Apparat builds Data Fidelity Registers: machine-readable, structured records gathering the facts an organisation publishes about itself.

Artificial intelligence doesn’t do magic. Machines are very efficient at gathering information from many sources, but sometimes there’s nothing to gather from. The web was built by humans, for humans, and not for machines: information is scattered across websites, documents, databases, PDFs and many other formats; sources aren’t always accurate, available or reachable. So an answer may be drawn from a good source, a stale one, or none at all, and nothing in it tells you which: a correct answer and an invented or approximate one look exactly alike. And as information passes from one system to the next, tracing it back to its source gets harder. What’s missing is a simple way for a machine to establish: what did this organisation actually publish? Where does this information come from? And has it stayed unchanged?

Every organisation should have a Data Fidelity Register.

It gathers the facts an organisation publishes about itself, and each one is bound to its original public source and sealed cryptographically. The register is published on the organisation’s own domain, and can be verified independently by anyone, without having to trust us. A single place where what an organisation has said is structured, dated, and checkable. That is what we are building towards, one register at a time: a web where the information organisations publish is easy for machines to read, to trace back to its source, and to verify at any time. Not big data. Small, reliable data.