The Entropy Defense Mechanism
"Complexity does kill companies... It's never too early to plant the seeds of simplicity." — Dharmesh Shah
What It Is
A deliberate imposition of artificial constraints and simplification rules to counteract the natural organizational drift toward complexity (the Second Law of Thermodynamics applied to business).
When To Use
- During periods of rapid growth
- When considering expanding the product portfolio
- When processes feel increasingly bureaucratic
- Before adding new product lines or features
The Problem: Entropy
STARTUP SCALED COMPANY
Simple ▓▓▓▓▓▓▓▓▓▓▓▓ →→→→ ▓▓▓░░░░░░░░░░ Complex
Fast ▓▓▓▓▓▓▓▓▓▓▓▓ →→→→ ▓▓▓▓░░░░░░░░░ Slow
Decisive ▓▓▓▓▓▓▓▓▓▓▓▓ →→→→ ▓▓░░░░░░░░░░░ Political
Without intervention, entropy wins.Core Principles
1. Acknowledge the Drift
Accept that without intervention, the company will become slower and more complex.
2. Impose Artificial Constraints
Use binary rules to force simplicity:
- "No meetings before 11am"
- "One feature in, one feature out"
- "All or nothing" policies (no permission management)
3. Calculate Dimensional Complexity
When adding a new product line, factor in that every future decision now requires choosing between A and B.
4. Simplify Binary Decisions
Make policies "all or nothing" (e.g., everyone is an insider) to remove administrative overhead.
How To Apply
STEP 1: Identify Complexity Sources
└── What decisions require the most meetings?
└── Where are processes slowing down?
STEP 2: Create Binary Rules
└── Turn gray areas into black/white
└── "If X, then always Y" rules
STEP 3: Measure Carrying Cost
└── New feature = Dev time + Future maintenance
+ Sales training + Support complexity
+ Every future decision now A vs B
STEP 4: Enforce "One In, One Out"
└── Add a feature? Remove one.
└── Add a product line? Kill one.Common Mistakes
❌ Measuring cost of new features only by engineering hours
❌ Adding "just one more" exception to binary rules
❌ Waiting until complexity is already painful to act
Real-World Example
In the early days, HubSpot enforced a rule where adding a new UI element required removing an existing one to maintain a constant level of complexity.
*Source: Dharmesh Shah, Lenny's Podcast*