HOW INTELLIGENCE WORKS · PLANNING · eduKateSG
How a Mind Builds a Route From the Present to a Future Goal
Planning is the intelligence process that turns a desired future into an ordered route of subgoals, dependencies, resources, checks, contingencies and actions.
Goal → current state → constraints → subgoals → dependencies → route → checkpoints → action → update → arrival or replan.
This article belongs to the How Intelligence Works series. The main hero owns the full intelligence city. This pillar isolates planning: how intelligence constructs an executable route rather than merely imagining possibilities, and how that route remains adjustable when the world returns new information.
The Route Problem
A goal describes a preferred future. It does not yet say how to reach it. Planning fills the space between present state and goal by arranging actions in an order that respects dependencies, cost, time, uncertainty and available capability.
The route can be simple: finish homework before dinner. Or it can be civilisational: build a transport system, educate a generation, prepare for a disaster, develop a scientific programme. The scale changes, but the core problem remains: what must happen first so that the next necessary state becomes possible?
A plan is not a list of wishes. It is a dependency map across time.
1. Planning Begins With a Goal State
Planning needs a representation of what success means. A vague goal such as “do better” provides weak routing. A stronger goal specifies enough of the desired state that progress can be recognised.
Goals may be quantitative, qualitative or structural. The important feature is that they constrain the route. They tell the system which changes count as progress and which attractive detours do not.
A plan therefore inherits the quality of its goal. If the goal is poorly chosen, excellent execution can produce the wrong future efficiently.
2. The Present State Must Be Mapped Honestly
A route cannot be planned from an imaginary starting point. The system needs an accurate account of current resources, knowledge, constraints, dependencies, time, risk and unfinished work.
Planning failure often begins with optimistic state representation: prerequisites are assumed complete, hidden debt is ignored, people are counted as available when they are already overloaded, or uncertainty is converted into certainty to make the schedule look cleaner.
A realistic plan starts with a realistic “here.”
3. Planning Decomposes Goals Into Subgoals
Large goals are rarely actionable in one step. Planning decomposes them into intermediate states that can be reached, checked and connected.
- What must be true before the final goal is possible?
- Which prerequisite must exist before that?
- Which tasks can proceed in parallel?
- Which tasks block everything downstream?
- Which milestone gives evidence that the route is still working?
This decomposition reduces complexity because the mind no longer needs to solve the entire future simultaneously. It can operate through bounded stages.
4. Dependencies Give the Plan Its Shape
Some actions can happen in any order. Others cannot. A dependency means one state must exist before another action becomes valid or useful.
| Dependency type | Planning question |
|---|---|
| Knowledge | What must be understood first? |
| Resource | What material, money, time or access is required? |
| Authority | What approval or permission must precede action? |
| Sequence | Which operation logically depends on an earlier result? |
| Capacity | Who or what must be free to perform the task? |
| Evidence | Which uncertainty must be reduced before commitment? |
5. Planning and Counterfactual Simulation Are Different
Counterfactual simulation explores possible futures. Planning selects and organises a route toward a chosen future.
Simulation asks, “What if we took route A?” Planning asks, “Given our goal, constraints and current state, which route will we execute, in what sequence, with which checkpoints and fallback?”
The companion article How Intelligence Works | Counterfactual Simulation owns scenario exploration. Planning uses those scenarios as inputs to route design.
6. Plans Need Checkpoints
A long route without intermediate checks can drift far before failure becomes visible. Checkpoints divide the plan into states where progress, assumptions and conditions can be re-evaluated.
A useful checkpoint asks:
- Did the expected intermediate state occur?
- Has a critical assumption changed?
- Is the remaining route still feasible?
- Has cost or risk moved outside tolerance?
- Should the next stage begin, pause, switch or stop?
Checkpoints transform planning from prediction into controlled adaptation.
7. Planning in Mathematics
Mathematical problem solving contains planning whenever the answer cannot be reached by one immediate operation. The learner must decide what intermediate quantities are needed, which representation to construct, which theorem or method applies and in what order the steps should occur.
