The useful life of a fact

The first record in a database can become wrong while you're still collecting the millionth.

A company changes its name. A product changes its price. A person leaves a job. The old record can remain complete, searchable and neatly formatted. Nothing about its appearance has to reveal that the world moved on.

I find that tension more interesting than the size of a collection. Each record adds something to maintain. How much attention it deserves depends on what someone needs it for.

Fresh enough for which decision?

Consider a company directory. Someone looking for businesses worth researching may still find an older record useful. Someone trying to reach a particular employee needs a more current answer. The company hasn't become less relevant. The decision has become more sensitive to change.

Freshness starts with that use. How current does the information need to be? What happens if the answer is old?

Updating everything more often could improve some answers while spending effort on facts that rarely change. A founding year and a live price ask for different kinds of attention. A broad research directory makes a different promise from a service someone relies on for an immediate decision.

This is what makes a single update schedule unsatisfying to me. It can describe a process accurately while telling a customer very little about whether a particular answer is dependable.

What the timestamp records

There is another distinction I would want a product to preserve: when a source was checked and when the underlying fact was established.

Looking at an old page again doesn't make its contents new. A recent check might confirm only that the same statement is still published. If a timestamp blurs those events together, the effort spent checking can produce more confidence than the source warrants.

That changes what I would want to know before making a dataset bigger. Which fields matter to the decision? What could establish that they are current? Where would the product need to show uncertainty?

A smaller collection with clear answers to those questions could be more useful for a particular purpose than a larger one whose age is impossible to establish. The product would have to earn that claim in use.

I still like the ambition of a broad index. It can help someone discover where to look. A carefully maintained set of details can help them act. Knowing which of those jobs a product promises to do gives the upkeep a purpose beyond keeping its update counter moving.