AI tools helped gather research and write the first draft of this article. I edited it and checked every study, figure and quotation against the paper's official record or full text, or the organisation's own page, listed at the end. The cover illustration is AI-generated. Details at the end.
When people look for someone who knows a subject, the first number they see is the follower count. It is large, public and easy to compare. An account with 400,000 followers looks more authoritative than one with 6,000, before anyone has read a word either of them wrote.
That reading mixes up two different things. Audience size says how many people, at some point and for some reason, chose to see an account's posts. Topic authority is narrower: whether this person reliably knows this subject, so that others in the field rely on what they say about it. The two can coincide. Often they do not.
This article asks one question: what does audience size tell you about someone's authority on a specific topic, and what should you look at instead? It looks at a large study that compared follower counts with other measures of influence, at research that measures influence topic by topic, at how Google describes expertise, and at why follower counts are a weak signal even before anyone asks about topics.
This is the ninth part of Influence & Incentives, a series on how other people, rules and rewards shape what we decide, write and publish, and the fourth in its short run about authors and creators. The previous parts asked what to count beyond output, how rewards change what gets published and how creator metrics get gamed. This part is only about one common mistake: reading reach as expertise. How to build an expert ranking that can be explained, and how to compare newcomers with established creators fairly, are questions for the next two parts.
What a follower count actually records
A follower count is the running total of decisions to follow. Each of those decisions was made at a different time, for a different reason. Some followers came for one viral post years ago. Some follow hundreds of accounts and read almost none of them. Some came for jokes, some for news, some for the subject the account is known for today. Some are no longer active, and some may never have been people.
All of that is compressed into one number for the whole account. The number has no topic. An account that writes about football, politics and cryptocurrency carries the same count into every one of those subjects. It also has no date: a count built up in 2018 looks the same as one built up last month.
So the count answers a real question, how many people might see the next post, but not the question people usually ask it: does this person know what they are talking about here?
The million follower fallacy
In 2010 Meeyoung Cha, Hamed Haddadi, Fabrício Benevenuto and Krishna Gummadi published a study with a direct title: "Measuring User Influence in Twitter: The Million Follower Fallacy" [1]. Their dataset covered 54 million users connected by about 2 billion follow links, who had produced 1.7 billion tweets [1]. They focused on about 6.2 million active users and compared three measures of influence: the number of followers, the number of times a user was retweeted, and the number of times a user was mentioned by others [1].
The main finding is in the abstract: "popular users who have high indegree are not necessarily influential in terms of spawning retweets or mentions" [1]. Indegree is the network term for the number of followers. The authors concluded that "indegree alone reveals very little about the influence of a user" [1].
The paper also shows who rises on each measure. The most followed accounts were news sources, politicians, athletes and celebrities. The most retweeted were content aggregators, business figures and news sites, which the authors described as trackers of trending topics and knowledgeable people [1]. The top users by the three measures had "little overlap" [1]. The authors' summary was that followers represent a user's popularity, retweets the content value of their tweets, and mentions the name value of the user [1].
Popularity, the value of what someone writes and how often others name them are three different things. A follower count measures only the first.
Big audiences help a message spread, and still predict poorly
It would be wrong to say audience size means nothing. A year later Eytan Bakshy, Jake Hofman, Winter Mason and Duncan Watts tracked 74 million diffusion events among 1.6 million Twitter users over two months in 2009 [2]. They found, "unsurprisingly", that the largest cascades of sharing tended to start with users who had been influential in the past and who had many followers [2].
But they also found that predicting which user or link would start a large cascade was "relatively unreliable" [2]. In some cases the biggest accounts were also the most cost-effective, but under a wide range of assumptions, the most cost-effective way to spread a message in their models was to work with many "ordinary influencers", people of average or even below-average influence, rather than a few big accounts [2].
This is a finding about spreading, not about knowing. It says a large audience raises the chance that a message travels. It says nothing about whether the message was right. That is the distinction this article is about: reach is a property of the audience, authority is a property of the person's work on a subject.
Authority belongs to a topic
If authority is about a subject, it has to be measured per subject. Researchers have done exactly that for a long time.
In 2010 Jianshu Weng, Ee-Peng Lim, Jing Jiang and Qi He proposed TwitterRank, a measure of influence that takes into account "both the topical similarity between users and the link structure" [3]. In their experiments it performed better than the method Twitter then used and than the standard PageRank algorithm [3]. The detail of the formula matters less than the design choice: the researchers did not accept a single influence score for an account. They computed influence within topics.
