If you only count public posts, privacy looks exactly like disappearance.
That is a measurement error.
A person who stops blogging and starts talking in a family WhatsApp group produces less searchable web material. A developer who moves work into a private repository contributes fewer visible technical artifacts. A creator who moves discussion into patron-only chat may appear quieter in public while spending more time talking with supporters.
The hard part is measuring that shift without responding by demanding access to everybody’s private messages.
Measure activity without reading the room
Surveys are one obvious tool.
Pew Research Center’s 2025 social-media survey asked a representative sample of U.S. adults which platforms they use. It found that about a third used WhatsApp, even though the contents of their private conversations were not part of the measurement. See Americans’ Social Media Use 2025.
That kind of evidence cannot tell us what people said, how meaningful the conversations were, or whether they used the app for one family thread or fifty active groups.
It can tell us that human activity exists outside publicly indexable posting.
Platforms can expose privacy-preserving signals too.
GitHub, for example, lets users choose to display counts of contributions made to private repositories while keeping the repository names and specific activity hidden from outsiders. See GitHub’s private contribution visibility settings.
That is a useful design pattern: show that work occurred without revealing the work itself.
Several weak measurements beat one confident mistake
A serious estimate of online participation might combine:
- representative surveys about platform use and frequency;
- aggregate active-user or membership figures where they are available;
- privacy-preserving counts of messages, sessions, or contributions;
- public posting volume;
- interviews or longitudinal panels that track how people’s habits change;
- traffic measurements that distinguish reading from publishing.
None of these alone measures “how alive the internet is.”
Message counts can be inflated by group chatter. Membership counts include inactive accounts. Surveys depend on memory and sampling. Platform statistics use definitions chosen by the platform. Public posts remain the easiest material for outsiders to inspect precisely because they are public.
The uncertainty does not disappear just because more metrics are added.
Privacy creates an intentional blind spot
That blind spot should be acknowledged rather than filled with speculation.
If a private service does not reveal message contents, an outside researcher cannot legitimately conclude that those messages are profound human conversation.
They also cannot conclude that nobody is talking because Google cannot index them.
The defensible answer is narrower: public visibility and human participation are different variables.
The Human Retreat does not require us to pry open the private internet to prove people are there.
It requires us to stop treating closed curtains as an empty house.