful ladder of intervention. A platform can start with context, then reduce distribution, then restrict repeat offenders, and reserve removal for the highest-risk cases. That ladder matters because myths vary in harm. A harmless historical misconception, a misleading diet claim, a false voting instruction and a dangerous medical falsehood should not all trigger the same response.

Repeat-offender rules also address a common loophole: misinformation actors may rely on volume. Even if each individual post is borderline, a page, channel or group that repeatedly pushes false claims can shape the information environment. Design choices that target repeated behaviour are therefore more structural than one-off fact checks.

The trade-offs: slowing myths without building a black box

Every anti-misinformation design choice creates a governance question. Who decides what counts as misleading? How are errors corrected? Can users appeal? Are labels applied consistently across languages and countries? Do researchers have access to enough data to test whether interventions work? Can platforms explain why a claim was demoted without revealing systems that bad actors can game?

Transparency is not a decorative extra. The Digital Services Act requires very large online platforms and search engines in the EU to assess systemic risks and gives users more information about recommender systems, including at least one option not based on profiling. [Digital Strategy]digital-strategy.ec.europa.euDigital StrategyThe Digital Services Act | Shaping Europe's digital futureThe Digital Services Act helps to make the online environment s… These obligations reflect a broader shift: platform design choices are now public-interest decisions, not merely internal product tweaks.

There is also a legitimacy problem. Labels or demotions may be accurate but still fail if users believe the platform is politically biased or commercially self-interested. Meta’s 2025 decision to replace US third-party fact-checking with a Community Notes-style model was framed by the company as a response to over-enforcement and perceived censorship, while critics warned that it could weaken misinformation controls. [About Facebook]about.fb.comAbout Facebook More Speech and Fewer MistakesAbout Facebook More Speech and Fewer Mistakes The dispute shows that the design of misinformation controls cannot be separated from public trust in the institution applying them.

A well-designed system therefore needs more than clever prompts. It needs published rules, meaningful appeals, independent research access, careful measurement, language coverage, crisis procedures and regular evidence review. Without those, interventions can become either performative safety theatre or opaque moderation that users cannot understand.

What good platform design looks like in practice

The best design strategy is layered. A single label, prompt or ranking change will not stop myths by itself, but a set of mutually reinforcing choices can reduce spread without treating every false belief as a deletion problem.

A practical design cluster would include:

The strongest misconception to avoid is that platform design must choose between doing nothing and deleting everything. Most of the important interventions sit between those poles. They change speed, visibility, context, incentives and repetition. That is where many myths either gain social life or lose momentum.

Amazon book picks

Further Reading

Books and field guides related to Can Platform Design Slow Myths Down?. Use these as the next step if you want deeper reading beyond the article.

Book

The Chaos Machine

By Max Fisher

Directly addresses platform incentives, recommendation systems and viral falsehoods.

See on Amazon

Book

Stolen Focus

By Johann Hari

Connects interface design and attention capture to poor information judgement.

See on Amazon

Book

The Hype Machine

By Sinan Aral

Covers social media design, misinformation, virality and platform interventions.

See on Amazon

Book

Hooked

By Nir Eyal

Explains habit-forming design patterns that can also accelerate myth sharing.

See on Amazon

Browse more on Amazon: The Chaos Machine Stolen Focus The Hype Machine

As an Amazon Associate I earn from qualifying purchases.

eBay marketplace picks

Marketplace Samples

Example marketplace items related to this page. Use the search link to explore similar finds on eBay.

Shop location

Using USA USA USAUKCanadaAustraliaIreland

Example eBay listing

RETRO TECHNOLOGY Framed Wall Art Poster Canvas Print Picture

Search eBay.co.uk: technology wall art

Browse similar on eBay.co.uk

Example eBay listing

Technology girl Framed Art Print Framed Wall Art Poster Canvas Print Picture

Search eBay.co.uk: technology wall art

Browse similar on eBay.co.uk

Example eBay listing

yellow technology tree Framed Art P Framed Wall Art Poster Canvas Print Picture

Search eBay.co.uk: technology wall art

Browse similar on eBay.co.uk

Example eBay listing

Technology Framed Art Print Framed Wall Art Poster Canvas Print Picture

Search eBay.co.uk: technology wall art

Browse similar on eBay.co.uk

Browse more on eBay.co.uk

Example items shown for inspiration; availability and pricing can change. Branchoria may earn a commission if you purchase through outbound eBay links.

