Defining the Hairy Problem

Jan 10, 2026

This is the fundamental issue each knowledge worker needs to solve on a daily basis.

Entropy reduction

If you see a random day in a knowledge worker’s job, it would sound something like this—

• “Boss has asked me to fix the sales numbers. What does that even mean?”
• “This design does not drive conversions, and I have no idea what to do.”

Entropy is why an email with the subject line “Quick sync?” ruins your morning.

The job of a knowledge worker is to reduce the entropy in the system—to act like a funnel that takes in high entropy (chaos) and converts it into low entropy (order).

Articulation is the tool with which one can achieve this. It is not just writing the problem down; it is constraint identification.

Clarity of thought vs intelligence

Standardized tests (SATs, GREs) are designed to measure raw intelligence, but they cheat: they give you well-defined problems. They remove the noise. Real life is nothing but noise. Therefore, real-world success is a function of Raw Intelligence × Clarity of Thought.

Clarity of thought and intelligence seem to be two different things. While one is largely innate, the other is an acquired skill.

Articulation is a skill, as it is trainable. The skill can be defined as the ability to define the shape of a problem. So, how do we train this “clarity of thought”? It starts by visualizing the problem physically.

The shape of the problem

A shapeless problem is a source of stress. It might look like—“My work is boring!”

A better problem statement would be—“I need to negotiate for two interesting tasks that help me improve my Python coding skills without increasing my total working hours.”

Let’s take a really complex, hairy issue—something that has a lot of moving variables, like deciding a career or finding the right house.

We can take inspiration from computers and try to solve it algorithmically. This would mean breaking down the problem into smaller ones until we hit a trivial problem with a known solution. This is called recursive decomposition and is a well-studied concept that is heavily used while designing solutions for complex problems in computer science.

Vague: Deciding a career

Decomposed: “Need to find a role that pays at least $X, requires no more than $Y$ hours of travel, and utilizes skill $Z$, within the next six months.”

This makes the problem far more solvable, which is exactly what we are trying to do.

Math is very rigorous when it comes to defining problems. Jacques Hadamard suggested the concept of well-posed problems.

A problem is only “well-posed” if:

• A solution exists
• The solution is unique
• The solution depends continuously on the data

While life doesn’t always have a unique solution like math, we should strive for the clarity that math demands.

This gives us an algorithm to convert a mess into a better-organized problem:

• Define constraints that the desired solution should have
• Restrict the scope of the solution

The cool thing about defining a problem is that it is a very effective tool for understanding the problem.

Automate the articulation

The highest form of this skill is when articulation becomes automated. Great problem solvers are really just framework developers. They find themselves defining the same problem types repeatedly, so they crystallize their methodology.

For example, Elon’s framework for solving hard problems is:

• Make the requirements less dumb. (Question the constraints immediately.)
• Delete the part or process. (If you are not adding back at least 10% of the time, you aren’t deleting enough.)
• Simplify or optimize. (Only do this after steps 1 and 2.)
• Accelerate cycle time.
• Automate.

Or Jack Dorsey’s protocol:

• Perfect every detail.
• Limit the number of details.

The payoff

“A problem well put is half solved.” — John Dewey

The hidden benefit of articulation is that once a problem is constrained and defined, you no longer have to be the one to solve it. A vague problem requires you to stress over it. A well-articulated problem can be outsourced—to a junior employee, to code, or to AI.