Hubbry Logo
Feed (Facebook)Feed (Facebook)Main
Open search
Feed (Facebook)
Community hub
Feed (Facebook)
logo
7 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
Contribute something
Feed (Facebook)
Feed (Facebook)
Feed (Facebook)
View on Wikipedia
from Wikipedia

Facebook's Feed for mobile devices
This article is part of a series about
Meta Platforms
Products and services
Facebook
Other products
People
Executives and board members
Notable employees
Related organizations
Business
Criticism
Litigation
Related

Facebook's Feed, formerly known as the News Feed, is a web feed feature for the social network. The feed is the primary system through which users are exposed to content posted on the network. Feed highlights information that includes profile changes, upcoming events, and birthdays, among other updates. Using a proprietary method, Facebook selects a handful of updates to show users every time they visit their feed, out of an average of 2,000 updates they can potentially receive. Over two billion people use Facebook every month, making the network's Feed the most viewed and most influential aspect of the news industry.[1] The feature, introduced in 2006, was renamed "Feed" in 2022.

History

[edit]

Before 2006, Facebook simply consisted of profiles, requiring the user to visit a profile to see any new posts.[1] On September 6, 2006, Facebook announced a new home page feature called "News Feed". The new layout created an alternative home page in which users saw a constantly updated list of their friends' Facebook activities.[2][3] Initially, the addition of the News Feed caused discontent among Facebook users, many of which complained that the feed was too intrusive, detailing every moment with timestamps,[4] and violating their privacy.[5] Some called for a boycott of the company.[6] In response to this dissatisfaction, CEO Mark Zuckerberg issued a statement clarifying that "We didn't take away any privacy options."[6] Following this, Zuckerberg later issued an open letter apologizing for a lack of information on new features and users' controls, writing that "We really messed this one up. [...] I'd like to try to correct those errors now."[7]

The News Feed has received multiple updates over the years since its original setup. In 2008, Facebook added a feedback button to each story in a user's feed, letting them tell the service about their personal preferences for their feed. However, the feedback button was removed in April,[8] and returned in July, with Facebook reportedly removing the first iteration of the feedback options due to a low impact on user satisfaction compared to other aspects of the algorithm.[9]

In March 2009, Facebook rolled out the option to "Like" a page to see updates from it in their feed, gave users customizable filters to determine what friends they wanted to see News Feed updates from,[10] and also added a publishing field at the top of the feed, previously exclusive to user profiles, for easy post creation.[11] The publishing field contained the text "What's on your mind?", a similar but also notably different question from Twitter's "What are you doing right now?"[11] A few weeks later, the company introduced controls to reduce content from app interactions, and enabled the feed to show photos in which friends were tagged.[10]

In December 2010, Facebook rolled out a new drop-down button, offering users the ability to view News Feed by categories, including only games, status updates, photos, links, Pages, or specific groups of people.[12]

In February 2011, Facebook added News Feed settings to let users specify if they want content from only the people and pages they interact with the most, rather than everyone.[13] In September, Facebook updated the feed to show top stories and most recent stories, rather than relying on a strictly chronological order.[10] Later the same year, it introduced the "ticker", a real-time extension of News Feed, located on the right side of the screen.[10][14] At the end of the year, news outlets reported that Facebook would be starting allowing advertisements through "Sponsored Stories" in News Feed for the first time.[15][16] Advertisements started rolling out on January 10, 2012, with a "Featured" tag declaring its paid status.[17][18] Advertisements were expanded to mobile in February 2012.[19][20]

In March 2013, Facebook held a press event to unveil new updates to News Feed, including a more minimalistic design with consistency across both the website and mobile devices. This included a new layout for posts, presenting friends' photos, shared articles, and maps with larger text and images, and brands' logos. New "sub-feeds" show updates in specific areas, such as posts from specific friends or interest updates.[21][10][22] However, the initial limited rollout of the new design saw a trend of lower user engagement, prompting the company to stop the rollout.[23] A year later, in March 2014, Facebook once again updated its News Feed, but in response to criticism from users, the company chose to scale back its efforts. While bringing bigger photos that span the width of the feed, font changes, and design tweaks to buttons and icons, the new design removed the drop-down menu, placing relevant entries in a navigation on the left side of the screen while removing some of the sub-feeds. It also simplified the comments system, altered the appearance of profile photos in the feed, and added a search bar at the top of the page.[24][25] News Feed product manager Greg Marra explained that "People don't like us moving their furniture around, because you break muscle memory".[24] Marra also stated that "Over the last year, we've spent a lot of time seeing what people were saying, what was working, what wasn't working, and we're rolling out the version that takes all of that feedback into account".[26]

In January 2018, following a difficult 2017, marked by accusations of relaying fake news and revelations about groups close to Russia which tried to influence the 2016 US presidential election (see Russian interference in the 2016 United States elections) via advertisements on his service, Mark Zuckerberg announced in his traditional January post:

We're making a major change to how we build Facebook. I'm changing the goal I give our product teams from focusing on helping you find relevant content to helping you have more meaningful social interactions

— Mark Zuckerberg

Following surveys of Facebook users,[27] this desire for change will take the form of a reconfiguration of the News Feed algorithms in order to:

  • Prioritize content of family members and friends (Mark Zuckerberg January 12, Facebook:[28] "The first changes you'll see will be in News Feed, where you can expect to see more from your friends, family and groups".)
  • Give priority to news articles from local sources considered more credible

These changes are expected to improve "the amount of meaningful content viewed".[29] However, a 2022 study shows that when news content is removed from the Feed, "many users will find almost nothing of value".[30]

In 2022, Facebook's parent company, Meta Platforms, announced it is renaming the "News Feed" to simply be named "Feed".[31]

Influence

[edit]

Approximately two billion people use the Facebook platform every month.[32] Approximately 62 percent of adults in the United States use social media to get news, meaning Facebook's influence has become a liability for the company.[33] During the 2016 U.S. presidential election, the Russian government used the Facebook platform to disseminate fake news that more frequently favored Donald Trump over Hillary Clinton.[33] As a social media platform, user-generated content and media created content can be shared vastly within the digital community. This has come with repercussions for Facebook, as they were accused of releasing personally identifiable information of approximately 82 million users to Cambridge Analytica.[34] The Cambridge Analytica Scandal drew much attention to the privacy settings and influence of the Feed on the Facebook platform. The Feed has become a significant contributor to the spread of misinformation; as former U.S. president Barack Obama put it, "misinformation...looks the same when you see it on a Facebook page or you turn on your television."[32]

After the 2016 election, journalist Margaret Sullivan called on Facebook inc. to hire an editor to monitor the News Feed to ensure accuracy and balance of news stories.[32] In late 2016, Facebook described plans to issue warning labels on certain News Feed posts. Facebook has a partnership with fact-checkers like Snopes.com and PolitiFact, and would display that a story is disputed if it has been debunked by one of those fact-checkers.[32]

Operation

[edit]

On the Facebook app, Feed is the first screen to appear, partially leading most users to think of the feed as Facebook itself.[32]

The Facebook Feed operates as a revolving door of articles, pages the user has liked, status updates, app activity, likes from other users photos and videos.[35] This operates an arena of social discussion. Algorithms are employed on the Facebook platform to curate a personalized experience for users that is predominantly featured in the Feed.[36]

Adam Mosseri is Facebook's vice president in charge of Feed and Chief Product Officer while Chris Cox runs the Facebook app and Feed.[32] On October 1, 2018, it was announced Adam Mosseri would become the head of Instagram.[37][38]

Algorithms

[edit]

Facebook's proprietary recommendation algorithms compare the merits of about 2,000 potential posts every time the app is opened, using a complex system based on providing a meaningful experience, over that of clicks, reactions, or reading time.[32] The Feed has been described as a filter bubble, showing users personalized results about information deemed interesting to them, in contrary to showing all information, even information that they disagree with.[32] Subsequently, the functionality of the Feed has been debated as to whether or not it is an echo-chamber.[32]