Endnotes

1. Source: nature.com
   Link: https://www.nature.com/articles/s44159-021-00006-y
   Source snippet

   NatureThe psychological drivers of misinformation belief and its...by UKH Ecker · 2022 · Cited by 1919 — In this Review, we describe the...

2. Source: techcrunch.com
   Title: Tech Crunch Twitter plans to bring prompts to ‘read before you retweet
   Link: https://techcrunch.com/2020/09/24/twitter-read-before-retweet/

3. Source: nature.com
   Link: https://www.nature.com/articles/s41586-021-03344-2

4. Source: partnershiponai.org
   Title: Partnership on AIIt matters how platforms label manipulated media. Here are
   Link: https://partnershiponai.org/it-matters-how-platforms-label-manipulated-media-here-are-12-principles-designers-should-follow/

5. Source: misinforeview.hks.harvard.edu
   Link: https://misinforeview.hks.harvard.edu/article/state-media-warning-labels-can-counteract-the-effects-of-foreign-misinformation/

6. Source: pmc.ncbi.nlm.nih.gov
   Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC11212665/

7. Source: nature.com
   Link: https://www.nature.com/articles/s41467-026-72597-0

8. Source: reuters.com
   Title: Musk’s X ineffective against surge of US election misinformation, report says
   Link: https://www.reuters.com/world/us/musks-x-ineffective-against-surge-us-election-misinformation-report-says-2024-10-31/

9. Source: publications.parliament.uk
   Link: https://publications.parliament.uk/pa/cm5901/cmselect/cmsctech/441/report.html

10. Source: ifo.de
    Link: https://www.ifo.de/en/cesifo/publications/2026/working-paper/ranking-engagement-how-social-media-algorithms-fuel-misinformation

11. Source: blog.youtube
    Title: The Four Rs of Responsibility, Part 2: Raising authoritative
    Link: https://blog.youtube/inside-youtube/the-four-rs-of-responsibility-raise-and-reduce/

12. Source: misinforeview.hks.harvard.edu
    Title: Misinformation Review Measuring the effect of Facebook’s downranking
    Link: https://misinforeview.hks.harvard.edu/article/measuring-the-effect-of-facebooks-downranking-interventions-against-groups-and-websites-that-repeatedly-share-misinformation/

13. Source: transparency.meta.com
    Title: Transparency Center Fact-Checked Misinformation
    Link: https://transparency.meta.com/features/approach-to-ranking/content-distribution-guidelines/misinformation/

14. Source: transparency.meta.com
    Link: https://transparency.meta.com/en-gb/policies/community-standards/misinformation/

15. Source: transparency.meta.com
    Title: how fact checking works
    Link: https://transparency.meta.com/en-gb/features/how-fact-checking-works/

16. Source: committees.parliament.uk
    Link: https://committees.parliament.uk/writtenevidence/449/pdf/

17. Source: committees.parliament.uk
    Link: https://committees.parliament.uk/writtenevidence/449/html/

18. Source: committees.parliament.uk
    Link: https://committees.parliament.uk/writtenevidence/138329/html/

19. Source: researchbriefings.files.parliament.uk
    Title: POST PN 0719
    Link: https://researchbriefings.files.parliament.uk/documents/POST-PN-0719/POST-PN-0719.pdf

20. Source: facebook.com
    Link: https://www.facebook.com/businessinsider/posts/according-to-a-report-viewed-by-the-verge-facebook-identified-a-massive-ranking-/10159357436404071/

21. Source: facebook.com
    Link: https://www.facebook.com/FastCompany/posts/a-new-study-found-the-platforms-recommendation-algorithm-steers-users-toward-sha/1359243662736759/

22. Source: facebook.com
    Link: https://www.facebook.com/business/help/315131736305613

23. Source: facebook.com
    Link: https://www.facebook.com/cnn/posts/youtube-says-it-will-start-reducing-its-recommendations-of-borderline-content-an/10159153613136509/

24. Source: facebook.com
    Link: https://www.facebook.com/business/help/297022994952764

25. Source: youtube.com
    Title: Stanford Seminar
    Link: https://www.youtube.com/watch?v=cjEQikrwbBg

26. Source: youtube.com
    Link: https://www.youtube.com/howyoutubeworks/our-policies/

27. Source: youtube.com
    Title: Designing for Trust: How Platforms Shape Online Belief
    Link: https://www.youtube.com/watch?v=0hY7lP5wW6w
    Source snippet

    The Role of Friction in Stopping Misinformation...

