General Criteria

From the documentation of the field

These are the criteria shared by all six primitives of Ismail's Primitives. They say what counts as evidence, what does not, and how a mechanism that performs a function is told from an artefact that resembles one. Each primitive's own page applies them in its terms. They state what the evidence looks like and say nothing about how it is gathered or weighed.

The unit of evidence

The unit of evidence is the operation of a mechanism, shown in records of decisions. The presence of a document, a method or a role is not evidence of possession, because a template that is filled in once and filed performs no function. Methodologies and error theory explain why.

Five properties of a mechanism that performs the function

A mechanism that performs a function has five properties. Each primitive page of Ismail's Primitives names its own forms of them.

• Operation. The mechanism changes decisions. There is at least one recorded case in which a decision was made differently because of it.
• Coverage. It operates wherever the situation arises, across decisions, units and periods, and not only where it was first needed.
• Currency. It is revised when circumstances change, and each revision is dated and tied to the change.
• Consistency. The same situation receives the same treatment each time.
• Traceability. Records link the evidence the mechanism used to the decision it informed.

Mechanism and artefact

A mechanism can carry a different name from the primitive it serves, as defined in Ismail's Primitives, and an artefact can carry the primitive's name exactly. The test is the function.

What counts as evidence

• Dated records of decisions and the basis on which they were made.
• Artefacts with revision histories.
• Practice observed across more than one case.
• Accounts from more than one party, including those who make decisions and those who rely on them.
• The structure's behaviour over time.

What does not count

• A description of itself, given by the structure alone.
• The presence of a named method or framework.
• A document that no decision followed.
• A one-off exercise that is not revised.
• Outcomes taken alone. Results over a period reflect much more than the structure, so a good run shows nothing about whether a mechanism exists, and a poor one shows nothing about whether it is missing.

Samples

The cases examined are chosen by a rule fixed before the evidence is read, not by the organisation's preference. The sample includes at least one case that succeeded, one that failed and one still in progress. It is widened until the findings stop changing, and the verdict states the sample.

Time

A verdict applies at a stated date. Evidence is dated, and evidence from before a change in the structure is not evidence about the structure after it.

Two levels of evidence

The same evidence supports two levels of possession. Ordinary possession needs a mechanism that performs the function and no standing failure of the kind the failure condition describes. Robust possession needs the stronger, quantitative form defined for the primitive in Ismail's Primitives, as each criteria page sets out. The standard states what each verdict asserts.