~/writing/math-foundations-for-reasoning-models $ ls
Reasoning Models as Representation Transformation
Planning
Planning is the formation of a structured path of possible actions or steps before or during task execution.
~/writing/math-foundations-for-reasoning-models $ cat topic.md
Planning
Definition
Planning is the formation of a structured path for reaching a goal.
Simple form:
goal -> possible steps -> chosen path
In reasoning models, planning can guide which transformations or actions come next.
Planning organizes future steps before or during execution.
Why Planning Is Useful
Some tasks cannot be solved well by reacting one token at a time.
Examples:
write a program
solve a multi-step problem
compare strategies
use tools in sequence
debug an error
Planning helps preserve global structure.
Planning keeps local steps aligned with a larger goal.
Plan vs Execution
A plan is not the same as successful execution.
Example:
plan:
do A, then B, then C
execution:
A works, B fails, C must change
The model may need to update the plan as new information appears.
Planning is a guide, not a guarantee.
Explicit And Implicit Planning
Planning can be explicit.
Example:
write down steps before answering
Planning can also be implicit.
Example:
internal representation favors a certain sequence of actions
Both are forms of structured control over future steps.
Planning can be visible in text or hidden in representation.
Planning As Representation
A plan is itself a representation.
It represents:
goal
current state
possible actions
ordering
constraints
expected outcome
The model can then transform or revise that representation.
Planning is representation of possible future computation.
Final Summary
- Planning forms a structured path toward a goal.
- It helps multi-step tasks stay organized.
- A plan is not the same as successful execution.
- Planning can be explicit or implicit.
- A plan is a representation that can be revised.
~/writing/math-foundations-for-reasoning-models $ open
Continue the series.
Scratchpad