Scope fidelity
Scope fidelity is the rule that a short name standing in for some bigger underlying activity has to honestly match what's under it. The name can't promise more than the actual details support, and it can't quietly leave out content the details include. Overclaim usually shows up as aspirational language: calling an intent "run a profitable freelance business" when the activity beneath it has only been specified as "sends monthly invoices" loads the name with freight (profitability, sustainability, business operations) the underlying details don't substantiate. Underclaim runs the other way: the name omits content the details actually produce, leaving later analysis to reconstruct what the label should have carried.
Definition
The constraint that a scoped description — the name of a scoped transition, a condition-rollup, an intent category, or any coarser-grained statement summarizing finer-grained material — accurately tracks what it summarizes. The compression must claim no more than the underlying material supports, and no less than what that material ratifies.
Overreach and under-reach
Scope fidelity fails in two directions.
-
Overreach: the name asserts characteristics the underlying structure doesn't ratify. Aspirational modifiers are the common instance — "the right candidate," "properly onboarded," "a seamless experience" — where the modifier carries rhetorical freight the specification doesn't substantiate. Overreaching names look like framework objects but collapse on verification, because there's nothing in the structure to check the modifier against. Overreach at parent-scale takes a related form: a parent name whose children don't collectively constitute what the name claims.
-
Under-reach: the name omits characteristics the structure does produce. Content the transition produces goes unnamed, and downstream analysis has to reconstruct what the name should have carried. Under-reach at parent-scale takes a related form: a name that drags child-scale detail upward instead of compressing — "user clicked signup, entered email, confirmed password, submitted" where "user signed up" is the available compression. Under-reach is harder to spot than overreach because the name isn't making a false claim — it's just making too small a claim, or one at the wrong altitude.
Three consistency checks
Three diagnostics test names, each at a different layer.
- Hot-air tests structure-existence: is there any specifiable structure underneath the name at all, or is the name floating?
- Re-sort tests structure-vs-examples: does the chosen structure actually fit the seed instances it was built to cover?
- Scope fidelity tests name-vs-structure: does the name accurately compress the chosen structure?
A name can pass hot-air and re-sort and still fail scope fidelity. The structure exists; it fits the examples; the name still oversells or undersells what the structure supports.
Example
A team running intent detection groups its trial users into an intent category called "users seriously evaluating for purchase." The category's specified structure: users who visited the pricing page at least twice in their first week.
Hot-air passes — the structure is specifiable. Re-sort passes — the seed users in the category all did visit pricing twice in their first week. Scope fidelity fails. "Seriously evaluating for purchase" claims an intent that two pricing-page visits don't ratify; users hit pricing for plenty of reasons (curiosity, comparison-shopping, checking what a colleague mentioned). Downstream apparatus will run on "evaluating buyers" while the structure underneath only supports "twice-pricing-page-visitors."
Two repairs: rename to what the structure ratifies ("repeat pricing-page visitors"), or add structure that ratifies the evaluation claim — team-account creation, demo bookings, invitations to other seats.