Cha and her colleagues also looked at topics, and their result needs careful reading. They followed three major subjects of 2009 for 60 days each: the Iranian presidential election, the H1N1 flu outbreak and the death of Michael Jackson [1]. They found that "most influential users can hold significant influence over a variety of topics" [1]. That can sound like the opposite of this article's argument. It is not quite. Those were three huge news events that everyone was talking about, and the measure was retweets and mentions, not whether anyone was right. A large account can move attention on many topics at once. That is reach working as designed, and it is exactly why reach is easy to mistake for expertise.
The same study offers the other half. When the authors inspected users whose influence grew most, they found that "users who limit their tweets to a single topic showed the largest increase in their influence scores" [1]. Their conclusion was that ordinary users can gain influence "by focusing on a single topic and posting creative and insightful tweets that are perceived as valuable by others" [1]. Focus, not size, was the route for people without a large audience.
Where topic authority shows up instead
If the follower count does not show authority, what does? The useful signals have one thing in common: they come from other people's judgements about the subject, not from the size of the crowd.
How others file you. In 2012 Naveen Sharma and colleagues built a system that inferred Twitter users' topics of expertise from the names and descriptions of the public Lists that other users had put them in [4]. A List called "climate scientists" or "rollup researchers" is a small, deliberate judgement by one reader about what an account is for. Many such judgements together say more about topic than a follower total, which records only that someone clicked "follow".
What Google asks. Google's guidance for site owners includes a self-assessment question: "Is this content written or reviewed by an expert or enthusiast who demonstrably knows the topic well?" [5]. It groups such signals under E-E-A-T, experience, expertise, authoritativeness and trustworthiness, and states that "trust is most important" [5]. Google also says E-E-A-T itself is not a specific ranking factor [5]. Note the word demonstrably: the question is about evidence in the work on a topic, not the size of the author's following.
For an editor or a reader, those ideas translate into a short list of things to look for:
- Share of work on the topic. Does most of what this person publishes deal with the subject, or is it one theme among twenty?
- Specific, checkable claims. Do they give figures, sources and mechanisms that can be verified, or mostly opinions and predictions?
- A record over time. Have their past statements on the topic held up? When they were wrong, did they correct it in public?
- Recognition from people in the field. Do others who work on the subject cite them, link to them or argue with them seriously? Who follows matters more than how many.
- Answers to hard questions. In replies and comments, do they engage with detailed questions on the subject, or only with praise?
None of these is a number you can read off a profile in a second. That is the point. Topic authority takes effort to observe because it takes effort to earn.
Size can also be bought
There is a second reason to treat follower counts carefully, before topics come in at all: the number can be manufactured.
In January 2019 the New York Attorney General announced a settlement with Devumi and related companies, which had sold "fake followers, 'likes,' views and other forms of online endorsement and activity" produced by bot and sock-puppet accounts [6]. The office described the settlement as establishing that "selling fake social media engagement and using stolen identities to engage in online activity is illegal" [6].
In August 2024 the US Federal Trade Commission announced a final rule on fake reviews that also covers social media. It "prohibits anyone from selling or buying fake indicators of social media influence, such as followers or views generated by a bot or hijacked account" [7]. The ban applies when the buyer knew or should have known the indicators were fake and they misrepresent the buyer's influence or importance for a commercial purpose [7].
The fact that regulators took these actions says something about how much weight people put on the count. How to spot gaming in creator metrics was the subject of the previous part. Here the point is simpler: even an honest follower count measures only reach, and the count you see may not be honest.
A worked example (fictional)
Both accounts and all figures below are invented to show the difference in use.
An editor needs someone to review an article about transaction fees on Ethereum rollups. Two candidates come up.
Account A has 400,000 followers. It posts about twenty times a day on crypto news in general: prices, exchange listings, regulation, memes. In the last three months it mentioned rollup fees in four posts, all of them links to other people's articles. No one working on rollups replies to it. Its posts are widely shared.
Account B has 6,000 followers. Almost everything it has written in two years is about rollup fees and data costs. Its posts include tables, links to the code and explanations of how fees are calculated. Two developers of rollup projects regularly reply to it and sometimes disagree. Last year it published a correction to one of its own fee estimates and explained what it had got wrong.