Facebook has been researching this situation since 2010,[32] and initially used an algorithm known as EdgeRank.[39] By late 2013, clickbait articles had become significantly prevalent, leading Facebook's chief product officer Chris Cox's team to hire survey panels to assess how Feed was working. As a result, Facebook began adding ever-increasing numbers of data points to its algorithm to significantly reduce clickbait.[32]

Effect on opinion

[edit]

A 2015 study published in Science concluded that Facebook's algorithms had a minimal effect on the news feed's diversity, though the study prompted academic criticism.[32]

Researchers at the MIT Media Lab Center for Civic Media produced an application called Gobo which allows users to see the results of adjusting the algorithm.[40][41]

See also

[edit]

References

[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia

Feed (Facebook)

View on Grokipedia
from Grokipedia
The Feed on , formerly known as the News Feed, is the platform's central algorithmic interface that delivers a personalized of content to users, including updates from friends, followed pages, groups, events, and advertisements, prioritized by predicted relevance and engagement potential. Launched on September 5, 2006, it aggregated and displayed real-time activities across profiles in reverse chronological order initially, fundamentally shifting from static profile views to a dynamic, centralized hub for social interaction. The introduction of the Feed dramatically increased user by surfacing relevant stories, with Facebook's stated objective to show content that fosters meaningful connections and prevents users from missing important updates, though it immediately provoked widespread backlash for exposing private activities without explicit , leading to over 700,000 users joining protest groups within days. Over time, the Feed evolved from chronological sorting to machine learning-driven algorithms emphasizing interactions like likes, comments, and shares to maximize time spent on the platform, which indicates boosts overall engagement metrics but can amplify polarizing or sensational content due to its reliance on user signals. Key defining characteristics include its adaptability to user feedback loops, where higher reinforces content visibility, contributing to Facebook's (now ') revenue growth through targeted ads integrated seamlessly into the stream; however, empirical studies highlight correlations with widened ideological divides and shifts toward lower-quality news consumption during periods of algorithmic tweaks prioritizing . Despite internal efforts to balance with reducing —such as demoting untrustworthy sources—the Feed's core mechanism remains centered on causal drivers of user retention, underscoring its role in shaping information ecosystems while facing ongoing scrutiny for unintended societal impacts like echo chambers, though some claims of have been challenged by data showing self-reinforcing user preferences as primary factors.

Historical Development

Inception and Initial Design

The Facebook News Feed was introduced on September 5, 2006, as a central homepage feature designed to aggregate and display real-time updates from users' friends and subscribed groups. Prior to its launch, users primarily accessed updates by manually visiting individual profiles, which limited visibility and engagement; the Feed aimed to address this by creating a consolidated, dynamic stream that surfaced activities such as relationship status changes, wall posts, photo uploads, and event RSVPs. Engineered by Ruchi Sanghvi and her team, the initial implementation pulled data from users' networks to generate short, automated "stories" in reverse-chronological order, emphasizing recency over algorithmic personalization at launch. The design prioritized ease of use and network effects, reasoning that centralized visibility would increase time spent on the platform by reducing the of navigation. Stories were generated from or friend-visible actions, without user for , which aggregated disparate updates into a single feed to foster passive consumption and social awareness. No advertising was integrated initially, focusing instead on organic content to drive retention; the Feed's structure supported scalability as expanded beyond college networks earlier that year. Launch provoked immediate user backlash, with over 700,000 joining protest groups within 24 hours, citing privacy invasions from the automatic surfacing of personal updates previously confined to profiles. In response, rolled out enhanced controls on September 8, , allowing users to hide specific stories or adjust visibility settings for News Feed and Mini-Feed (a profile-side summary). This iteration underscored early tensions between platform utility and user autonomy, shaping subsequent refinements while establishing the Feed as the core mechanism for content discovery.

Key Algorithmic and Feature Evolutions (2006-2018)

The News Feed launched on September 5, 2006, presenting users with a reverse-chronological stream of aggregated activities from their friends, such as status updates, posts, and photo uploads, replacing the previous siloed profile views. Initially devoid of sophisticated beyond recency, the feature aimed to centralize social updates but triggered widespread protests, leading to the rapid addition of customizable visibility controls by September 8, 2006. By 2009, the Feed incorporated the "Like" button, activated on February 9, which introduced a lightweight engagement metric to signal user interest and influence future content prioritization. That year also saw a shift to real-time streaming, eliminating the prior 30-minute delay for stories to appear, enabling instantaneous updates akin to emerging platforms like Twitter. In 2010, Facebook unveiled dual Feed modes—"Top News" for algorithmically curated highlights and "Most Recent" for chronological order—marking the formalization of EdgeRank, an early scoring system that weighted stories by user-poster affinity (relationship strength), edge weight (interaction type, with comments valued over likes), and time decay (favoring fresh content). The December 2011 rollout of Timeline restructured user profiles as comprehensive chronological histories, integrating richer content types like life events and milestones into the Feed, which began surfacing these as dynamic updates to enhance narrative continuity. evolved into more opaque models by 2013, with an August update boosting high-engagement posts from Pages that users had previously overlooked, based on predicted click-through and response rates derived from billions of interactions. In 2014, refinements prioritized "quality" signals—such as sustained comments and shares—over sheer volume, demoting spammy or low-interaction content through classifiers trained on user feedback. Subsequent iterations from 2015 to 2018 increasingly leveraged for personalization, incorporating over 1,000 signals per post, including demographic predictions and latent factors from user behavior graphs. A pivotal 2016 redesign emphasized "meaningful interactions" by elevating posts from close friends and family—those eliciting emotional responses like extended comments—over viral publisher content, reducing the latter's reach by up to 20% in tests to combat . By January 2018, under CEO Mark Zuckerberg's directive, the algorithm further deprioritized passive consumption of news and videos, favoring content sparking reciprocal conversations among strong ties, as measured by reply chains and group engagements, in response to surveys indicating user fatigue with non-social feeds. This period's evolutions transformed the Feed from a simple aggregator into a predictive , processing 100,000+ ranking variants per user session to maximize time spent while adapting to shifting patterns.

Modern Updates and Shifts (2019-2025)

In 2019, refined its News Feed to prioritize original videos longer than three minutes that garnered repeat views, while demoting low-quality or manipulative content sharing schemes. The platform also began downranking posts promoting misleading health information or unverified "cures" in , aiming to curb amid growing scrutiny. Mobile display optimizations followed in August, limiting overlaid text to three lines and adopting a 4:5 for images and videos to enhance visual appeal. By 2020, the algorithm emphasized "news ecosystem quality" (NEQ), boosting posts from original news sources with clear authorship in July while demoting unattributed articles. November saw intensified NEQ measures to reduce election-related , reflecting heightened regulatory and public pressure during the U.S. presidential cycle. These adjustments built on prior meaningful interactions focus but increasingly incorporated video prioritization, setting the stage for short-form content dominance. Facebook launched in the U.S. in September 2021 and expanded globally in February 2022, integrating short-form videos prominently into the Feed—often at the top—to rival . A June 2022 overhaul shifted the Feed toward video-centric recommendations, de-emphasizing text-based posts from pages in favor of engaging, algorithmically surfaced videos from diverse creators. The "News Feed" branding was dropped in February 2022 to distance from journalistic connotations, renaming it simply "Feed." From 2023 onward, played a larger role in Feed curation. Meta doubled AI-driven recommendations by July 2023, expanding suggested content beyond users' direct connections to predict engagement via models. This reduced reliance on the , introducing more posts from strangers and prioritizing original content while penalizing aggregators reposting excessively. In October 2024, updates further amplified non-connection recommendations, incorporating local exploration and AI-assisted features like prompts. In 2025, Meta enhanced visibility with an October algorithm tweak prioritizing recent videos—showing 50% more same-day posts—and incorporating user feedback signals like "Not Interested" for finer control over video feeds. Starting December 16, interactions with (e.g., chats on topics like ) will inform personalized Feed content and ads, excluding sensitive categories such as or from ad targeting, with options via preferences. These evolutions mark a transition from socially anchored feeds to AI-orchestrated discovery engines optimized for prolonged user retention through predictive relevance.

