Mutations#

The accounting team wants to flag unresolved invoices for review when their customer's payments are on hold. The operation has both a selection rule and an intended change. If each application implements them separately, one may flag invoices that another excludes. A mutation keeps the selection and the change together as a definition the teams can share.

That definition is part of the dataset. The rule for changing an invoice can be examined alongside the invoice, just as a query or schema can. The team can examine which invoices qualify and which fields the operation changes. The rule remains available for inspection and change as the team's understanding develops.

Evolution#

The review policy may change as the team learns more about disputes. Initially, it might flag every unresolved invoice associated with a payment hold. Later, the team might limit that operation to invoices above a particular amount. The change belongs to the history of the mutation definition.

That history matters when someone asks why an invoice was flagged last month but would not be flagged today. The current invoice and current rule may no longer explain the earlier decision. Keeping the rule's evolution makes its earlier requirements available for examination with the relevant data. This applies Reification to the operations that change the dataset itself.

Exploration#

A proposed rule needs examination before it becomes an accepted change. The team considering an amount threshold needs to see which invoices would qualify and what would happen to them. Trying the idea should not require flagging those invoices and then undoing the experiment. A preview must leave the dataset unchanged so that examining a proposal remains separate from adopting it.

For example, the team might compare a rule covering every unresolved invoice with one covering only invoices above the proposed threshold. The useful evidence is the affected invoices and the intended changes. Those results let people discuss the proposal using actual data rather than different interpretations of its wording. The definition they examine can then be the basis of the operation they agree to use.