← Back to Algorithmic Amplification
Dynamics dynamics/algorithmic-amplification

Page history

Every change to the Algorithmic Amplification page: what changed, when it went live, and how it was reviewed.

8 open items

This implies research on algorithmic political effects only started after 2016. The page's own sources include earlier work, such as the 2006 Salganik et al. experiment and 2010-2012 coverage of EdgeRank and YouTube's ranking changes. The statement is unsourced and the timing is overstated. Reword it to say that attention increased after 2016, and cite the source.

Vague attribution ('some researchers') with no cited source. None of the listed sources show this ideological-concentration effect. The Guess et al. 2023 source found no significant change in polarization. Name the specific studies or remove the claim.

Stated as a general tendency without a citation in the passage. Brady 2017 and Rathje 2021 measure user sharing of moral-emotional and out-group content, not anger/outrage/group solidarity specifically, and the page itself says those studies did not examine ranking systems.

Unsourced causal assertion about virality mechanics and its effect on the 'pace and character of political discourse'. The claim that algorithms 'identify and rapidly distribute content showing early signs of viral spread' is also unsupported.

Unsourced claim about YouTube recommendation pathways. Research on this is mixed, and some audits found little evidence of such pathways. No cited source supports it, and the text gives no qualification. The related 'increasingly niche' wording in the radicalization connection also lacks support.

Unsourced and imprecise. No TikTok data or study is cited for the age-demographic claim. Similarly, the Instagram and coordinated Twitter trending bullets describe mechanisms with no cited evidence.

No cited source supports this. The cited TechCrunch source covers EdgeRank in 2010, not the 2006 launch. The News Feed launch (September 2006) is well known, but its original ordering is characterized loosely as 'chronological'.

The page links algorithmic amplification to seven movements, but nothing in the body or sources supports any link. The Tea Party (2009) and Occupy (2011) arose in the period the page describes as the early rollout of ranked feeds. This implies an unsupported association.

Flagged by the review on Oct 26, 2026; open for 0 days as of Oct 26, 2026.

See the findings

Lifecycle

Sep 19, 2025 – today · select a mark to jump to it
  • Writing
  • Independent review
  • Fixes
  • Minor upkeep

Changes

  1. Independent review Review 3 of 3

    Reviewed: 13 findings — 3 fixed, 2 dismissed, 8 open

    Factual accuracy 8Tone 5
    Reviewed by
    Not recorded
    Why it ran
    Not recorded
    Text changed
    +650 / −60 words
    Show changes Hide changes
    Tone

    Before “These changes altered how political information reached users on each platform.”

    After “These changes modified how content was ranked and shown to users on each platform.”

    Why: This is an unsupported causal claim stated as fact. None of the cited sources show how the individual changes altered political information reach. Attribute it or limit it to what is documented, for example 'These changes modified how content was ranked on each platform.'

    Tone

    Before “affecting the pace and character of political discourse”

    After “Political content that generates immediate strong reactions can be amplified across networks more quickly than content with lower engagement signals.”

    Why: 'Character of political discourse' is a vague evaluative claim with no measurable referent. Replace it with a documented effect, for example the measured speed of spread, or attribute it to researchers.

    See all 13 findings in the review report
    Full review report
  2. Independent review Review 2 of 3

    Reviewed: 15 findings — 11 fixed, 1 dismissed, 3 open

    Factual accuracy 9Tone 6
    Reviewed by
    Not recorded
    Why it ran
    Not recorded
    Show changes Hide changes
    Factual accuracy

    Before “Algorithmic amplification has been observed to affect democratic processes in several ways:”

    After “Researchers have examined whether and how algorithmic amplification affects democratic processes, including the following proposed effects:”

    Why: Overstated. The bullets that follow are uncited and presented as observed effects. Some are contradicted or complicated by the page's own cited research: the 2023 Science field experiment found no significant change in polarization or political knowledge, and the bullet on reduced exposure to diverse viewpoints is not supported by any cited source. Wording like 'has been hypothesized' or 'researchers have examined' would be more accurate, or each effect should be tied to a source.

    Factual accuracy

    Before “Reducing exposure to diverse political viewpoints through personalized filtering”

    After “Proposed reduction in exposure to diverse political viewpoints through personalized filtering”

    Why: Stated as an observed effect with no citation. The page's own sources and its 'debated' hedging elsewhere suggest the evidence on filter-bubble effects is mixed. The claim is unsupported as written.

