Provisionality
Adapted from the Cognitive Dimensions of Notations framework (Green & Petre, 1996; Blackwell & Green, 2003). See About the Dimensions for framework references.
Provisionality is the degree to which users can sketch, explore, and make tentative marks without committing to final decisions. In the Cognitive Dimensions framework, this is sometimes described using the pencil metaphor: the ability to make "faint, blurry marks meaning something more or less like this" before committing to precise, permanent strokes.
In complex professional workflows, provisionality is not a luxury — it is essential. Solutions to complex problems are inherently provisional. There is rarely a single correct answer, and users need to explore multiple approaches, compare alternatives, and refine their work iteratively. Software that forces premature precision or permanent commitment at every step actively works against how professionals think and create.
Provisionality is the natural antidote to Premature Commitment. Where Premature Commitment forces users to decide too early, provisionality gives them space to defer decisions, work roughly, and commit only when ready. The two dimensions sit at opposite ends of the same spectrum: high provisionality means low premature commitment, and vice versa.
Non-Destructive Editing​
The most common implementation of provisionality in professional software is non-destructive editing — the ability to apply changes without permanently altering the source data. Adjustment layers in Photoshop, parametric modelling history in 3D tools, and formula-based transformations in data analysis tools all preserve the original while allowing exploration. The key property is reversibility without information loss: you can always get back to the original state, and you can adjust the parameters of any intermediate step.
Draft States and Parallel Explorations​
Beyond non-destructive editing, true provisionality means supporting draft states — incomplete, rough, or experimental configurations that the system treats as legitimate working states rather than errors. A 3D artist should be able to rough out a pose on an unfinished rig. A video editor should be able to drop placeholder clips into a timeline without the system complaining about mismatched formats.
Parallel explorations take this further: the ability to maintain multiple alternative versions simultaneously and compare them side by side. Version branching, comp alternatives, or "what-if" scenarios all support this. The challenge is keeping multiple provisional states manageable without creating confusion about which version is "real."
When Provisionality is Missing​
The absence of provisionality is felt most acutely during exploration. When every action is permanent and precise, users become cautious. They stop experimenting. They commit to the first approach that seems workable rather than exploring alternatives. The software subtly pushes users towards conservative choices — exactly the opposite of what professional creative work demands.
Testing​
- Can users explore different approaches without irreversible consequences? Try an experimental workflow and then attempt to reverse course.
- Are there draft or preview states? Can users see the effect of a change before committing to it?
- Can users maintain and compare multiple alternative versions of their work?
- Is "undo to checkpoint" available, or only step-by-step undo? In complex workflows, users need to be able to roll back to a known good state, not just reverse individual operations.
- Does the system treat incomplete or rough work as a valid state, or does it force completeness at every step?
Conclusion​
Supporting provisionality means trusting users to work at their own pace and in their own order, committing to decisions when they are ready — not when the software demands it. The best professional tools make provisional exploration feel natural and safe, reserving permanent commitment for the moments when users explicitly choose it.
Check After Improvement​
- Premature Commitment
- Error-Proneness