Clarity

The Lens

When you do not know what to do, something needed for the next action is missing. You may not know the goal, your current position, the available actions, or what those actions would produce. Clarity is a useful way to describe having that information: a visible path through the possible states of the task.

Knowing what to do appears to involve five parts:

• Current state: an observable account of where you are.
• Goal state: a concrete result to reach.
• Action space: the moves available to you.
• Transition model: how each move changes the state, including its expected probability of success.
• Path: a sequence connecting the current state to the goal.

An undefined part leaves the planning process without a required input. Modeling “I don't know what to do” this way gives you something to inspect and repair without treating it as a character flaw.

The route expresses your current model of the situation; observing the action's result can change that model.

Read this diagram

Current state → Available action → Expected change → Goal state.

Why This Is Useful

Without an action space, you cannot generate the next action. Without a transition model, you cannot judge whether an action is worth taking. Undefined EV variables prevent sustained motivation. What feels like procrastination is the absence of a clear, executable path.

With clarity, the next action becomes obvious and requires no deliberation. The expected-value calculation returns a high value, motivation emerges, and execution has a low activation cost. Specifying the task makes the behavior runnable.

The Model: Clarity and Expected Value

Clarity determines whether the variables in the Expected Value calculation are available:

$$EV = \frac{reward \times probability}{effort \times time\_distance}$$

(This model is a useful heuristic for understanding motivation patterns, not a quantifiable prediction formula. The value is in identifying which variable to adjust, not calculating exact numbers.)

Defining the variables means knowing what you will get, that the action will work, what it costs, and that the result is soon. High EV then tends to produce motivation, which often enables action.

The Diagnostic Rubric

These 7 questions locate the missing input:

Ask them in order when stuck. The first one you cannot answer cleanly identifies where clarity broke down.

Examples: Clarity vs Lack of Clarity

Example 1: "Work on the project"

“Make progress” does not specify a result. “The project is messy” does not specify its current state. The next action could be almost anything, and the effort could be 30 minutes or 8 hours. The EV calculation cannot run, so motivation fails and the task is postponed.

Now specify the goal: implement a user authentication endpoint with passing tests. An Express server is already running, but there is no authentication code. The next action is to create the /auth route handler file, which will hold the login logic.

That is step 1 of 4 toward working authentication. Allow 20 minutes to scaffold it, then test. The goal, starting state, action, outcome, path, and effort are all defined. High motivation follows and execution begins immediately.

Example 2: "Get in shape"

“Be healthier” does not define an observable target. “Exercise more” leaves the type and frequency open. Possible workouts are unbounded and “eventually” provides no timeline. A vague reward divided by unbounded effort and distant time leaves EV near zero.

A specified version is to complete 30 consecutive days of 20-minute gym sessions. The current state is Day 0, with no established gym habit. Tomorrow at 7am, go to the gym for a 10-minute warmup and 10-minute lift.

Day 1 is expected to cost ~6 units of willpower and begin establishing the pattern. The path runs from Day 1 to Day 30, with the cost decreasing through the 30x30 pattern. The commitment is bounded at 20 minutes per day for 30 days. These defined variables make execution possible.

State Transitions: From Fog to Executable

Before clarity, you may stare at a screen or whiteboard while searching an unbounded space. The search consumes capacity without producing an action.

Planning changes this through rigorous simulation: model the states, estimate transition probabilities, and generate a sequence of actions. Afterward, concrete instructions and a causal model are available. The next action is obvious, so execution needs little deliberation.

Achieving Clarity

Primary methods:

1. Use The Braindump to put vague thinking on paper, where missing variables can be identified.
2. Use the Clarity Bear Protocol with AI or another person to question the problem until the target is concrete and observable.
3. Ask bounded questions to limit the region being searched.
4. Use discretization to give an abstract goal observable milestones and clear completion states.
5. Give an LLM the fuzzy account of what is blocking you and have it identify the undefined variable.

The forcing question:

"What is the literal next physical action?"

If you cannot answer in one sentence using observable verbs, return to the diagnostic. The required clarity is still missing.

When to Use This Lens

Use the rubric when you stare at work without starting, feel overwhelmed by its scope, or repeatedly say you should work on something without doing it. Switching between tasks without finishing suggests that “done” is undefined. Procrastination typically indicates that no executable path is visible.

“I don't know what to do” is the trigger to run the 7-question diagnostic.

Common Anti-Patterns

Clarity and AI Usage

AI's primary value here is finding and specifying an undefined EV variable. Start by giving it the raw account of what feels unclear. It identifies the issue, such as an abstract reward or an unbounded action space, then proposes 3 concrete next actions with defined outcomes. You select one and execute.

AI as Accelerator requires clarity to provide value. A vague goal produces vague advice. A defined state and a specific question produce an actionable model.

Clarity Across Contexts

The Meta-Clarity Loop

Acting on a clear model produces data that can improve it. Define the current state, act, observe the result, update the transition model, then plan again. Repeating this cybernetic loop tends to improve the model of the possible states and how to move between them. Clarity enables execution, and execution supplies information for the next round of clarity.

Related Concepts

• Expected Value - The variables the diagnostic makes explicit
• Motivation - What emerges when the EV variables are defined
• Clarity Bear Protocol - Building clarity through questions
• Procrastination - Undefined variables preventing a next action
• Question Theory - Bounded questions that reduce the search space
• Algorithmic Complexity - Unbounded search caused by missing inputs
• The Braindump - Externalizing thought so gaps can be inspected
• Salience - Defined goals becoming distinguishable from noise
• Discretization - Giving milestones observable completion states
• Agency - Clarity needed to express intent and act
• Activation Energy - Lower startup costs when search overhead is removed
• AI as Accelerator - Specifying the problem that AI will accelerate
• Execution Resolution - Matching detail to where intervention is needed
• Cybernetics - Acting, observing, and updating the model

Key Principle

Treat a missing next action as a missing input to planning. Find whether the goal, current state, available actions, transition model, or effort bounds are undefined, then specify what is missing until the path becomes visible.

This framework emerged from N=1 observation and has proven useful in practice. Test whether the diagnostic rubric helps you identify stuck points in your own system.