Removing personalized feed ranking would change what many people see and how much they use a platform, but it would not reliably “fix” social media. Controlled experiments found lower activity in chronological feeds, altered exposure to political and untrustworthy material, and measurable changes in some political views when ranking was switched on. They did not show a consistent reduction in polarization, misinformation, or every other harmful-content measure.
The answer depends on what “algorithm” means, which platform is tested, whose feed is changed, how long the change lasts, and whether the outcome is exposure, engagement, knowledge, attitudes, toxicity, or enjoyment.
As an Amazon Associate I earn from qualifying purchases.
“Removing the algorithm” is not one intervention
Every large social platform uses software somewhere in the feed and recommendation system. A chronological feed removes or reduces one kind of ranking, but it does not necessarily remove recommendations, moderation, search ordering, notifications, advertising selection, or the social network people have already built.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →| System component | What it does | What a chronological-feed experiment tests |
|---|---|---|
| Feed ordering | Chooses which eligible posts appear first, often using predicted engagement, relevance, safety, or other signals. | Whether replacing personalized ordering with newest-first ordering changes exposure and behavior. |
| Recommendations | Suggests accounts, posts, videos, or communities a person does not already follow. | Only changes if the experiment also removes or alters recommendations. |
| Content moderation | Removes, labels, limits, or demotes posts under platform rules. | Usually remains in place, so a chronological feed is not an unmoderated feed. |
| Social graph | Determines which accounts a person follows, blocks, mutes, or regularly sees. | Usually remains unchanged at the moment ranking is switched off. |
| Platform objective | Defines what the ranking system optimizes, such as recency, engagement, diversity, safety, or a combination. | Can be replaced with a different objective instead of being eliminated. |
This distinction matters because changing ordering cannot erase posts a user already chose to follow, accounts they discovered through earlier recommendations, or habits formed during previous exposure.
#1 Best Overall
What the major experiments found
Facebook and Instagram during the 2020 US election
A field experiment reported by Guess, Malhotra, Pan, and coauthors in Science in 2023 assigned consenting US users to reverse-chronological feeds on Facebook and Instagram or to the platforms’ normal feeds. The intervention lasted three months during the 2020 election period.
Chronological ordering substantially reduced time spent on the platforms and overall activity. It also changed the content participants encountered. On both platforms, they saw more political content and more content classified as untrustworthy. On Facebook specifically, they saw less content classified as uncivil or containing slur words, more posts from moderate friends, and more material from sources with ideologically mixed audiences.
Those exposure changes did not translate into significant changes in issue polarization, affective polarization, political knowledge, or other key measured attitudes during the study. The result is not evidence that ranking never matters. It is evidence that, for these users, platforms, outcomes, and three-month election-period intervention, reverse chronology did not produce broad attitude change.
Algorithmic versus chronological feeds on X
A separate field experiment recruited active US X users in 2023 and assigned them to an algorithmic or chronological feed for approximately seven weeks. The final post-treatment sample contained 4,965 participants. The study was published in Nature in 2026.
Among participants who began with chronological feeds, switching on the algorithm increased engagement and shifted some policy priorities and views on current events in a conservative direction. Switching the algorithm off for participants who initially had the algorithmic feed did not create a comparable reversal. Neither direction of switching significantly changed affective polarization or self-reported partisanship.
The authors observed a mechanism that helps explain the asymmetry. The algorithmic feed promoted conservative content and political activist accounts relative to the chronological feed. Some users consequently became more likely to follow conservative activist accounts. Those accounts remained in their follow networks after the feed was changed back to chronological ordering, so the later setting could not simply restore the earlier information environment.
The study authors’ overall conclusion was: “Overall, we conclude that social media feed algorithms can play an important role in shaping political opinions and online behaviour.” That conclusion applies to the tested intervention and population, not to every platform or every ranking system.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsA cross-country randomized Twitter study
A large randomized study that used a reverse-chronological control group found that algorithmic ranking amplified political content differently across countries and political groups. This demonstrates that ranking can affect exposure, but it does not establish that removing ranking would solve misinformation, toxicity, or polarization across all platforms. Older Twitter findings should not be presented as a description of the current X ranking system.
