Maintenance
Part of How internet communities change: a reading framework
Internet communities comparison in plain terms
An internet communities comparison on five observable axes: the membership gate, governance, moderation labor, exit cost and who owns the container.
People compare online communities constantly, and almost always on the wrong axes. Size, activity and health are the three that get used, and none of them means anything without a definition nobody supplies. The properties that determine what a place is like to be in are rarely mentioned, because they are less visible from outside.
An internet communities comparison in plain terms uses five checkable axes instead: the membership gate, governance, moderation labor, exit cost and who owns the container. Reddit, Discord and English Wikipedia differ on all five. So do Stack Overflow, MetaFilter and Hacker News. 4chan and Nextdoor break the pattern in their own ways.
This page sets out what can be compared honestly, what cannot, and how to run a comparison whose conclusion would change if the facts were different.
What to take away
- Comparablethe membership gate, the governance model, where moderation labor falls, the exit cost, and who owns the container. MetaFilter charges $5 once to join; English Wikipedia lets anyone edit without an account.
- Not comparablesize, activity, health, and anything requiring a count of people.
- Most published comparisons are ranking two places by how much the writer liked them, with the axes chosen afterward.
- A comparison is only useful if it names the decision it is meant to inform. Where to post a question and where to move a group weight the axes differently.
The five axes that work
Axis / The question
- Membership gate
- What must someone do to post? · You can attempt it · Arrival rate, and how much norm-teaching the place has to do
- Governance
- Who decides, and how are they chosen? · Usually published, and the gap between the published version and practice is itself visible · How conflicts end, and whether they end
- Moderation labor
- Who does the work, and are they paid? · Visible from who acts and how fast · Whether the place survives a surge of attention
- Exit cost
- What do you lose by leaving? · Enumerable: content, connections, standing, name · How much dissent the place can absorb before people leave
- Ownership of the container
- Who can close it, and who holds the archive? · A fact about the arrangement · Whether the community can outlive a disagreement with its host
Filled in for named places, the same five axes produce a table you can argue with instead of a ranking.
| Community | Membership gate | Governance | Moderation labor | Exit cost | Who owns the container |
|---|---|---|---|---|---|
| Free account, then each subreddit's own rules | Subreddit moderators set rules; Reddit staff can ban or quarantine a subreddit | Unpaid volunteer moderators | Karma and posting history stay behind | Reddit Inc., public as RDDT on the NYSE since March 2024 | |
| Discord | An invite link from someone inside; no public directory | The server owner holds final say and assigns roles | Unpaid volunteer moderators; Discord staff handle platform reports | A server's history stops being visible to you once you leave | Discord Inc. |
| English Wikipedia | No account needed to edit; starting an article needs four days and ten edits | Editor-written policies; administrators elected at Requests for Adminship; Arbitration Committee elected yearly | Volunteer editors; Wikimedia Foundation staff do not write articles | Low: edit history and user page remain, and the CC BY-SA license allows a fork | Wikimedia Foundation |
| Stack Overflow | Account to post; reputation gates privileges at 15, 125, 2,000 and 10,000 points | Moderators elected by the community; Stack Exchange staff hold final say | Unpaid elected moderators; review queues open at 500 reputation | Reputation and badges do not transfer; posts stay under CC BY-SA | Stack Exchange Inc., bought by Prosus in 2021 |
| MetaFilter | $5 one-time signup fee | Paid moderators apply published guidelines; a member-led group bought the site from founder Matt Haughey in 2017 | Paid staff moderators | $5 and your posting history | MetaFilter Network LLC |
| Hacker News | Free account, no email required; karma gates comment downvoting at 501 and flagging at 31 | Y Combinator employees moderate; no elected moderators | Paid Y Combinator staff | Karma and comment history, little else | Y Combinator |
| 4chan | No account; a captcha and a name field | Volunteer janitors and moderators are appointed; the owner has final say | Unpaid volunteers | Almost none; old threads are pruned | Hiroyuki Nishimura, who bought the site in 2015 from founder Christopher Poole |
| Nextdoor | Verified legal name and home address | Unpaid Neighborhood Leads apply guidelines; Nextdoor staff handle reports | Unpaid volunteer leads | Your verified account and local feed | Nextdoor Holdings Inc., public as KIND on the NYSE |
MetaFilter charges $5 once to join and pays its moderators. Reddit and Discord lean on unpaid volunteers, and so do Nextdoor's Neighborhood Leads and 4chan's janitors.
Discord has no public directory, so entry runs through an invite from someone already inside. 4chan needs no account at all, and its threads expire, which drops exit cost to almost nothing.
Stack Overflow sorts arrival by a reputation score. Upvoting opens at 15 points, downvoting at 125, editing another person's post at 2,000, and moderator tools at 10,000. English Wikipedia sorts arrival by patience: four days and ten edits before an account can start an article.