28. Source: youtube.com
    Title: The Role of Friction in Stopping Misinformation
    Link: https://www.youtube.com/watch?v=QZ0s3_s9yXw
    Source snippet

    How Algorithmic Ranking Influences Viral Myths...

29. Source: youtube.com
    Title: How Algorithmic Ranking Influences Viral Myths
    Link: https://www.youtube.com/watch?v=A-n826H4H7s
    Source snippet

    Beyond Content Moderation: Platform Architecture and Misinformation...

30. Source: youtube.com
    Title: Beyond Content Moderation: Platform Architecture and Misinformation
    Link: https://www.youtube.com/watch?v=5Ue819b78kU
    Source snippet

    Designing Social Media to Reduce Emotional Contagion...

31. Source: youtube.com
    Title: Designing Social Media to Reduce Emotional Contagion
    Link: https://www.youtube.com/watch?v=v_WXMEWiZi8

32. Source: digital-strategy.ec.europa.eu
    Link: https://digital-strategy.ec.europa.eu/en/policies/digital-services-act
    Source snippet

    Digital StrategyThe Digital Services Act | Shaping Europe's digital futureThe Digital Services Act helps to make the online environment s...

33. Source: rshare.library.torontomu.ca
    Link: https://rshare.library.torontomu.ca/articles/conference_contribution/To_Share_or_Not_to_Share_Randomized_Controlled_Study_of_Misinformation_Warning_Labels_on_Social_Media/29950115

34. Source: theguardian.com
    Link: https://www.theguardian.com/technology/2024/sep/20/trump-tweets-false-label-credibility-supporters

35. Source: communitynotes.x.com
    Link: https://communitynotes.x.com/guide/en/about/introduction

36. Source: about.fb.com
    Title: About Facebook More Speech and Fewer Mistakes
    Link: https://about.fb.com/news/2025/01/meta-more-speech-fewer-mistakes/

37. Source: about.fb.com
    Title: About Facebook An Update on Our Work to Keep People Informed and Limit
    Link: https://about.fb.com/news/2020/04/covid-19-misinfo-update/

38. Source: theguardian.com
    Link: https://www.theguardian.com/technology/2025/jan/07/meta-facebook-instagram-threads-mark-zuckerberg-remove-fact-checkers-recommend-political-content

39. Source: theguardian.com
    Title: twitter aims to limit people sharing articles they have not read
    Link: https://www.theguardian.com/technology/2020/jun/11/twitter-aims-to-limit-people-sharing-articles-they-have-not-read

40. Source: support.google.com
    Link: https://support.google.com/youtube/answer/13813322?hl=en

41. Source: Wikipedia
    Title: Community Notes
    Link: https://en.wikipedia.org/wiki/Community_Notes

Additional References

1. Source: algorithmwatch.org
   Link: https://algorithmwatch.org/en/dsa-explained/
   Source snippet

   AlgorithmWatchA guide to the Digital Services Act, the EU's law to rein in...More transparency on recommender systems and online adverti...

2. Source: researchgate.net
   Link: https://www.researchgate.net/publication/374859677_Misinformation_warning_labels_are_widely_effective_A_review_of_warning_effects_and_their_moderating_features

3. Source: apnews.com
   Link: https://apnews.com/article/0fa4fec0f703369b93be248461e8005d

4. Source: prosocialdesign.org
   Link: https://www.prosocialdesign.org/library/accuracy-prompts

5. Source: eu-digital-services-act.com
   Link: https://www.eu-digital-services-act.com/

6. Source: cepr.org
   Link: https://cepr.org/voxeu/columns/fact-checking-reduces-circulation-misinformation-we-should-not-get-rid-it

7. Source: brookings.edu
   Link: https://www.brookings.edu/articles/how-misinformation-spreads-on-social-media-and-what-to-do-about-it/

8. Source: lawsocietywa.asn.au
   Link: https://lawsocietywa.asn.au/wp-content/uploads/2025/04/Annexure-J-New-Article-Metas-Approach-to-Disinformation-Phoebe-Galbally.pdf

9. Source: 5rightsfoundation.com
   Link: https://5rightsfoundation.com/wp-content/uploads/2024/10/MisinformationBriefingRiskybyDesign.pdf

10. Source: bipartisanpolicy.org
    Link: https://bipartisanpolicy.org/wp-content/uploads/2023/10/BPC_Tech-Algorithm-Tradeoffs_R01.pdf