Rules
Part of Online status and identity: five concepts explained
Online status and identity statistics are instrument readings, not proof
Online status and identity statistics are instrument readings: what a counter counts, why it stops describing anything, and why comparisons fail.
What to take away
- A follower count on Instagram, a view count on YouTube or TikTok, a like count or a rank badge is a reading from an instrument the platform designs, calibrates and can redefine. Online status and identity statistics are those readings, not facts about a person.
- No platformMeta, YouTube and TikTok included — publishes the full definition behind those counts, so two accounts' numbers cannot be compared as if one ruler produced both.
- The figures in this field that carry a value, a named publisher and a date are regulatory rather than counter-based. They are listed below.
- Display a measure and people optimize it. Once that happens, it stops describing what it was built to describe.
- If you need a series rather than a reading, keep your own dated log. The platform's version of your history changes when its definition does.
The dated figures that survive checking
Counters arrive with no published definition and no audit. The checkable paper trail in this field is legal, and it marks the moments when platforms changed what they show and to whom.
| Item and named source | Value | Date | What it records |
|---|---|---|---|
| Federal Trade Commission penalty on Facebook | $5 billion | July 2019 | Privacy failures found in an investigation |
| Facebook profiles reached through a third-party app, company disclosure | up to 87 million | April 2018 | Profiles in a data pull, not engagement |
| Section 230, Communications Decency Act, US Congress | enacted | February 1996 | A liability shield for interactive services |
| Children's Online Privacy Protection Act, US Congress | enacted 1998, rule effective April 21, 2000 | 1998 to 2000 | Children under 13 as a legal category |
| California Age-Appropriate Design Code, AB 2273 | signed, then blocked by a federal court | September 2022 and September 2023 | Design duties toward younger users |
| Online Streaming Act, Parliament of Canada | royal assent | April 2023 | Regulatory scope over streaming and uploaded content |
| Online News Act, Parliament of Canada | royal assent, platform news blocking followed | June 2023 | Link sharing between platforms and publishers |
Not one row is a follower count, a view count or a rank. Those numbers exist, but no service publishes the definition behind them, so no comparable figure can be quoted for any of them.
The rows above are checkable because their publishers are named: the Federal Trade Commission, Congress, the California Legislature and the Parliament of Canada.
The $5 billion penalty is quotable because a named agency published it on a known date, and you can read the FTC announcement yourself.
Two of those rows govern what US apps may show younger users. The compliance detail, and the difference between the two rules, is set out in COPPA and California's design code.
Why a counter is a reading from an unnamed instrument
Every visible number rests on a definition the service keeps private.
An Instagram or X follower count counts accounts in a relation, including dormant ones, duplicates and automated ones. A YouTube view counts an event whose threshold is a design decision about duration and repeats, and YouTube has moved that threshold more than once. Reactions are taps, and cheapness is the point of them. A rank counts a formula, the least stable of the four.
None of that is dishonest. Each is a real count of a defined event. But the definition can move without the number's name changing: YouTube has rewritten what counts as a view, and Instagram has hidden public like counts from many users. The units shift under a number that looks unchanged.
That is the ordinary problem of construct validity, the term the psychometricians Lee Cronbach and Paul Meehl gave to the gap between what a measure counts and what readers assume it counts. Which of these signals is worth reading at all is worked through in online status and identity.
Example: two accounts, one number, no common ruler
Take two Instagram accounts displaying the same follower count. One gathered it when a post traveled. The other bought reach, and the platform suggested an audience to it. The displayed value matches; the instruments do not.
Add a selection problem. The accounts whose numbers you see are the ones the ranking chose to surface, so they are a sample of what performed rather than a sample of accounts. Conclusions about typical outcomes drawn from them are survivorship bias at work — the statistician Abraham Wald's question about which aircraft never came back, run on a feed.
What optimization does to a counter
The reader is not the only thing that changes. Once a count is public, people act on it, and acting on it changes the behavior the count was measuring. That is Goodhart's law — the economist Charles Goodhart's rule that once a measure becomes a target it stops being a good measure — running on a live display.
The table below sets out what four counters measured before anyone optimized them and what they measure after.
Counters before and after optimization
Counter
- Followers
- Rough interest
- Views
- Whether shown
- Reactions
- Weak approval signal
- Rank or level
- Standing among peers
Before optimization
- Followers
- Trade willingness, history, luck
- Views
- Matched ranking rewards
- Reactions
- Social obligation to reciprocate
- Rank or level
- Time spent, mostly
After optimization
- Followers
- Views
- Reactions
- Rank or level
What it measured before anyone optimized it
- Followers
- Rough interest in a person
- Views
- Whether something got shown
- Reactions
- A weak signal of approval
- Rank or level
- Standing among peers
What it measures once everyone does
- Followers
- Willingness to trade a follow, plus history, plus luck
- Views
- How well it matched what the ranking rewards
- Reactions
- A social obligation among people who reciprocate
- Rank or level
- Time spent, mostly
The last row is the honest summary of most rank systems: they measure persistence. Adding scored progress to an activity that had none is the subject of the literature on gamification. Persistence is a real thing to measure. It is not what a rank badge is read as meaning.
Of the four, followers and views sit closest to signal, because both are expensive to manufacture at scale on a platform that polices fake accounts. Reactions are the cheapest and the noisiest, and rank is the one most directly targeted by the people it measures.
Keeping your own dated log
- Fix one day each month and one place to store the record, such as a document, a notes app or a spreadsheet.
- Write the raw countersfollowers, views on a fixed recent item, reactions, rank.
- Add the date and anything the platform announced that month about how counts work.
- Append. Never overwrite an earlier entry.
- Compare definitions first when you compare two entries, then the numbers.
What you end up with is a series measured by an instrument that changed underneath it. That is imperfect, and it still beats one current reading with no history. If the comparison habit is affecting how you read your own standing, the case is set out in critiques of social media.
Common questions
Do the figures above describe my account?
No. They describe rules, penalties and enforcement dates. They matter because they mark when a platform changed what it displays and to whom.
Is a large follower count meaningless?
No. It means something happened at scale. It does not mean the specific thing readers assume, and it cannot be compared with another account's count.
Should I stop reading my own counters?
That is your decision. The useful version is to record them and watch the series instead of the single reading.
Why will no platform publish the definition?
A published definition of a scored quantity invites direct optimization against it. Definitions also change often enough that any published version ages fast.