Operational Mechanics

Content Inventory and Sourcing

The content inventory for the Feed comprises the pool of potential stories eligible for display to a given user, drawn primarily from posts generated by their direct social connections. This includes non-deleted content shared by friends, followed Pages, and Groups the user has joined, such as status updates, photos, videos, links, and reactions. On average, this inventory yields over 1,000 candidate posts per user per day, incorporating recent activity since the user's last session as well as unread or interaction-bumped items to prioritize unseen relevant content. Sourcing originates from user-generated uploads and publications within the platform's ecosystem, where individuals and entities post directly to their profiles, Pages, or Groups. The system aggregates these via a feed aggregator that queries underlying databases for actions (e.g., posts, comments), associated objects, and summaries, forming an initial set of candidates before any ranking or filtering. Content violating Meta's Community Standards, such as hate speech or misinformation, is excluded from the inventory during preprocessing to prevent eligibility for display. Beyond organic user connections, the inventory incorporates sponsored advertisements from Meta's advertising platform, where businesses bid to place targeted content, and algorithmic recommendations of posts from unconnected sources predicted to align with user interests based on broader signals like similar engagements. This expanded sourcing reflects the Feed's evolution to include diverse formats, though core inventory remains anchored in explicit connections to mitigate over-reliance on inferred preferences.

Ranking and Personalization Algorithms

Facebook's News Feed employs systems to personalize content for over 2 billion users by predicting engagement likelihood and . The process begins with an inventory of potential posts from connected friends, followed pages, groups, and recommended content, generating thousands of candidates per session. models then analyze signals—such as user relationships, post recency, content type, and historical interactions—to compute prediction scores for actions like comments, reactions, shares, or hides. These scores contribute to a final that determines display order, prioritizing content expected to foster "meaningful social interactions" over passive consumption. Key ranking factors include the strength of user-poster relationships, where posts from close friends or receive higher weights to demonstrated . Engagement metrics heavily influence scores: comments and expressive reactions (e.g., "love," "haha," "sad") carry more weight than likes, as they signal deeper involvement, while shares amplify reach through network effects. Content characteristics also factor in; original posts are favored over reposts to reward creators, and like photos or videos outperforms links, which are often downranked to reduce clickbait. Recency decays scores over time, ensuring timeliness, but the algorithm balances this with enduring relevance for content. Peer-reviewed analyses confirm that high-engagement posts dominate feeds, with videos and photos overrepresented compared to text or external links. Personalization draws from extensive user data, including past behaviors, device type, session context, and inferred interests from interactions across . Models predict not only positive but also negative signals like hides or unfollows to suppress undesired content, with iterative training on billions of daily interactions refining accuracy. Since 2018, the system has evolved from simpler formulas—factoring affinity, edge weight, and time decay—to sophisticated multi-stage neural networks handling over 15,000 signals per post. Updates in 2023 emphasized originality to curb aggregation, while 2025 integrations incorporate AI-driven interactions for broader recommendation tuning. This -optimized approach, while effective for retention, has been critiqued in studies for amplifying polarizing content that elicits strong reactions, as algorithms treat such signals as proxies without inherent quality filters.

Delivery and User Experience Optimization

Delivery of content in the Facebook Feed occurs after ranking, where server-generated story candidates are transmitted to the client device for final presentation. To optimize this process, Facebook employs client-side ranking, which reorders stories using locally cached unseen content and newly fetched items, reducing dependency on network latency. This technique prioritizes stories with fully loaded media, such as images and videos, to minimize visual placeholders like spinners or gray boxes, thereby accelerating perceived load times especially under variable connectivity conditions. Media assets within Feed stories, including photos and videos, are served through Facebook's (CDN), which incorporates advanced caching strategies to reduce backbone traffic and delivery latency. The CDN evolution includes predictive caching based on user behavior patterns, ensuring frequently accessed content is stored closer to end-users, which has been shown to improve photo and video serving speeds globally. For video-specific delivery integral to the Feed, a unified coordinates ranking signals with server-side encoding and mobile client adaptations, enabling to match network conditions. User experience optimization focuses on reducing friction in content consumption, such as through memory-efficient data structures on Android devices. By replacing standard collections like HashSet with primitive-based alternatives (e.g., LongArraySet), Feed rendering cuts object allocations by approximately 30%, leading to fewer garbage collection pauses and smoother scrolling with reduced frame drops. Additionally, AI models interpret sparse user feedback signals, such as "Show More" or "Show Less" interactions, by generating embeddings for users and posts via transformer-based neural networks fine-tuned on billions of interactions. This enhances by upweighting preferred content types across the Feed, improving relevance for users even with infrequent feedback and allowing greater control over curation. These optimizations collectively aim to maintain low end-to-end latency, with Feed systems targeting under 50 milliseconds for at scale for billions of daily , fostering sustained without compromising device performance. Prefetching mechanisms further aid delivery by anticipating user actions, such as pre-loading linked content in anticipation of taps, though primarily documented in contexts. Empirical adjustments via ensure these techniques adapt to diverse hardware and network environments, prioritizing causal improvements in retention over theoretical ideals.

Core Features and User Interactions

Supported Content Types

The Facebook Feed supports a diverse array of content types designed to facilitate sharing and interaction among users, Pages, and Groups. Primary organic content includes text-based status updates, which allow users to post short messages or thoughts. Photographs and image albums enable visual sharing, often accompanied by captions. Videos, ranging from user-uploaded clips to short-form Reels, constitute a significant portion of Feed content, with Reels optimized for algorithmic promotion due to high engagement rates. Shared links to external websites or articles appear as previews with headlines, descriptions, and thumbnails, promoting information dissemination beyond the platform. Live video broadcasts provide real-time streaming, fostering immediate audience interaction through comments and reactions. Event creations and invitations notify users of upcoming gatherings, integrating calendar-like functionality into the Feed. Posts from followed Pages and Groups aggregate professional, community, or interest-based content, such as updates from businesses or discussion threads. Algorithmically recommended content, including suggested posts from non-followed sources, expands visibility based on inferred user interests derived from past interactions. App activity and likes from connected applications or third-party integrations also surface, though less prominently in recent updates prioritizing direct user-generated material. Advertisements, seamlessly integrated, mimic organic formats like video or image ads but are labeled for transparency. While Stories offer ephemeral sharing in a dedicated , cross-promotions or highlights occasionally appear in the main Feed to drive . This multifaceted support underscores the Feed's role in curating personalized streams, though prioritization favors video and interactive elements as of 2025.

Engagement Mechanisms

Facebook's Feed incorporates interactive features that enable users to respond to content, generating signals for algorithmic personalization and ranking. These mechanisms, including reactions, comments, and shares, are designed to capture explicit user feedback, with the platform's systems interpreting them to predict future interactions and surface relevant posts. Explicit signals such as liking, commenting, or resharing contribute to content prioritization, while the absence of engagement can limit visibility. Reactions represent a core engagement tool, launched globally on February 24, 2016, as an extension of the original "Like" button introduced in 2009. Users can select from six emoji-based options—Like, Love, Haha, Wow, Sad, or Angry—to convey nuanced sentiments without composing text, addressing limitations of the binary Like system. Internal data from Meta indicates that reactions, particularly Love and Haha, correlate with higher post persistence in feeds, as they signal emotional resonance and are weighted similarly to likes in ranking models but differentiated for content type relevance, such as positive reactions favoring uplifting posts. Comments facilitate extended discourse through threaded replies, allowing users to elaborate on posts and engage with others' responses. This mechanism is valued for its depth, as composing comments requires greater cognitive effort than reactions, leading algorithms to assign higher predictive weight to them in determining content quality and user interest. Meta's systems analyze comment volume, sentiment, and reciprocity—such as replies between friends—to amplify posts sparking meaningful conversations, with empirical observations showing that posts receiving early comments expand distribution more rapidly than those reliant solely on reactions. Shares enable users to redistribute content to their networks, either publicly or to specific groups, thereby extending reach beyond the original . As a high-effort action implying endorsement, shares serve as a potent virality signal, prompting to boost similar content for the sharer and their connections. Distribution models treat shares as multiplicative, with one share potentially exposing content to hundreds via network effects, though Meta has adjusted weights post-2018 to prioritize shares among close ties over mass dissemination to mitigate spread.

