Portfolio · Written for this search · text
Text
Systems desk
Text is the plainest interface we have. The Systems desk treats it as the layer where a concept, a circuit and a record can all be read by a person and checked by a machine.
- Concept
- Plain text source
- Parse and check
- Run or render
- Recorded output as text
What we mean by text
Here, text means a sequence of characters with an agreed encoding and a stated grammar. It is not a file format owned by one tool. It can be read in an editor, kept in version control and compared line by line.
The Systems desk uses the word narrowly on purpose. A note, a circuit listing and a log of outcomes are all text. They differ only in the grammar that governs them. See the portfolio for the wider set of notes.
Where text sits in the work
A circuit is first written down before anything is run. The description in the quantum circuit model note is a sequence of operations on named registers. Held as text, that sequence can be reviewed before any simulator or device sees it.
The other end of the pipeline is also text. Outcomes from measurement and sampling arrive as strings of bits with counts attached. When classical and quantum steps are chained, as in hybrid execution, text is the hand-off between them.
Limits and discipline
Text is not self-checking. A listing can parse and still describe the wrong experiment. The desk therefore separates two checks: does the text obey its grammar, and does it say what the author meant. Only the first can be automated in full.
We also record what was not captured. If a run keeps no seed, no version or no date, the note says so. A text record is only as useful as the context it carries. Further notes on method are listed under documentation.
Finding from this desk
No measurement is claimed here. The position is a design one: keep the source and the output of every piece of work as plain text, so that each step can be read, compared and rerun.