Research
Long-form research behind Nah?. These are working documents, not marketing. They contain the numbers that argue against the project as well as the ones that argue for it.
Figures are labelled by origin throughout. A company claim is what a company says about itself. An estimate comes from an outside tracker and is modelled rather than measured. Where something could not be verified, the documents say so.
The documents
Concept audit: Nah? ten months in
11 September 2026. Where the project actually stands, where the specs and the decision records contradict each other, who else is in this space, whether one person can build it, and a server architecture designed so any family could run its own.
What makes people come back
12 September 2026. How the private-social apps are really doing, which return mechanics work, why BeReal collapsed, what the research says about habits and notifications, whether anyone pays, and why the calm ones keep dying.
Circles and the backend question
13 September 2026. A person belongs to many overlapping circles at once. Can Mastodon, the wider fediverse or Matrix express that? What per-circle digests, dying circles and archives should look like.
Spike results: which runtime, and how many databases
13 September 2026. Four candidate servers measured against thresholds written in advance, plus the SQLite topology question and the digest storm. The result overturned the framework choice.
Related
- Decisions — the architecture decision records
- Ten Months In — the post that summarises both documents