Customization and Control Options

Users can adjust Facebook Feed content through post-level interactions, such as selecting "Interested" or "Not interested" on individual posts, which signals to show more or fewer similar items from that source or topic. Hiding a specific post or opting to see fewer from a , page, or group further refines visibility, with options to snooze content for 30 days or unfollow entirely to remove it from the Feed without unfriending. These adjustments are processed via to personalize future rankings, though their long-term impact depends on ongoing user signals and algorithmic weighting. The platform provides a dedicated Feed preferences , accessible primarily from desktop, where users can manage categories like reducing recommendations on specific topics, prioritizing posts from friends and family over public content, or limiting political and news-related items. Mobile users can access similar controls through the app's under Settings > Preferences > Feed, including toggles for video autoplay, photo grid views, and notifications tied to Feed interactions. As of 2025, Meta has enhanced these with AI-driven tools allowing users to fine-tune and video recommendations separately, emphasizing user-selected interests to counterbalance algorithmic defaults. For chronological viewing, the Feeds tab—introduced in 2022 and available in the app's navigation—displays posts in reverse chronological order from followed friends, groups, Pages, and favorites, bypassing the default ranking algorithm that prioritizes predicted engagement. This tab supports temporary switches to "Most Recent" mode, though the primary Feed reverts to algorithmic sorting unless manually adjusted, with a 2025 filter bar update adding explicit toggles between ranked, recent, and favorites-focused views to increase user agency over content sequencing. Additional controls include custom friend lists for segmented viewing (e.g., Close Friends or Acquaintances), which influence prioritization in the main Feed, and ad-related preferences that indirectly affect sponsored content density by refining targeting based on user data exclusions. These options collectively enable granular control, but empirical analyses indicate that heavy reliance on defaults perpetuates algorithmic curation, with user-initiated tweaks requiring consistent application to meaningfully alter the experience.

Economic and Data Dimensions

Integration with Advertising

Advertising on the Facebook Feed is integrated via sponsored posts that mimic the visual and interactive format of organic content, such as status updates, photos, and videos from connected users and pages, while bearing a "Sponsored" label for disclosure. These ads occupy slots within the personalized algorithmic sequence of the Feed, typically appearing every few organic posts to maintain user engagement without overwhelming the stream. The system ensures ads are contextually relevant, drawing from advertiser-selected objectives like awareness, traffic, or conversions, and are delivered across desktop and mobile interfaces. Ad placement in the Feed relies on Meta's real-time auction mechanism, where impressions are allocated based on advertiser bids competing for user-specific opportunities. Upon an advertiser defining a —using criteria like age, location, interests, behaviors, or custom lists such as uploads—the eligible ads enter the . Meta calculates a total value score for each ad, comprising the advertiser value (bid amount times the machine learning-predicted estimated action rate, reflecting the probability of user actions like clicks or purchases) plus an ad quality score (evaluating via anticipated minus negative signals like hides or reports). This score determines auction winners, with higher values securing Feed visibility; ads thereby contend against both rival ads and organic content for prioritization in the broader Feed process, which weighs predicted across all items. models, informed by aggregated user data including on-platform interactions and off-platform signals from tools like the Meta Pixel, refine these predictions to favor ads likely to yield positive outcomes for users and advertisers alike. This integration underpins ' primary monetization, as Feed-delivered ads generate the bulk of advertising revenue. For the full year 2024, Meta reported total revenue of $164.50 billion, with advertising comprising approximately 97-98% of that figure across and affiliated apps; family daily averaged 3.24 billion in December 2024, providing a vast inventory for ad s. Algorithmic refinements, including auction optimizations introduced in updates like those enhancing action rate estimations, have sustained revenue growth amid competitive pressures, with ad impressions scaled via automated bidding strategies that prioritize return on ad spend. Official disclosures from Meta emphasize these mechanics' role in balancing with viability, though independent analyses note potential over-reliance on engagement metrics that may amplify sensational content.

Utilization of User Data

Facebook's Feed employs systems that process extensive user data to personalize content ranking for over 2 billion users, predicting which posts will generate engagement and perceived value. These systems evaluate thousands of signals per post, including user interactions such as likes, , comments, shares, and clicks, to compute affinity scores and probabilities. Behavioral metrics like time spent viewing content, scrolling patterns, and activity frequency further refine predictions of interest and dwell time. Social graph data, encompassing mutual friends, group memberships, and relationship types (e.g., family or close friend status), informs prioritization by weighting posts from stronger connections higher. Contextual signals, including device type, operating system, and (where user permissions allow), enable adjustments for usability and immediacy, such as favoring recent content. Post-specific attributes—recency, media type (e.g., video versus photo), , and author details—are cross-referenced against a user's historical patterns to forecast outcomes like sharing likelihood or social ripple effects (e.g., induced comments from recipients). Over 100 models aggregate these inputs into multitask predictions, combining behavioral forecasts with survey-derived valuations of content "worth," such as time investment justification. Real-time processing scores more than 1,000 potential items per user daily, yielding a dynamic score that balances against diversity rules to mitigate echo chambers. This framework extends to recommended content from non-followed sources and advertisements, where inferred interests from aggregated data drive targeting while maintaining feed cohesion. User utilization emphasizes predicted meaningful interactions over raw volume, demoting low-engagement or negative-signal content based on hides, reports, and reduced viewing metrics. Empirical tuning via offline simulations and online tests ensures causal links between signals and outcomes like sustained platform retention. As of June 2025, these processes remain core to Feed operations, with ongoing refinements to handle evolving user behaviors amid regulations.

Empirical Impacts

Evidence on Social Connectivity and Engagement

Empirical analyses indicate that Facebook's feed , which ranks content based on predicted user interactions such as likes, comments, and shares, has driven sustained platform since its introduction in , with daily active users averaging over 2 billion by 2023 and session times influenced by algorithmic to maximize and recency. However, this often manifests as passive scrolling rather than active relationship-building, with internal Meta documents from 2021 revealing that tweaks in 2017-2018 prioritized "meaningful social interactions" from friends and family to counteract declining metrics like comments and shares, reportedly boosting user-reported satisfaction by promoting closer ties over viral content. Research on social connectivity yields mixed findings, with some evidence suggesting the feed supports maintenance of weak ties and larger networks that correlate with reduced ; for instance, users with broader friend networks report lower isolation levels, as curates feeds to include diverse connections beyond chronological posts. Conversely, longitudinal studies link higher feed engagement to increased social comparison and rumination, exacerbating , particularly among heavy users who spend more time viewing idealized posts from acquaintances rather than engaging deeply with close contacts. Quasi-experimental designs further illuminate causal dynamics: reducing exposure to positive or envy-inducing content in algorithmically curated feeds lowers users' positive posting tendencies but does not consistently enhance perceived connectivity, implying that algorithmic amplification of emotional content may sustain engagement at the expense of genuine relational depth. Cross-national surveys associate greater daily social media time, including Facebook feed usage, with elevated loneliness scores, especially for passive consumers who view without interacting, though active communication features mitigate this for some demographics. These patterns hold across age groups, with young adults showing stronger negative associations due to heightened vulnerability to upward comparisons facilitated by the feed's personalized ranking. Overall, while the algorithm fosters superficial connectivity through network expansion, empirical data underscores a net tendency toward diminished well-being from substitutive online interactions over offline bonds.

