# What does algorithm mean on social media?

Canonical: https://postmesh.app/glossary/algorithm

By PostMesh team. Published 2026-10-31; updated 2026-10-31.

On social media, the algorithm is the set of automated systems a network uses to choose and rank the posts each person sees, based mostly on what that person watches, likes, shares and follows. Each app, and often each feed inside an app, uses its own.

## What algorithm means

In general, an algorithm is a set of steps for solving a problem. Merriam-Webster's first definition is a specific step-by-step procedure for solving a problem or accomplishing some end. Its second sense is the one people mean online: a computerized process that evaluates data to select or rank things without a human checking each case, and especially the hidden system a platform uses to curate content for each user based on their behavior and profile.

When someone says "the algorithm" on social media, they usually mean whatever decides what shows up in their feed, and whether their own posts get seen. "The algorithm hates me" is a complaint about reach. "The algorithm knows me too well" is about a feed that seems to read minds. Merriam-Webster even lists "the algorithm" and "game the algorithm" as phrases in their own right.

The singular is a simplification. Instagram says it doesn't have a single algorithm deciding what people see. It uses a variety of algorithms, classifiers and processes, and each part of the app, from Feed and Stories to Explore, Reels and Search, has its own. The same is true elsewhere: a network's home feed, its search and its recommendations are usually ranked differently.

## Where the word comes from

The word is named after a person. Merriam-Webster traces algorithm through Middle English algorisme and Medieval Latin algorismus to the Arabic name al-Khwārizmī, an Islamic mathematician who flourished around AD 825. It dates the first known use of algorithm in the step-by-step sense to 1811.

- Around AD 825: The mathematician al-Khwārizmī, whose name gives us the word
- 1811: First known use of algorithm as a step-by-step procedure
- June 18, 2020: TikTok explains how its For You feed recommends videos
- March 31, 2023: Twitter publishes the code behind its recommendations
- May 31, 2023: Instagram publishes its updated ranking explainer

*Sources: Merriam-Webster, TikTok Newsroom, X Engineering blog and Instagram's blog.*

## How social media algorithms work

The networks describe the same basic process in different words. Instagram's explainer and Twitter's 2023 engineering post both break ranking into stages.

1. Gather candidates: recent posts from accounts you follow, plus posts from accounts you don't
2. Read the signals: your activity, the post, the person who posted, your history with them
3. Predict: how likely you are to watch, like, comment, share or follow
4. Rank: order posts by those predictions, weighted by what matters most
5. Filter and balance: remove rule-breaking or blocked content and avoid repeats

*Based on Instagram Ranking Explained and Twitter's Recommendation Algorithm post.*

The signals are the part people care about, and the networks have said a fair amount about theirs. What follows is each one's own explanation.

## Instagram's algorithm

In Instagram Ranking Explained, [Instagram](/platform/instagram)'s Adam Mosseri lists the most important Feed signals, roughly in order: your activity, such as posts you've liked, shared, saved or commented on; information about the post, including how quickly people are engaging with it; information about the person who posted; and your history of interacting with them. Instagram then predicts how you'll react. In Feed, the five predictions it watches most closely are whether you'll spend a few seconds on a post, comment, like, share, or tap the profile photo.

Other surfaces weigh things differently. In Explore, the key predicted actions are likes, saves and shares, and Instagram says popularity signals matter much more there than in Feed or Stories. For [Reels](/glossary/reel), it looks most at how likely you are to reshare, watch to the end, like, and go to the audio page. It also says it makes some reels less visible, including low-resolution or watermarked ones, reels with borders or that are mostly text, and reels already posted on Instagram.

You have some control. Instagram points to Favorites, a Following view that shows the last 30 days of posts from accounts you follow in time order, snoozing suggested posts for 30 days, marking recommendations Not interested, the Sensitive Content Control, and the Why you're seeing this post option in a post's menu. It also says it doesn't suppress content to push people into buying ads, and addresses [shadowban](/glossary/shadowban) worries directly in the same post.

## TikTok's algorithm

[TikTok](/platform/tiktok)'s help center says its recommendations draw on three groups of factors: user interactions, such as what you like, share, comment on, and watch in full or skip; content information, such as sounds, hashtags, view counts and the country where it was posted; and user information, such as language, location and device. For the For You feed, it says user interactions, including time spent watching, are generally weighted more heavily than the rest.

Its 2020 newsroom post added useful detail. It said a strong signal, like finishing a longer video, counts for more than a weak one, like sharing a country with the creator, and that neither follower count nor a history of hit videos is a direct factor, even though bigger accounts tend to get more views. TikTok also says it tries to diversify your feed with new creators and topics, generally avoids showing you videos you've already seen, and doesn't recommend content that misses its For You feed eligibility standards. Our [For You page](/glossary/fyp) entry goes further, and our guide to [why a TikTok isn't getting views](/blog/why-is-my-tiktok-not-getting-views) covers what to check.

## YouTube's algorithm

[YouTube](/platform/youtube) describes its recommendation system as trying to anticipate what a viewer wants and give them a satisfying experience. It says it uses viewing behavior, likes, dislikes, subscriptions and feedback, including satisfaction surveys, and that it also considers a channel's reputation and quality, with external evaluators helping it judge how an average viewer would see content. Recommendations show up mainly on the Homepage and in the Up Next panel.

Search works a little differently. YouTube says search results are ranked on relevance, engagement and quality, with the balance shifting by the type of search, and that for topics like news, health and personal finance it gives prominence to high-quality sources. You can pause, edit or delete your watch and search history, which changes what it recommends.