A diversified ranking design in 2026
A 2026 researcher-designed experiment tested a third approach instead of leaving engagement-based ranking unchanged or eliminating ranking altogether. Its diversified-extremity design reduced exposure to intergroup, moralized, emotional, and toxic content. It also improved the accuracy of participants’ perceptions of social norms while maintaining comparable platform enjoyment against the study’s alternatives.
This result supports testing specific ranking objectives. It does not show that one model can be copied unchanged onto every network, because ranking effects depend on the platform’s content, users, recommendation inventory, and implementation.
Rank #3
Study results side by side
| Study | Participants and setting | Feed change | Duration | Main measured results |
|---|---|---|---|---|
| Guess, Malhotra, Pan et al., Science (2023) | Consenting users in the United States during the 2020 election | Facebook and Instagram default feeds versus reverse chronology | Three months | Lower time and activity; more political and untrustworthy exposure on both platforms; less uncivil/slur-word content and more moderate, ideologically mixed sources on Facebook; no significant change in several political-attitude measures. |
| Gauthier, Hodler, Widmer et al., Nature (2026) | 4,965 active US X users | Algorithmic versus chronological feed | Approximately seven weeks | Algorithm activation increased engagement and shifted some policy priorities and current-event views in a conservative direction; switching it off did not comparably reverse those changes; no significant change in affective polarization or self-reported partisanship. |
| Large randomized Twitter study | Users in multiple countries and political groups | Algorithmic ranking versus reverse chronology | Study-specific | Political amplification differed by country and political group, showing that effects are not uniform. |
| Diversified-extremity experiment (2026) | Research participants assigned to alternative ranking designs | Diversified ranking versus tested alternatives | Study-specific | Lower intergroup, moralized, emotional, and toxic exposure; more accurate social-norm perceptions; comparable platform enjoyment. |
Would a chronological feed stop misinformation?
No. The Facebook and Instagram experiment found that chronological ordering increased exposure to political and untrustworthy content on both platforms. That may happen because a newest-first feed gives every followed account a more direct route into the timeline, including accounts that post low-quality or misleading material. Personalized ranking can also suppress some posts, but its objective may be engagement rather than truthfulness.
Free tools Windows power users keep installed
One-click scans. No signup required.
Chronology can improve one content measure while worsening another. On Facebook, participants encountered less content classified as uncivil or containing slur words, yet political and untrustworthy exposure rose. Therefore, “less algorithmic” is not the same as “more accurate,” “less toxic,” or “safer.” A platform must define the target outcome and measure it directly.
Would removing ranking reduce polarization?
The strongest evidence does not support a universal yes. The three-month Facebook and Instagram intervention did not significantly change issue polarization or affective polarization, and it did not significantly change political knowledge. The X experiment likewise found no significant change in affective polarization or self-reported partisanship, even though algorithm activation changed engagement and some policy views.
These null results have a precise meaning: the tested feed changes did not move those measured outcomes enough to reach statistical significance during the study periods. They do not prove that algorithms have no effect on other attitudes, behaviors, or groups. Conversely, a significant shift in selected policy views does not prove that ranking alone causes broad social polarization.
Why turning ranking off may not undo earlier effects
Follow networks persist
When a ranked feed repeatedly surfaces an account, a user may follow it. A chronological setting then displays the account’s posts in time order rather than deleting the account. The X experiment found that some users exposed to conservative activist accounts became more likely to follow them, and those follows remained after the feed switch.
Recommended Free Tools
Habits and attention persist
Ranking can alter which accounts a person notices, how often they open an app, and which topics they choose to pursue. A lower-engagement chronological feed may reduce total activity, but it cannot instantly reverse choices made during the earlier period. This is why a switch-off test is not always the mirror image of a switch-on test.
Exposure is only one part of influence
A ranking system can affect exposure without changing a broad attitude measure. People may see more political posts yet learn little, reject what they see, discuss it elsewhere, or change only a narrow policy priority. Researchers therefore need separate measures for exposure, engagement, knowledge, beliefs, social-network composition, and polarization.
What could platforms do instead of choosing “algorithm” or “no algorithm”?