Research on Polarization and Opinion Formation

A 2023 analysis of data from all active U.S. adult users revealed that the user was exposed to 50.4% like-minded political content in their feed, compared to 14.7% content from opposing viewpoints, with only 20.6% of users seeing more than 75% like-minded material. This pattern arises from the algorithm's prioritization of engagement-driven content, which often aligns with users' existing networks and interactions, fostering selective exposure without extreme isolation for most. Field experiments testing causal effects have consistently found limited algorithmic influence on polarization. In a preregistered involving 23,377 consenting U.S. users from September 24 to December 23, 2020, downranking like-minded content reduced such exposure from 53.7% to 36.2% (P < 0.01) and increased neutral content to 35.9%, while also lowering uncivil or misinformation-laden posts; however, pre- and post- surveys detected no significant changes in eight key attitudinal outcomes, including affective polarization and ideological extremity, ruling out effects larger than ±0.12 standard deviations. Similarly, during the 2020 U.S. , switching users to reverse-chronological feeds on increased exposure to political and untrustworthy content but yielded no detectable shifts in issue polarization, affective polarization, or political . On opinion formation, evidence suggests Facebook feeds shape immediate engagement and awareness but exert minimal causal sway over enduring beliefs. A deactivation experiment around the 2020 election showed that ceasing access reduced news knowledge by 0.098 standard deviations (P < 0.05) and slightly lowered belief in misinformation, yet produced no meaningful changes in overall political knowledge, polarization, or Trump voting intention (point estimate -0.026 units, P = 0.054). These findings challenge earlier narratives of algorithm-fueled echo chambers driving deep attitudinal shifts, as users' offline networks and predispositions appear to dominate long-term opinion dynamics. Critiques of these studies highlight potential methodological limitations, such as the short duration of interventions (e.g., ) and reliance on consenting participants, which may limit generalizability to habitual users or broader populations. Nonetheless, the large-scale, data-driven designs—drawing from millions of interactions—provide robust empirical counterweight to claims of substantial algorithmic causation in polarization or opinion entrenchment.

Findings on Misinformation Propagation and Mental Health

A randomized deactivation experiment involving 19,857 U.S. users during the six weeks preceding the 2020 presidential election (September 23 to November 4) demonstrated that platform access causally increased belief in specific claims, such as allegations of millions of fraudulent ballots, with deactivation yielding a directional reduction in such beliefs (0.042 standard deviation increase in fact knowledge, p=0.11). This effect persisted in instrumental variable analyses accounting for self-selection, indicating that exposure via the algorithmic news feed contributes to entrenched false beliefs, though overall political knowledge effects were mixed (-0.033 SD, p=0.26). Analysis of over 900 million public posts during the 2020 U.S. election period revealed that content from domains publishing received six times more clicks than factual news, driven by higher user engagement rates that the feed prioritizes through metrics like reactions and shares. False or misleading articles, comprising a small fraction of total content, achieved disproportionate virality due to emotional and novelty, amplifying reach by up to 10-fold compared to true stories in analogous datasets, with similar dynamics inferred for 's engagement-optimized feed. However, a 2023 experiment randomizing users to chronological versus algorithmic feeds found the latter reduced exposure to untrustworthy content (with chronological increasing it by an estimated 0.014 proportion on ), suggesting algorithmic ranking demotes low-quality sources to some extent, though critics contend the study's short duration () and participant selection underestimated long-term amplification of polarizing . Theoretical models and empirical simulations indicate that recommendation algorithms like Facebook's, which maximize predicted engagement, inherently favor sensational over factual reporting, as novel falsehoods elicit stronger reactions (e.g., , surprise) than routine truths, leading to broader diffusion networks and reduced truth crowding-out. Recent data from 2025 show rising ideological polarization and prevalence on the platform, correlating with algorithmic changes that inadvertently boost biased content through personalized echo chambers, despite interventions like labels, which Meta reports reduce shares by 10-20% but fail to curb initial propagation. The staggered rollout of to U.S. colleges from 2004 to 2006, analyzed via 430,000 National College Health Assessment responses, established a causal link between feed-driven access and deterioration, with campus-wide adoption increasing severe depression reports by 7% and anxiety disorders by 20%, effects intensifying with prolonged exposure (e.g., two semesters versus one). These outcomes stemmed from mechanisms like upward social comparison, where users' feeds highlight peers' curated successes, prompting and self-doubt, alongside disrupted sleep from habitual scrolling. Longitudinal data from youth cohorts confirm dose-response patterns: adolescents averaging over three hours daily on platforms including faced twice the risk of depression and anxiety symptoms compared to lighter users (n=6,595, ages 12-15), with experimental limits to 30 minutes per day reducing depression severity by over 35% in those with high baseline symptoms. Systematic meta-analyses of 50+ studies report small but consistent positive correlations (r≈0.10-0.15) between time spent on feeds and internalizing disorders like depression, mediated by factors such as exposure and , though causation is stronger in experimental designs than purely observational ones. Peer-reviewed evidence attributes these effects to the feed's algorithmic emphasis on emotionally charged content, which heightens rumination and isolates users in loops, with college populations showing heightened vulnerability during peak usage periods. While correlational studies dominate, deactivation and rollout experiments provide causal substantiation, revealing no offsetting benefits in social connectivity sufficient to mitigate harms; instead, feed-induced isolation from real-world interactions exacerbates symptoms, as evidenced by rising and uptake post-adoption. Academic sources, often institutionally inclined toward highlighting platform risks, align with these findings but warrant scrutiny against self-reported data limitations, yet replication across datasets reinforces the feed's role in propagating declines through passive consumption patterns.

Controversies and Debates

Claims of Ideological Bias

Conservatives have long alleged that Facebook's news feed and practices exhibit a systemic left-leaning ideological , suppressing right-wing viewpoints while amplifying progressive ones. These claims gained prominence during the 2016 U.S. , with Republican lawmakers citing internal metrics showing reduced visibility for conservative pages after algorithm tweaks in 2017-2018, which purportedly prioritized "meaningful social interactions" over news diversity. A 2023 congressional report noted that despite adjustments to the , right-leaning pages experienced sustained lower engagement compared to left-leaning counterparts, attributing this to opaque ranking factors favoring established media outlets often aligned with liberal perspectives. A pivotal example cited in these claims is the October 2020 demotion of a article on Hunter Biden's laptop, which Facebook restricted pending ; CEO later attributed this to erroneous FBI warnings about potential Russian disinformation but acknowledged in 2022 that the platform erred by not treating it as newsworthy, effectively aiding Democratic narratives during the election. In August 2024, Zuckerberg publicly stated that the Biden administration had pressured Meta to censor COVID-19-related content, including true information on side effects and satirical posts, and expressed regret for complying, describing the interference as "wrong" and indicative of broader overreach that chilled conservative discourse. These admissions fueled accusations that the feed's enforcement of community standards disproportionately targeted right-leaning users, with internal data leaks suggesting fact-checkers, often sourced from left-leaning organizations, applied stricter scrutiny to conservative claims. Whistleblower allegations have amplified these concerns, though not all directly affirm partisan bias; former employee testified in 2021 that algorithms prioritized divisive content for engagement but stated she was unaware of research proving explicit political favoritism, focusing instead on general amplification of extremes that could disadvantage minority viewpoints. Critics, including Republican committee chairs, pointed to employee demographics—Zuckerberg himself noting in 2020 that most staff lean left—as a causal factor in moderation decisions, leading to deplatforming of figures like post-January 6, 2021, while permitting equivalent left-wing . By January 2025, Meta's decision to discontinue third-party programs, with Zuckerberg labeling prior moderation as "censorship," was hailed by conservatives as an implicit concession to claims, though the company maintained changes aimed at reducing government-influenced overreach rather than admitting ideological favoritism. Empirical studies present mixed evidence on these claims, with some attributing perceived to user-driven chambers rather than algorithmic intent. A 2023 analysis of U.S. users found the median feed comprised 50.4% like-minded political content versus 14.7% cross-cutting, but concluded the algorithm did not exacerbate polarization beyond baseline preferences, countering narratives of deliberate conservative suppression. Another study from the same year, using de-identified data, detected asymmetric segregation where conservatives encountered more ideologically homogeneous feeds, potentially reinforcing claims of right-wing isolation, yet emphasized that news feed ranking itself introduced minimal compared to chronological alternatives. Skeptics of the bias allegations, including reports from think tanks like ITIF, argue that both parties decry inconsistencies, but conservative complaints dominate due to higher reliance on platforms for unfiltered reach, with left-leaning sources occasionally claiming the opposite—algorithmic promotion of right-wing —though such counter-claims often rely on selective metrics from potentially biased academic outlets. Despite algorithmic transparency efforts, such as Meta's 2023 collaborations with researchers, opacity in weights persists, sustaining distrust among conservatives who view mainstream validations as tainted by institutional left-wing skews in academia and media.