    See all 15 findings in the review report
    Full review report
  3. Edit

    Edited: text revised

    +31 / −25 words
    Reason
    Not recorded
    Made by
    Not recorded
    Show changesHide changes

    Before- **YouTube Recommendation Sequences**: Recommendation algorithms that can lead viewers from mainstream political content toward increasingly niche channels with fringe or contested claims - **Facebook Engagement Prioritization**: News feed algorithms that boost high-engagement political posts, which tend to generate more comments and reactions than other content - **Coordinated Twitter Trending Activity**: How coordinated behavior can leverage trending algorithms to amplify specific political narratives or hashtags - **TikTok Political Content**: Short-form video algorithms that rapidly …

    After- **YouTube Recommendation Sequences**: Recommendation algorithms that can lead viewers from mainstream political content toward lower-viewership channels covering contested political topics - **Facebook Engagement Prioritization**: News feed algorithms that boost high-engagement political posts, which tend to generate more comments and reactions than other content - **Coordinated Twitter Trending Activity**: How coordinated behavior can leverage trending algorithms to amplify specific political narratives or hashtags - **TikTok Political Content**: Short-form video algorithms that rapidly …

    BeforeThe 2016 election cycle drew widespread attention to the potential political effects of engagement-driven algorithms, as observers noted that polarizing content often generated higher engagement than other political discourse. Subsequent elections and political events have raised further questions about how algorithmic systems interact with misinformation and high-engagement political content, though researchers continue to debate the precise causal mechanisms and their magnitude.

    AfterFollowing the 2016 election cycle, researchers and journalists began examining the potential political effects of engagement-driven algorithms, noting that content generating strong reactions often achieved higher engagement than other political discourse. Subsequent elections and political events have raised further questions about how algorithmic systems interact with misinformation and high-engagement political content, though researchers continue to debate the precise causal mechanisms and their magnitude.

    BeforeAlgorithmic amplification has been observed to affect democratic processes in several ways: - Prioritizing high-engagement content, which can favor sensationalized material over policy-focused discussion - Creating information asymmetries where fringe perspectives can achieve mainstream visibility - Increasing the potential reach of contested or unverified claims - Reducing exposure to diverse political viewpoints through personalized filtering - Enabling coordinated amplification of specific narratives by political actors - Shifting political communication strategies toward content optimized …

    AfterAlgorithmic amplification has been observed to affect democratic processes in several ways: - Prioritizing high-engagement content, which can favor content designed to generate strong reactions over substantive policy analysis - Creating conditions where lower-circulation or minority perspectives can achieve broad visibility - Increasing the potential reach of contested or unverified claims - Reducing exposure to diverse political viewpoints through personalized filtering - Enabling coordinated amplification of specific narratives by political actors - Shifting political communication …

  4. Independent review Review 1 of 3

    Reviewed: 7 findings — 5 fixed, 2 open

    Relevance 1Factual accuracy 2Tone 4

    Recorded after the fact. The findings were reconstructed later by comparing the page's text, so the reviewer's original reasoning and the model used were not kept.

    Reviewed by
    Not recorded
    Why it ran
    Not recorded
    Text changed
    +228 / −149 words
    Show changes Hide changes
    Factual accuracy

    Before “Algorithmic amplification has been observed to affect democratic processes in several ways: - Prioritizing high-engagement content, which can favor sensationalized material over policy-focused discussion - Creating information asymmetries where fringe perspectives can achieve mainstream visibility - Increasing the potential reach of contested or unverified claims - Reducing exposure to diverse political viewpoints through personalized filtering”

    Why: These effects are presented as observed facts ('has been observed to affect'), but the academic literature on several of these claims — particularly filter bubbles reducing exposure to diverse viewpoints — is actively contested. Studies such as Guess et al. (2023) and work from NYU's Center for Social Media and Politics have found limited or mixed evidence for some of these effects. Framing them as observed outcomes rather than findings under active debate overstates scientific consensus.

    Tone

    Before “Recommendation algorithms that can lead viewers from mainstream political content toward increasingly niche channels with fringe or contested claims”

    Why: "Fringe" is a loaded descriptor that implies a value judgment about the content or channels being described. A neutral alternative would describe what makes the content distinct without characterizing it pejoratively — for example, "lower-audience or ideologically niche channels making contested or unverified claims."

    See all 7 findings in the review report
    Full review report
  5. Edit

    Edited: Mechanism type added

    Mechanism type added
    Reason
    Not recorded
    Made by
    Not recorded
  6. Edit

    Edited: Platforms changed, Related movements changed

    Platforms changedRelated movements changed
    Reason
    Not recorded
    Made by
    Not recorded

  7. MinorMinor edit · Image removed

  8. Edit

    Edited: 1 section removed, text revised

    1 section removedImage added+0 / −2 words
    Reason
    Not recorded
    Made by
    Not recorded
    Show changesHide changes

    Before# Algorithmic Amplification

  9. Page created

    First version

    441 words
    Reason
    Not recorded
    Made by
    Not recorded
Spotted an error on this page?
This site is under active development. Content and features may change.