Run a review
Use a repeatable review cadence to keep Maps accurate, useful, and alive.
A Map becomes valuable when you return to it. The first version reveals structure, but the review process turns that structure into durable knowledge.
Use reviews to decide what should be preserved, revised, connected, or removed.
When to review
Review after the Map has changed shape, not only after a fixed amount of time. A review is useful when new sources were added, a decision changed, a draft became stable, or the next action is no longer obvious.
For small Maps, review after each focused work session. For larger projects, review before sharing the Map with someone else or before asking AI for a high-stakes output. The review is the moment where rough conversation becomes project memory.
Review the Map, not only the latest answer
AI workflows often focus on the newest output. That is understandable, but it misses the point of a Map. The value is in the whole working surface: sources, questions, decisions, drafts, and next actions.
During a review, scan the Map before opening individual Points. Ask:
- Which area has become clear?
- Which Point is now stale?
- Which open question is blocking progress?
- Which decision needs stronger evidence?
- Which useful Block should become context for future work?
This keeps attention on the shape of the project instead of the last message in a conversation.
Use four review actions
Most review work fits into four actions.
Preserve
Preserve material that is stable enough to reuse. A definition, decision, rubric, or summary can become a clean Point or Block.
Do not preserve everything. Preserve the material that will make the next question easier.
Revise
Revise material that is directionally useful but not yet precise. Open the Point, edit the language, ask BaseHalf for alternatives, or split a crowded Block into smaller pieces.
Revision is where a rough AI answer becomes a durable artifact.
Connect
Connect Points when one should inform another. A source may support a claim. A decision may constrain a draft. A review task may depend on a weak concept.
References keep context reusable without forcing you to copy it everywhere.
Retire
Retire material that no longer helps. Merge duplicates, archive stale questions, or delete temporary notes that would confuse the next session.
Good Maps are not bigger by default. They are clearer.
What to delete
Deletion is part of review. Keep a Point only if it still changes the work.
Delete or merge:
- duplicate summaries that say the same thing in different words
- old drafts that are no longer candidates
- questions that were answered and no longer explain the reasoning path
- risks that do not produce a test, mitigation, or decision
- references that make the graph look busy without adding useful context
If a piece feels emotionally expensive to delete, ask a practical question: would this help the next person make a better move? If not, it can leave the Map.
End with the next action
Every review should leave the Map easier to continue.
Before you stop, create or update one visible next action. It can be a task, a question, or a Point that needs another pass. The point is to make the next session start from the current state, not from memory.
The review cadence is simple: preserve what matters, revise what is useful, connect what should travel, and retire what has stopped helping.
Review checklist
Use this checklist when a Map starts to feel crowded:
- Every source Point has a reason to exist.
- Every decision Point includes the reason for the decision.
- Every open question still blocks something.
- Every task has enough context to be resumed.
- Every Reference changes what should be considered.
- The next action is visible without rereading the whole Map.
If the checklist exposes too much clutter, do not reorganize everything at once. Retire the noisiest Points first, then reconnect the material that still matters.