Ownership decides what survives a fight with the host. English Wikipedia's text is licensed CC BY-SA, so a dispute with the Wikimedia Foundation can end in a fork. A subreddit banned by Reddit staff does not get forked; it becomes inaccessible.
Two of the axes have a general literature behind them worth borrowing from. Exit cost is the ordinary economics of switching barriers, and moderation labor is a special case of who does unpaid work in a shared space, treated at length under content moderation.
Five axes of comparison
Axis
- Membership gate
- What must you do to post?
- Governance
- Who decides, how chosen?
- Moderation labor
- Who works, are they paid?
- Exit cost
- What do you lose leaving?
- Container ownership
- Who can close it?
Question
- Membership gate
- Arrival rate, norm-teaching
- Governance
- How conflicts end
- Moderation labor
- Survives attention surges
- Exit cost
- Dissent absorbed before leaving
- Container ownership
- Outlives host disagreement
What it predicts
- Membership gate
- Governance
- Moderation labor
- Exit cost
- Container ownership
Each is checkable without asking anyone for a number, and each changes daily life. Exit cost is skipped most often, and it explains the most.
A community nobody can afford to leave talks differently from one people can walk out of. Neither is plainly better. What is lost on the way out is set out under online status and identity.
Why size cannot be compared
A membership figure counts accounts that once registered. It includes people who left, people who never posted, duplicates and abandonments, and it does not include the people reading without an account, who are usually the majority.
Two places reporting the same figure can have wildly different numbers of humans participating. A subreddit's subscriber count and a Discord server's member count can land in the same spreadsheet column while measuring different things.
Active membership is worse, because "active" is a definition each place makes for itself, out of events it defines, over a window it chooses. The reasons a count of activity cannot travel between systems are the same ones set out under benefits of social media.
There is also a deeper problem: size is not a property that ranks. A place of thirty and a place of thirty thousand are not the same thing at different scales, but different kinds of thing.
The thirty know each other. The thirty thousand cannot, so they need rules, roles and enforcement, and those are the things worth comparing instead.
Why health cannot be compared
"Healthy" smuggles in a purpose. A community optimized for newcomer support looks unhealthy to someone who values depth, and one optimized for expertise looks hostile to someone who values access. Both readings are correct given their premise, and the premise is what is being argued about.
Ask instead what the place is for, and whether its structure serves that. A place for beginners with a high membership gate is misconfigured. A place for specialists with no gate will spend all its time on basics and lose the specialists.
That is a comparison with a fact of the matter in it, and it does not require anyone to agree about health. The families of complaint that get filed under this heading are sorted under critiques of social media.
A comparison procedure
- Name the decision.Where should I ask this question, where should I spend my evenings, where should we move the group. Different decisions weight the axes differently, and without one there is no basis for weighting at all.
- Fill the five axes for each candidate,from observation, not from the description on the front page. A signup form shows the gate; a front page shows what the place wants to look like.
- Note what you cannot observe,which will usually include how the place treats a disagreement, since that only shows up in one.
- Check the age and stage.A young community and a settled one differ in ways that have nothing to do with their design, and the stages are described under internet communities.
- State the conclusion as conditional.For this decision, on these axes, with these gaps. Anything stronger is a preference with a table attached.
What comparisons are usually for
Most published comparisons of online spaces are arguments in a dispute. Somebody wants people to move from a subreddit to a Discord server, or wants to justify having moved, and the axes are chosen to produce the answer. That does not make them useless, but the axes deserve more scrutiny than the conclusions.
Two ways to check. First, ask whether any axis would favor the other candidate, and if none does, the selection is doing the work. Second, ask whether the writer has actually participated in both, over enough time to see a conflict.
Comparisons written from the outside describe the front page, which is the part designed to be described. The general reason outside descriptions of a group are unreliable is covered under generational digital culture, where the same error is made about age groups.
Common questions
Can I compare two places I have only read about?
You can compare their published rules and their stated governance, which is real information. Comparing English Wikipedia's deletion policies with MetaFilter's moderation guidelines is fair. Comparing what two Discord servers feel like is not, and the difference between those tasks is where most bad writing lives.
Is a bigger community always harder to moderate?
Harder per moderator, yes, and the relationship is not simple: a large place with a strong gate can be easier than a small one with none. The gate and the labor are the axes, not the size.
What about comparing a community across time?
That is usually the more informative comparison and the harder one, because the earlier state is only visible through whatever survived. What survives, and what it leaves out, is set out under internet culture archives.
Which axis matters most?
Exit cost matters most, for most decisions people actually face. It determines how much a place can ask of you before leaving becomes the reasonable response.
It is also the axis nobody looks at until it is too late to look. Leaving a Discord server costs you the history, leaving MetaFilter costs $5, and leaving English Wikipedia costs almost nothing.