A strong solver often plans before calculating:
What do I need to know at the end? What information would give me that? What earlier quantity produces that information?
Working backwards from the goal is therefore a planning technique, not merely a mathematical trick.
8. Planning in Learning
Students often mistake intention for planning: “I will study more.” A learning plan becomes real when the goal, sequence, practice type, feedback point and schedule are explicit.
- Which prerequisite must be repaired first?
- Which topic deserves retrieval rather than rereading?
- When will practice be spaced?
- Which assessment will reveal whether transfer occurred?
- What should change if performance remains weak?
The plan should be built around learning mechanisms rather than only hours spent.
9. Planning Under Uncertainty
Real plans operate before the future is known. Good planning therefore does not require every future state to be predicted precisely. It identifies uncertainty and builds adaptation points.
The companion article How Intelligence Works | Uncertainty owns the map of what is not settled. Planning converts that uncertainty into reserves, alternatives, trigger conditions and reversible commitments.
10. Planning Failure Atlas
| Failure | What happens | Repair |
|---|---|---|
| Goal vagueness | The route has no clear destination | Define success state |
| Starting-state fiction | Hidden debt or missing prerequisites are ignored | Audit current state |
| Dependency blindness | Tasks begin in the wrong order | Map prerequisites explicitly |
| Schedule optimism | Uncertainty and variation are removed from estimates | Add buffers and ranges |
| No checkpoints | The route drifts before failure is detected | Add stage gates |
| Plan inertia | The route survives after conditions change | Define replanning triggers |
| Contingency overload | Too many branches make action impossible | Prepare only material alternatives |
11. Teams Need Shared Plans and Local Freedom
Collective planning requires enough shared structure that contributions align, but enough local freedom that specialists can adapt within their domains.
A strong shared plan identifies goal, interfaces, dependencies, owners, deadlines, evidence gates and escalation paths. It should not prescribe every local action when the people closest to the work need room to respond intelligently.
The plan should coordinate the handoffs without pretending the centre can see every local road.
12. Institutions Turn Planning Into Operating Memory
Institutions plan through budgets, calendars, programmes, maintenance schedules, emergency procedures and long-term strategies. These externalise future coordination so that action does not depend on one person’s memory.
The institutional plan becomes more intelligent when assumptions, dependencies and triggers are recorded—not only final dates and targets. This allows successors to understand why the route was built and when it should be changed.
13. Artificial Intelligence and Planning
AI systems can help decompose goals, generate candidate sequences, estimate dependencies, compare routes and monitor progress. Agentic systems may also execute parts of a plan through tools.
The planning risk increases when generated steps become actions. A plausible sequence is not necessarily authorised, feasible or safe. Reliable AI planning therefore requires explicit permissions, state tracking, tool boundaries, checkpoints, reversible actions where possible and confirmation before high-impact commitments.
The plan needs an owner even when the planner is partly automated.
14. The Planning Audit
- Goal: What future state defines success?
- Present state: What is actually true now?
- Constraints: What limits the available routes?
- Subgoals: Which intermediate states are necessary?
- Dependencies: What must happen before what?
- Owners: Who carries each action and handoff?
- Resources: What time, information, money or capability is required?
- Checkpoints: Where will the route be tested?
- Triggers: What change forces replanning?
- Return: How will actual consequence revise the next plan?
15. CivDJ Reading: Planning Is Controlled Release Across Time
In the CivDJ frame, planning is a sequence of controlled releases. The system does not dump the entire future at once. It prepares the next state, verifies fit, releases, observes the return and then allows the following stage to proceed.
Different Masters may own different stages. The plan preserves their interfaces and ensures that one stage does not proceed before its prerequisites are satisfied.
A strong plan is a future made navigable one verified handoff at a time.
16. Return to the Route
A future goal begins as a point beyond the current map.
Planning turns that point into subgoals, roads, dependencies and checkpoints. The route is never guaranteed because the world can change while we travel.
Intelligent planning therefore does two things at once: it commits enough structure for coordinated action and preserves enough flexibility to rebuild the route when reality returns a different map.