The ingredient label for data

A food label lists what is inside, where it came from, and when it expires. A part carries a spec sheet and a lot number. Data that feeds a decision usually arrives with none of this, and we have gotten used to it. The ingredient label for data fits on four lines.

Source

The specific origin, named and linkable. This filing, this feed, this reading, at this address.

Read time

When the value was observed. A fact is true as of a moment, and the moment is part of the fact.

Steps since

What happened between the source and the value in your hand: parsed, joined, scored, summarized. The label carries the chain, so the value is not a mystery.

Signer

Who recorded it, as a signature over the exact bytes, checkable against a published key. This line is what makes the other three verifiable instead of merely stated.

Read the label and you know what you are feeding a model. Skip it and you are trusting a value for reasons you could not write down.

One caution on the signer line, because it is the one most often oversold. A signature proves who recorded the data and that it has not changed. It does not prove the data is correct. The label makes the data accountable and checkable; judging whether the source is right is still your job, and the label is what makes that job possible by pointing you at the source.

DRM3 puts this label on every fact and dataset it serves: trusted data for AI, with the source, the time, the steps, and a receipt you can check.