The 2026 diversified-extremity result illustrates a more practical design question: what should the system optimize? Possible objectives include recency, predicted engagement, source or viewpoint diversity, safety, user controls, or a combination. Each can produce trade-offs.
| Ranking approach | Primary objective | Potential benefit | Known caution from the evidence |
|---|---|---|---|
| Reverse chronological | Show recent posts from followed accounts | Simple, predictable ordering and lower activity in the Meta experiment | More political and untrustworthy exposure occurred on both Facebook and Instagram. |
| Engagement-based personalization | Predict interactions such as clicks, replies, or viewing time | Higher engagement and stronger content relevance for some users | On X, it promoted conservative content and activist accounts relative to chronology and shifted selected views. |
| Diversified ranking | Balance relevance or engagement with exposure limits and content diversity | Reduced intergroup, moralized, emotional, and toxic exposure while maintaining comparable enjoyment in the 2026 experiment | Results from one design may not generalize unchanged to another platform. |
| User-selected controls | Let people choose ordering, topic limits, or recommendation intensity | Gives users more agency and permits different preferences | A setting change does not erase existing follows, habits, or moderation decisions. |
A credible redesign should publish the objective, identify which posts are eligible, measure unintended exposure, and test whether effects persist after the setting changes. It should not claim success merely because people spend more or less time in the app.
How to judge a proposal to remove algorithms
Use these questions before accepting a claim that chronology will make a platform healthier:
Best Value
- What's included in the book: - Practical strategies for disconnecting from screens and social media - Inspiring stories of others who have gone on this transformative journey - A path to spiritual renewal and clarity in a digital world - The sacred art of listening to the voice of God - Tips for reconnecting with what truly matters in life - Ways to foster deeper connections with God and others - Tools for living with intention and purpose in a world of technology - Guidance for finding balance
- What is being removed? Is the proposal about feed ordering, recommendations, moderation, notifications, or all of them?
- What content remains eligible? Does the feed contain only followed accounts, or are recommended posts still inserted?
- What outcome is the goal? Specify misinformation, incivility, toxicity, political exposure, knowledge, polarization, time spent, or enjoyment.
- Whose result is being reported? Check the country, platform, user population, political context, and sample size.
- How long was the intervention? A result after seven weeks or three months does not establish a permanent effect.
- Was the result measured after network changes? New follows and unfollows can preserve exposure after ranking changes.
- Were multiple outcomes tested? A significant result on one policy view alongside null results for partisanship and polarization requires a narrow interpretation.
- Did the platform cooperate throughout the test? Algorithms can change during a study, limiting how broadly its findings apply.
What users can do without waiting for a platform-wide change
Where a service offers them, users can select a following-only or latest-posts view, reduce recommended content, mute or unfollow accounts that repeatedly post unwanted material, and use lists or notifications for sources they deliberately choose. These controls can change an individual’s exposure, but they do not remove platform-wide ranking, alter other users’ feeds, or reverse follows already made.
It is also useful to treat a chronological feed as a preference and transparency tool rather than a guarantee of quality. Review the accounts in the feed, check original sources for consequential claims, and remember that newest-first ordering gives equal timing priority to reliable and unreliable posts.
Limits on what these experiments can prove
The studies differ in platform, year, ranking system, participant recruitment, duration, political environment, and outcome definitions. The Facebook and Instagram results describe a 2020 US election intervention, not every Meta user or every later product version. The X findings describe active US X users during a 2023 intervention and a platform that may change its ranking at any time. The cross-country Twitter study demonstrates variation across places and political groups, not a universal effect of removing ranking.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →A null finding for one attitude is not proof of no effect on another outcome. A statistically significant change in a selected view is not proof that algorithms alone caused a broad social harm. Strong causal claims should stay within the tested intervention, population, period, and measurement.
Further reading
Social Media Algorithms, edited by Adele Kuzmiakova and published by Arcler Education Incorporated in 2025, is a 361-page background book on algorithm design, recommendation systems, and societal effects. It is useful for readers seeking technical and social context, but it is not evidence that a chronological feed or any other particular intervention works.
The practical verdict
Removing personalized ranking can reduce engagement, reshape exposure, and sometimes change political views. It can also increase exposure to political or untrustworthy posts, leave social-network changes intact, and fail to move polarization or knowledge. The better question is not whether algorithms are inherently good or bad, but which ranking design produces which measurable outcomes for which users—and whether the platform is willing to test those trade-offs transparently.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