Criticisms Regarding Content Amplification

Facebook's feed has faced criticism for prioritizing user metrics, such as likes, shares, and reactions, which systematically amplifies sensational and divisive content over neutral or informative material. Internal company research from revealed that these algorithms "exploit the human brain's attraction to divisiveness," as often correlates with emotionally charged posts that provoke outrage or strong reactions rather than substantive discussion. Critics argue this design incentivizes creators to produce rage-inducing content, as evidenced by the platform's shift to "meaningful social interactions," which weighted reactions—including anger—equally with positive signals, thereby boosting polarizing posts in users' feeds. Whistleblower , a former Facebook product manager, testified before in October 2021 that the algorithms exacerbate societal harms by amplifying and foreign interference, drawing from leaked internal documents showing the platform's awareness of these effects but prioritization of growth over mitigation. Haugen highlighted how the system manipulates user behavior through personalized recommendations that favor high-engagement content, increasing exposure to extreme views and contributing to events like the , 2021, U.S. Capitol riot via unchecked amplification of inflammatory narratives. She contended that while could adjust algorithms to reduce such amplification—such as by demoting divisive posts—the company resisted changes that might lower overall time spent on the platform, as engagement directly correlates with advertising revenue. Empirical analyses have supported these concerns, with a 2024 study finding that engagement-based ranking on platforms like amplifies emotionally charged content, including anger-expressing posts, beyond what users would select independently, potentially deepening affective polarization. Additional indicated algorithmic recommendations promote climate denial and misogynistic material, as these topics generate disproportionate interactions; for instance, a 2022 investigation showed 's feed surfacing denialist content attacking mitigation efforts despite user demographics leaning toward acceptance of anthropogenic warming. Critics, including Haugen, note that such amplification persists due to the profit-driven nature of the model, where algorithmic opacity allows unchecked escalation of harmful content loops, though Meta has disputed the extent, claiming user choices and chronological feeds mitigate extremes.

User and Creator Responses

Users have historically expressed significant discontent with changes to Facebook's News Feed, particularly regarding perceived invasions of and algorithmic manipulation. In September 2006, the introduction of the News Feed feature, which aggregated friends' activities into a centralized stream, prompted widespread protests from users who viewed it as an unauthorized breach of control over personal information visibility, leading to organized groups like "Students Against Facebook News Feed" and temporary dips in user engagement. Facebook responded by adding privacy controls, which mitigated but did not eliminate the outcry, as users reported feelings of illusory loss of despite technical opt-outs. Further controversy arose in June 2014 when Facebook disclosed an experiment manipulating the News Feeds of 689,003 users to study , altering the proportion of positive or negative posts to observe mood shifts, which elicited accusations of unethical without explicit consent. Public reaction included demands for regulatory oversight from privacy advocates and calls for user boycotts, though measurable exodus was limited; ethicists criticized the study for conflating from with informed participation in research. Surveys indicate persistent user dissatisfaction with the Feed's relevance and transparency amid ongoing algorithmic tweaks favoring engagement over utility. A 2014 survey found 43.2% of users perceived their News Feeds as less relevant than six months prior, coinciding with shifts toward personalized but opaque ranking. By 2018, Pew Research Center reported that 64% of U.S. Facebook users aged 18 and older did not fully understand how the News Feed prioritized content, contributing to frustration over algorithmic "black box" decisions. Satisfaction hit record lows in 2019 per the American Customer Satisfaction Index, linked to scandals amplifying distrust in Feed curation, though many users continued access due to habitual and social dependencies. In response to claims of ideological and content amplification in the Feed, users have voiced concerns over exposure to polarizing or rage-inducing material, with anecdotal reports of feeds shifting toward unsolicited partisan content post-2020 election tweaks. Empirical studies, such as those from the 2020 U.S. election period, reveal users often self-select ideologically aligned sources, which algorithms reinforce, leading to complaints of chambers rather than deliberate platform ; however, conservative users have alleged under-amplification until 2023 adjustments boosted right-leaning page interactions. Pew data from 2024 shows 74% of users encounter via the Feed, but satisfaction varies, with some reducing usage to avoid amplified or mental health strains from divisive posts. Content creators have adapted to Feed controversies by optimizing for algorithmic signals like emotional , often criticizing reduced organic reach as a barrier to . Post-2018 of "meaningful interactions," creators reported sharp declines in post views—sometimes over 50% for non-follower audiences—prompting diversification to platforms like or for better monetization. In response to amplification biases favoring , many shifted toward short-form video like , which saw algorithmic boosts in 2025, though complaints persist about unpredictable drops tied to perceived inconsistencies. Creators alleging ideological suppression, particularly conservatives pre-2023 tweaks, have lobbied for transparency, with some evidence of right-leaning videos gaining traction after adjustments, yet overall, high-engagement creators continue prioritizing controversy to combat reach erosion. Despite vocal critiques, surveys link creator continuance to perceived usefulness in audience building, underscoring economic incentives overriding dissatisfaction.

Major Investigations and Lawsuits

In June 2022, the U.S. Department of Justice (DOJ) settled a lawsuit against Meta Platforms, alleging that Facebook's ad delivery algorithms violated the Fair Housing Act by disproportionately excluding users based on protected characteristics such as age, gender, and family status when serving housing advertisements in users' feeds. The complaint highlighted how machine learning models optimized ad distribution for advertiser goals like cost efficiency, resulting in discriminatory outcomes independent of explicit targeting, with evidence from audits showing older users and those with children receiving fewer ads for desirable neighborhoods. Under the agreement, Meta committed to altering its algorithms, disabling certain "lookalike audience" tools for housing-related ads, and submitting to independent audits for five years to mitigate disparate impacts. In December 2020, the (FTC) filed an antitrust lawsuit against Meta, claiming the company maintained an unlawful monopoly in personal social networking by acquiring competitors like in 2012 and in 2014, thereby entrenching control over feed-based engagement that drives network effects and . The suit argues that these actions suppressed innovation in algorithm-driven content recommendation and user interaction features central to Facebook's feed, with the case advancing to a in April 2025 after evidentiary disputes over Meta's internal documents. A related antitrust alleging misrepresentation of data practices to stifle was dismissed in September 2025 for lack of admissible evidence on harms. In July 2019, the FTC imposed a $5 billion on —the largest ever for violations—resolving allegations that the platform deceived users about data controls, including third-party access exploited in the scandal, which enabled targeted feed content for political influence campaigns affecting millions. The order mandated comprehensive program reforms, independent assessments, and restrictions on facial recognition use, indirectly constraining algorithmic personalization in feeds reliant on user data inferences. In May 2024, MIT professor and the Knight First Amendment Institute sued Meta, contending that of the does not shield the company from First Amendment obligations to allow users third-party tools or opt-outs to customize or bypass proprietary feed recommendation algorithms, which curate content based on engagement predictions rather than chronological order. The suit challenges Meta's policy prohibiting modifications to its ranking systems, arguing it limits user autonomy over algorithmic gatekeeping without qualifying as editorial immunity for intermediary platforms. In October 2024, the preliminarily found Meta in breach of the for failing to transparently explain decisions on and , including algorithmic demotions in feeds, prompting formal proceedings separate from ongoing probes into systemic risks like amplification. These DSA investigations scrutinize how recommendation systems prioritize content, requiring platforms to disclose design choices affecting visibility.