## X's algorithm and its open source code

[X](/platform/twitter) has gone furthest in showing its work. On March 31, 2023, Twitter's engineering blog published a walkthrough of its recommendation algorithm and released the code. It said the system distills roughly 500 million Tweets a day down to a handful on your For You timeline, pulls about 1,500 candidates for each request, ranks them with a neural network of about 48 million parameters trained to predict positive engagement such as likes, reposts and replies, and then filters out things like posts from blocked accounts. At the time, it said the For You timeline averaged 50% posts from accounts you follow and 50% from accounts you don't.

- 500 million: Tweets a day the 2023 system narrowed to a handful
- 1,500: candidate posts pulled for each timeline request
- 50%: average share of For You from accounts you don't follow, in 2023

*From Twitter's Recommendation Algorithm, X Engineering blog, March 31, 2023.*

X keeps publishing. Its current open source repository describes the code that decides what appears in the For You feed: it combines posts from accounts you follow with posts found through machine learning, filters them, and ranks them with a transformer model. Its notes from August 2026 correct a common misreading of the published weights. They scale your own predicted chance of an action, such as liking or reporting a post, not raw counts, so one report doesn't simply cancel out a fixed number of likes. X's help center adds that you can switch to a Following timeline that shows only accounts you follow. If you want to find posts directly instead of waiting for the feed, [X Advanced Search](/glossary/x-advanced-search) does that.

## What the networks say matters most

| Network | Signals it emphasizes | Controls it offers |
| --- | --- | --- |
| Instagram | Your activity, the post's popularity, your history with the poster | Favorites, Following, Not interested, snooze suggestions |
| TikTok | Likes, shares, comments, watching in full or skipping | Not interested, refreshing your For You feed |
| YouTube | Viewing behavior, likes, dislikes, subscriptions, surveys | Pause, edit or delete watch and search history |
| X | Predicted engagement: likes, reposts, replies | Switch to the Following timeline |

Notice what's common. Every network leans most heavily on what people actually do with posts: watching, liking, sharing, commenting. That's why the advice that holds up is plain: make things your audience genuinely wants to watch, read and share, and judge results with something like [engagement rate](/glossary/engagement-rate) rather than raw follower counts. Chasing any single signal is risky, since Instagram says it adds and removes signals and predictions over time.

> If you cross-post, note that Instagram says it makes reels already posted on Instagram, and watermarked reels, less visible. Our guide on [whether cross-posting hurts engagement](/blog/does-cross-posting-hurt-engagement) goes through each network's rules.

## Questions

### What does "the algorithm" mean on TikTok or Instagram?

It's shorthand for the systems that decide which posts appear in your feed and which of your posts get shown to others. Each app, and each feed within it, has its own.

### Does follower count affect the algorithm?

Not directly on TikTok, which says follower count isn't a direct factor. Instagram does consider information about the poster, such as how often people have interacted with them recently.

### Can you see why a post was recommended to you?

On Instagram, open the post's menu and tap Why you're seeing this post. TikTok and YouTube let you tell them you're not interested, which changes future recommendations.

### Is the X algorithm really open source?

X has published the core code for its For You recommendations, first in March 2023 and in an updated repository since, along with notes on how the ranking weights work.

### Can I turn the algorithm off?

Partly. Instagram has a Following view in time order, X has a Following timeline, and YouTube lets you pause watch history. Recommendation feeds like For You can't be fully switched off.

### Where does the word algorithm come from?

From the name of al-Khwārizmī, a mathematician who flourished around AD 825. Merriam-Webster dates the step-by-step sense of the word in English to 1811.

## Related terms

- [FYP](https://postmesh.app/glossary/fyp)

- [Shadowban](https://postmesh.app/glossary/shadowban)

- [Engagement rate](https://postmesh.app/glossary/engagement-rate)

- [Reel](https://postmesh.app/glossary/reel)

- [X Advanced Search](https://postmesh.app/glossary/x-advanced-search)

- [Short-form video](https://postmesh.app/glossary/short-form-video)

## Sources

- [Merriam-Webster: algorithm (Wayback Machine copy, October 2026)](https://web.archive.org/web/20261015055508/https://www.merriam-webster.com/dictionary/algorithm)

- [Instagram: Instagram Ranking Explained, by Adam Mosseri (May 31, 2023; Wayback Machine copy, October 2026)](https://web.archive.org/web/20261021150653/https://about.instagram.com/blog/announcements/instagram-ranking-explained)

- [TikTok Help Center: How TikTok recommends content (Wayback Machine copy, June 2026)](https://web.archive.org/web/20260601141220/https://support.tiktok.com/en/using-tiktok/exploring-videos/how-tiktok-recommends-content)

- [TikTok Newsroom: How TikTok recommends videos #ForYou (June 18, 2020; Wayback Machine copy, October 2025)](https://web.archive.org/web/20251005094804/https://newsroom.tiktok.com/en-us/how-tiktok-recommends-videos-for-you)

- [How YouTube Works: Recommendations (Wayback Machine copy, October 2026)](https://web.archive.org/web/20261021190039/https://www.youtube.com/howyoutubeworks/recommendations/)

- [X Engineering: Twitter's Recommendation Algorithm (March 31, 2023; Wayback Machine copy, March 2026)](https://web.archive.org/web/20260310113313/https://blog.x.com/engineering/en_us/topics/open-source/2023/twitter-recommendation-algorithm)

- [GitHub: xai-org/x-algorithm, X For You Feed Algorithm (Wayback Machine copy, October 2026)](https://web.archive.org/web/20261009201226/https://github.com/xai-org/x-algorithm)

- [X Help Center: About your For you timeline on X (Wayback Machine copy, September 2026)](https://web.archive.org/web/20260929015103/https://help.x.com/en/using-x/x-timeline)