Benefits of social media, in six rules: how to test any claim

The case for networked communication is usually made badly, by people selling it, and the case against is made loudly, by people who have already decided. Between the two sits a set of real advantages that hold under specific conditions and fail outside them. The conditions are the useful part, and they are the part that gets left out.

This page states each benefit as a rule with its conditions attached, so that a reader can tell when a claimed advantage applies to them.

What to take away

The six benefits, in one list:

Each of the six rules below carries a condition:

Rule one: the useful connections are the weak ones

People who can tell you something new are not the ones you talk to daily, because your close circle shares your information.

People at the edge, met once, followed loosely, remembered vaguely, connect to circles you cannot see. The sociology of weak ties predates the internet by decades, but maintaining many weak ties became nearly free.

The classic case is the job tip from someone you met once and never spoke to again. LinkedIn built its product on that claim, and its People You May Know feature exists to manufacture those ties.

Condition: the benefit holds when the network is large and varied. A network of people who already share your background reproduces your close circle at scale and adds nothing. A network built by following whoever is recommended will converge on the same few sources everyone else follows, and the ties will be weak without being useful.

Rule two: rare needs meet rare peers

For a common situation, the people around you are enough. For an uncommon one, an unusual illness, an obscure craft, a minority language, or a specialized job, those who share it may be spread thin. No town holds two of them.

Networked communication is the only mechanism that has ever reliably let these people find each other. The communities they form are among the most durable described under internet communities.

PatientsLikeMe began as a forum for people with ALS, a condition too rare for any single town to gather a roomful of patients. Subreddits for conditions such as Ehlers-Danlos syndrome do similar work.

Condition: the benefit matches how unusual the need is, enormous for a one-in-a-million situation and negligible for a one-in-ten situation.

Rule three: coordination that used to need an institution

Organizing people to act at once, a protest, fundraiser, search, clean-up, or disaster response, once required an organization with a list and a way to reach everyone on it. That list and reach now belong to anyone.

The pattern of mutual aid, which used to depend on a standing institution, can be assembled in hours and dissolved when its work is done.

Occupy Sandy, the volunteer relief effort after Hurricane Sandy in 2012, ran on shared spreadsheets and social media rather than a single nonprofit. The Cajun Navy, a volunteer boat-rescue group, forms through Facebook groups when a flood hits.

Condition: this holds for things that are simple to explain, urgent, and do not need sustained commitment. Work that requires people to keep showing up for months still needs the institution.

Rule four: the record you did not know you were keeping

Public text conversations persist. A question gets answered once, and the answer remains for the next person who asks it. Problems get solved in the open, so someone absent can find the solution years later.

Much practical knowledge now exists only this way, in old threads, comments, and answers. It was never written up anywhere else. Stack Overflow is the type case: an error you hit today may be answered by a post written more than a decade ago. Usenet archives from the 1980s are still readable.

Condition: the benefit requires the container to persist and to be searchable. Conversation in ephemeral or closed containers produces none of it.

The shift toward those containers, from Snapchat in the 2010s to Discord servers and Slack, quietly destroys this benefit while the others survive. The problems for anyone reconstructing the past are set out under early social networks, where the record has mostly gone.

Rule five: access to people who would not take your call

An expert who would never answer a stranger's letter or call may still answer a public question. A public answer reaches everyone asking it, at the same cost as answering one person. Reddit's AskHistorians, Stack Overflow and MathOverflow all run on this: one volunteer answer serves every beginner who finds it.

The gap between beginner and expert has never been shorter. A beginner can now place a question where an expert will see it.

Condition: the question has to be good, public, and asked where the expert is. The same mechanism that lets one reply reach many also lets one expert be flooded by many askers.

Experts who answered freely in a field's early period mostly stop once the volume rises. Standing, as described under online status and identity, determines whose question is seen, and it is not distributed by merit.

Rule six: reach without permission

Anyone can publish. No editor, no gatekeeper, no cost. A person with something to say and no institution behind them can, occasionally, be heard by a very large number of people, and the mechanism that produces this is the same one described under memes and virality.

Blogs in the early 2000s, YouTube after 2005 and TikTok more recently all widened that door. The Ice Bucket Challenge of 2014 spread through personal posts rather than any organization's campaign.

Condition: occasionally. Reach without permission is real and it is a lottery. Most of what is published reaches almost nobody, and the distribution of attention is more unequal than in any medium with gatekeepers.

Reading a benefits claim

Four questions cover most claims:

Run any pitch, headline or argument through those four. A claim that cannot answer the first two is usually about the technology, not about the benefit. A claim that fails the last two describes a benefit that is already fading.

Common questions

Are the benefits worth the costs?

For whom, under which condition. For the person with a rare need and no local peers, almost certainly. For the person whose needs are met and whose weak ties are already plentiful, possibly not. There is no answer that does not depend on the case.

Why do the benefits seem to have shrunk over time?

Two of the six rules depend on conditions that have weakened: persistence and searchability of the record, and the willingness of experts to answer in public. Both were strongest early and have declined as containers closed and volume rose.

Is a bigger network always better?

Larger and more varied is better for weak ties. Larger and more similar is not. Growth by recommendation tends to produce the second.

Why does this page not cite research?

Because the research is contested, mostly correlational, and used selectively by every side. The benefits described here are mechanisms a reader can verify from their own experience, which is a better foundation than a figure that will be disputed next year.