Policy Reforms and Platform Adjustments

In response to growing concerns over user time expenditure and content quality, Facebook announced a significant algorithm adjustment on January 19, 2018, prioritizing "meaningful interactions" from friends and family over public posts, videos, and news from Pages or publishers, aiming to foster personal connections rather than passive media consumption. This shift, detailed by CEO , responded to internal data showing users derived greater satisfaction from social interactions, resulting in an estimated 5-10% reduction in time spent on the platform while increasing reported among surveyed users. To address misinformation propagation, particularly following the 2016 U.S. election, Facebook implemented policy reforms in 2017, partnering with third-party organizations under the International Fact-Checking Network to identify and downrank false stories in the Feed, reducing their distribution by an average of 80% according to platform metrics. These measures expanded in 2019 with tweaks that penalized content from repeated violators and introduced a "click-gap" metric to detect engagement bait, further curbing low-quality or misleading posts. Independent analyses, including a 2024 study from the , confirmed that subsequent iterations of the reduced exposure to untrustworthy election-related content by at least 24%, countering claims of amplification. Regulatory pressures, including the 2018 Cambridge Analytica scandal and subsequent EU GDPR enforcement, prompted transparency reforms such as the 2019 launch of a dedicated News tab separate from the main Feed to isolate journalistic content, alongside algorithm demotions for non-compliant publishers. In 2020, updates weighted credible, original news sources more heavily in rankings, informed by partnerships with over 80 fact-checkers globally, though critics noted potential over-reliance on left-leaning verifiers raised questions of ideological skew in demotions. By 2023, Meta further adjusted the Feed to deprioritize political and news content overall, citing internal research linking high civic exposure to user dissatisfaction and polarization, with studies estimating a 20-30% drop in news visibility post-change. Recent adjustments as of 2025 emphasize user controls and entertainment formats, including expanded options to customize Feed preferences for over posts and AI-driven personalization that gives "second chances" to undelivered content based on predicted relevance, aiming to boost without amplifying divisive material. These reforms, while reducing reach per platform audits, have drawn scrutiny for potentially entrenching echo chambers, as evidenced by a 2023 Nature study showing the algorithm's role in segregating liberal and conservative news exposure. Meta maintains these changes align with empirical signals from billions of interactions, prioritizing causal factors like predicted user reactions over raw metrics.