On audience size, A is about sixty-seven times larger. For spreading the finished article, A may well be more useful. For reviewing it, B is the obvious choice. The evidence of topic authority, focus, checkable work, peer recognition and a public correction, sits almost entirely with the smaller account. Neither account is "better" in general. They are good at different jobs, and the follower count only describes one of them.
Five questions before calling someone an authority
- On which topic? Authority is about a subject. If you cannot name the subject, you are looking at popularity.
- How much of their work is on it? A few posts among thousands is interest, not depth.
- Can you check what they say? Look for sources, figures and mechanisms, not only conclusions.
- Who in the field takes them seriously? Citations, links and serious disagreement from people who work on the subject count for more than the total.
- What happened when they were wrong? A public, specific correction is evidence of care. Silence is not evidence of being right.
What this article cannot tell you
The studies cited here used Twitter data from 2012 or earlier, most of it collected in 2009, when the service, its users and its feed worked very differently from today's X; they show a pattern, not current figures. Their measures were retweets, mentions and diffusion, which are signals of attention, not of being right; no study here measured expertise directly. The Lists-based method and TwitterRank are research systems, described from their official abstracts. Google's guidance describes what Google says it values, not how its ranking works in detail. The two legal actions show that follower counts can be bought, not how common that is today. The worked example is invented. Nothing here assesses any real author or account, and this part does not propose a ranking formula or a way to compare newcomers with established creators; the next parts of the series take those up.
How this article was made
| Task | What the AI did | What a person did | How it was checked |
|---|---|---|---|
| Research | Collected studies and rules on influence and follower counts into research notes | Chose the question and which sources to rely on | Figures and quotations checked against the full paper, official paper records and the organisations' own pages [1]–[7] |
| Draft | Wrote the first draft from the notes | Edited the text, the structure and the worked example | A separate review compared figures and claims with the sources |
| Images | Generated the cover photograph-style illustration | Chose the illustration; approved the diagrams, which were drawn in code from the article's figures | Labelled as AI-generated in the disclosure line and alt text |
| Not done by AI | — | The decision to publish and responsibility for the text | — |
Conflicts of interest: none. The author has no commercial relationship with the researchers, companies or agencies named; the accounts in the example are fictional.
Responsible for this article: Dmitriy Lastovka, author. Corrections: contact form.
Sources
Labels: paper (peer-reviewed study, read via its full text or official record), official (the organisation's own page or announcement).
- Meeyoung Cha, Hamed Haddadi, Fabrício Benevenuto, Krishna P. Gummadi, "Measuring User Influence in Twitter: The Million Follower Fallacy", Proceedings of the International AAAI Conference on Web and Social Media (ICWSM), 2010. https://doi.org/10.1609/icwsm.v4i1.14033 Paper, full text and official record.
- Eytan Bakshy, Jake M. Hofman, Winter Mason, Duncan J. Watts, "Everyone's an influencer: quantifying influence on Twitter", Proceedings of the Fourth ACM International Conference on Web Search and Data Mining (WSDM 2011). https://doi.org/10.1145/1935826.1935845 Paper, official record and abstract.
- Jianshu Weng, Ee-Peng Lim, Jing Jiang, Qi He, "TwitterRank: finding topic-sensitive influential twitterers", Proceedings of the Third ACM International Conference on Web Search and Data Mining (WSDM 2010). https://doi.org/10.1145/1718487.1718520 Paper, official record and abstract.
- Naveen Kumar Sharma, Saptarshi Ghosh, Fabrício Benevenuto, Niloy Ganguly, Krishna P. Gummadi, "Inferring who-is-who in the Twitter social network", ACM SIGCOMM Workshop on Online Social Networks (WOSN), 2012. https://doi.org/10.1145/2342549.2342563 Paper, official record and abstract.
- Google Search Central, "Creating helpful, reliable, people-first content", read 15 November 2026. https://developers.google.com/search/docs/fundamentals/creating-helpful-content Official.
- New York State Attorney General, "Attorney General James Announces Groundbreaking Settlement With Sellers Of Fake Followers And 'Likes' On Social Media", 30 January 2019. https://ag.ny.gov/press-release/2019/attorney-general-james-announces-groundbreaking-settlement-sellers-fake-followers Official.
- Federal Trade Commission, "Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials", 14 August 2024. https://www.ftc.gov/news-events/news/press-releases/2024/08/federal-trade-commission-announces-final-rule-banning-fake-reviews-testimonials Official.