References

  1. https://www.facebook.com/business/news/News-Feed-FYI-A-Window-Into-News-Feed
  2. https://about.fb.com/news/2016/06/building-a-better-news-feed-for-you/
  3. https://wallaroomedia.com/facebook-newsfeed-algorithm-history/
  4. https://www.influize.com/blog/the-history-of-facebook
  5. https://www.cnet.com/pictures/facebook-then-and-now-pictures/
  6. https://shorensteincenter.org/resource/facebook-friends/
  7. https://epjdatascience.springeropen.com/articles/10.1140/epjds/s13688-025-00585-3
  8. https://powerdigitalmarketing.com/blog/facebook-advertising-and-news-feed-algorithm-history/
  9. https://www.facebook.com/UMassAmherst/posts/though-facebook-can-limit-untrustworthy-content-new-research-suggests-often-choo/947782580727434/
  10. https://www.nbcnews.com/tech/social-media/timeline-facebook-s-privacy-issues-its-responses-n859651
  11. https://slate.com/technology/2013/09/facebook-news-feed-turns-7-why-its-the-most-influential-feature-on-the-internet.html
  12. https://www.infegy.com/blog/the-news-feed-the-revolution-of-media-consumption
  13. https://about.fb.com/news/2006/09/facebook-launches-additional-privacy-controls-for-news-feed-and-mini-feed/
  14. https://privacyinternational.org/examples/2133/facebook-introduces-news-feed
  15. https://www.washingtonpost.com/technology/interactive/2021/how-facebook-algorithm-works/
  16. https://www.officetimeline.com/blog/facebook-history-timeline
  17. https://www.weforum.org/stories/2016/09/facebooks-news-feed-is-10-years-old-this-is-how-the-site-has-changed/
  18. https://techcrunch.com/2010/04/22/facebook-edgerank/
  19. https://markinuity.com/educational/the-evolution-of-the-facebook-algorithm/
  20. https://www.shorensteincenter.org/resource/facebook-friends/
  21. https://about.fb.com/news/2022/02/launching-facebook-reels-globally/
  22. https://www.socialmediacollege.com/blog/meta-is-doubling-ai-recommendations-in-your-feed
  23. https://fortune.com/2024/10/04/facebook-update-meta-ai-dating-local-explore/
  24. https://techcrunch.com/2025/10/07/facebook-updates-its-algorithm-to-give-users-more-control-over-which-videos-they-see/
  25. https://about.fb.com/news/2025/10/improving-your-recommendations-apps-ai-meta/
  26. https://transparency.meta.com/features/explaining-ranking/fb-feed-recommendations/
  27. https://engineering.fb.com/2021/01/26/core-infra/news-feed-ranking/
  28. https://transparency.meta.com/features/ranking-and-content/
  29. https://www.facebook.com/business/help/718033381901819
  30. https://tech.facebook.com/engineering/2021/1/news-feed-ranking/
  31. https://www.scienceopen.com/hosted-document?doi=10.14293/S2199-1006.1.SOR-.PPOU7F9.v1
  32. https://www.researchgate.net/publication/339307584_How_biased_is_the_sample_Reverse_engineering_the_ranking_algorithm_of_Facebook%27s_Graph_application_programming_interface
  33. https://about.fb.com/news/2018/05/inside-feed-news-feed-ranking/
  34. https://academic.oup.com/pnasnexus/article/4/3/pgaf062/8052060
  35. https://engineering.fb.com/2016/10/20/networking-traffic/client-side-ranking-to-more-efficiently-show-people-stories-in-feed/
  36. https://research.facebook.com/blog/2016/4/the-evolution-of-advanced-caching-in-the-facebook-cdn/
  37. https://engineering.fb.com/2024/12/10/video-engineering/inside-facebooks-video-delivery-system/
  38. https://engineering.fb.com/2016/06/27/android/memory-optimization-for-feeds-on-android/
  39. https://ai.meta.com/blog/facebook-feed-improvements-ai-show-more-less/
  40. https://naina0405.substack.com/p/system-design-tech-case-study-pulse-e03
  41. https://www.facebook.com/business/help/1514372351922333
  42. https://www.facebook.com/help/1155510281178725
  43. https://www.socialmediaexaminer.com/facebook-content-strategy-2025-whats-actually-working-right-now/
  44. https://www.facebook.com/government-nonprofits/blog/demystifying-news-feed
  45. https://developers.facebook.com/docs/graph-api/reference/page/feed/
  46. https://about.fb.com/news/2021/01/how-does-news-feed-predict-what-you-want-to-see/
  47. https://www.facebook.com/business/help/1082519118875784
  48. https://www.outfy.com/blog/facebook-reels-vs-stories-vs-posts/
  49. https://blog.hootsuite.com/facebook-algorithm/
  50. https://about.fb.com/news/2016/02/reactions-now-available-globally/
  51. https://about.fb.com/news/2021/04/incorporating-more-feedback-into-news-feed-ranking/
  52. https://about.fb.com/news/2018/01/news-feed-fyi-bringing-people-closer-together/
  53. https://www.facebook.com/help/231515559920930
  54. https://www.facebook.com/help/1913802218945435
  55. https://www.facebook.com/help/371675846332829
  56. https://www.facebook.com/help/964154640320617
  57. https://www.facebook.com/help/218728138156311
  58. https://www.facebook.com/help/1774857799411802
  59. https://www.facebook.com/business/ads
  60. https://www.facebook.com/business/news/good-questions-real-answers-how-does-facebook-use-machine-learning-to-deliver-ads
  61. https://www.facebook.com/business/ads/ad-auction
  62. https://investor.atmeta.com/investor-news/press-release-details/2025/Meta-Reports-Fourth-Quarter-and-Full-Year-2024-Results/
  63. https://www.facebook.com/business/help/430291176997542
  64. https://about.fb.com/news/2023/06/how-ai-ranks-content-on-facebook-and-instagram/
  65. https://www.science.org/doi/10.1126/science.abp9364
  66. https://www.cnn.com/2021/10/27/tech/facebook-papers-meaningful-social-interaction-news-feed-math
  67. https://cyberpsychology.eu/article/view/14419
  68. https://academic.oup.com/jcmc/article/29/5/zmae017/7750016
  69. https://pmc.ncbi.nlm.nih.gov/articles/PMC7820562/
  70. https://cyberpsychology.eu/article/view/21188
  71. https://pmc.ncbi.nlm.nih.gov/articles/PMC11373151/
  72. https://pmc.ncbi.nlm.nih.gov/articles/PMC9817115/
  73. https://www.sciencedirect.com/science/article/pii/S245195882100018X
  74. https://pmc.ncbi.nlm.nih.gov/articles/PMC3335399/
  75. https://www.nature.com/articles/s41586-023-06297-w
  76. https://www.pnas.org/doi/10.1073/pnas.2321584121
  77. https://www.science.org/content/article/study-found-facebook-algorithm-didnt-promote-political-polarization-critics-doubt
  78. https://engineering.nyu.edu/news/misinformation-facebook-got-six-times-more-clicks-factual-news-during-2020-election-study-says
  79. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5316506
  80. https://www.uoc.edu/en/news/2025/fake-news-and-polarization-on-facebook
  81. https://transparency.meta.com/features/approach-to-misinformation
  82. https://mitsloan.mit.edu/ideas-made-to-matter/study-social-media-use-linked-to-decline-mental-health
  83. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3919760
  84. https://www.hhs.gov/sites/default/files/sg-youth-mental-health-social-media-advisory.pdf
  85. https://www.sciencedirect.com/science/article/pii/S0165032724014265
  86. https://pmc.ncbi.nlm.nih.gov/articles/PMC10129173/
  87. https://itif.org/publications/2023/10/26/the-facts-behind-allegations-of-political-bias-on-social-media/
  88. https://www.congress.gov/118/meeting/house/115561/documents/HHRG-118-IF16-20230328-SD038.pdf
  89. https://www.bbc.com/news/world-us-canada-62688532
  90. https://www.politico.com/news/2024/08/26/zuckerberg-meta-white-house-pressure-00176399
  91. https://www.reuters.com/technology/zuckerberg-says-biden-administration-pressured-meta-censor-covid-19-content-2024-08-27/
  92. https://www.cnn.com/2024/08/27/business/mark-zuckerberg-meta-biden-censor-covid-2021
  93. https://committees.parliament.uk/oralevidence/2884/html/
  94. https://www.npr.org/2025/01/07/nx-s1-5251151/meta-fact-checking-mark-zuckerberg-trump
  95. https://www.science.org/doi/10.1126/science.ade7138
  96. https://english.elpais.com/science-tech/2023-07-27/right-wing-bias-a-macro-study-confirms-that-facebook-disinformation-is-consumed-by-conservatives.html
  97. https://solomonmg.github.io/pdf/science.adk1211.pdf
  98. https://www.theverge.com/2020/5/26/21270659/facebook-division-news-feed-algorithms
  99. https://www.cbsnews.com/news/facebook-whistleblower-frances-haugen-60-minutes-polarizing-divisive-content/
  100. https://www.npr.org/2021/10/05/1043377310/facebook-whistleblower-frances-haugen-congress
  101. https://www.technologyreview.com/2021/10/05/1036519/facebook-whistleblower-frances-haugen-algorithms/
  102. https://theconversation.com/facebook-whistleblower-frances-haugen-testified-that-the-companys-algorithms-are-dangerous-heres-how-they-can-manipulate-you-169420
  103. https://pmc.ncbi.nlm.nih.gov/articles/PMC11894805/
  104. https://www.globalwitness.org/en/campaigns/digital-threats/climate-divide-how-facebooks-algorithm-amplifies-climate-disinformation/
  105. https://www.theguardian.com/media/2024/feb/06/social-media-algorithms-amplifying-misogynistic-content
  106. https://www.scientificamerican.com/article/facebook-whistleblower-testified-that-companys-algorithms-are-dangerous-heres-why/
  107. https://www.technologyreview.com/2021/03/11/1020600/facebook-responsible-ai-misinformation/
  108. https://techcrunch.com/2006/09/06/facebook-users-revolt-facebook-replies/
  109. https://faculty.ist.psu.edu/xu/papers/ecra.pdf
  110. https://www.theguardian.com/technology/2014/jun/29/facebook-users-emotions-news-feeds
  111. https://www.nytimes.com/2014/06/30/technology/facebook-tinkers-with-users-emotions-in-news-feed-experiment-stirring-outcry.html
  112. https://www.forbes.com/sites/gregorymcneal/2014/06/28/facebook-manipulated-user-news-feeds-to-create-emotional-contagion/
  113. https://epic.org/documents/in-re-facebook-psychological-study/
  114. https://www.ignitesocialmedia.com/twitter-marketing/users-find-facebook-less-interesting-less-used/
  115. https://www.pewresearch.org/short-reads/2018/09/05/many-facebook-users-dont-understand-how-the-sites-news-feed-works/
  116. https://martech.org/americans-record-lowest-satisfaction-with-facebook-since-2015-according-to-asci/
  117. https://www.reddit.com/r/facebook/comments/1ejie2k/anyone_else_get_the_feeling_that_facebook_is/
  118. https://www.npr.org/2023/07/27/1190383104/new-study-shows-just-how-facebooks-algorithm-shapes-conservative-and-liberal-bub
  119. https://www.pewresearch.org/journalism/2024/06/12/facebook-users-experiences-with-news/
  120. https://www.facebook.com/groups/2106162202855060/posts/3127678320703438/
  121. https://aicontentfy.com/en/blog/impact-of-social-media-algorithms-on-content-distribution
  122. https://stackinfluence.com/2025-social-media-algorithm-changes-engagement/
  123. https://www.reddit.com/r/facebook/comments/1nz4fog/seriously_whats_going_on_with_facebook_reels_i/
  124. https://targetinternet.com/resources/is-the-facebook-algorithm-biased
  125. https://www.researchgate.net/publication/358350313_Facebook_Users%27_Satisfaction_and_Intention_to_Continue_Using_It_Applying_the_Expectation_Confirmation_Model
  126. https://www.propublica.org/article/facebook-doj-advertising-discrimination-settlement
  127. https://www.justice.gov/usao-sdny/pr/united-states-attorney-resolves-groundbreaking-suit-against-meta-platforms-inc-formerly
  128. https://www.ftc.gov/legal-library/browse/cases-proceedings/191-0134-facebook-inc-ftc-v-ftc-v-meta-platforms-inc
  129. https://www.pbs.org/newshour/nation/historic-antitrust-trial-could-force-meta-to-break-off-instagram-whatsapp
  130. https://www.reuters.com/legal/government/meta-defeats-antitrust-lawsuit-over-facebook-user-data-2025-09-30
  131. https://www.ftc.gov/news-events/news/press-releases/2019/07/ftc-imposes-5-billion-penalty-sweeping-new-privacy-restrictions-facebook
  132. https://www.wired.com/story/meta-section-230-users-algorithm/
  133. https://www.cjr.org/the_media_today/meta-facebook-lawsuit-algorithms-ethan-zuckerman.php
  134. https://ec.europa.eu/commission/presscorner/detail/en/ip_25_2503
  135. https://digital-strategy.ec.europa.eu/en/news/commission-preliminarily-finds-tiktok-and-meta-breach-their-transparency-obligations-under-digital
  136. https://about.fb.com/news/2018/04/inside-feed-meaningful-interactions/
  137. https://www.wired.com/story/facebook-click-gap-news-feed-changes/
  138. https://www.umass.edu/news/article/team-led-umass-amherst-debunks-research-showing-facebooks-news-feed-algorithm-curbs
  139. https://arxiv.org/html/2507.19373v1
  140. https://buffer.com/resources/facebook-algorithm/
Add your contribution
Related Hubs
Contribute something
User Avatar
No comments yet.