← Reference · Nestor G Pestelos Jr

Systems Theory · Operations Research

Theory of Constraints

Reference entry · last updated August 26, 2026

The Theory of Constraints (TOC) is a management and systems engineering methodology asserting that any manageable system is limited in achieving higher throughput by a very small number of constraints, typically exactly one active bottleneck at any given time.[1] Formulated by physicist Eliyahu M. Goldratt in 1984, TOC establishes that system optimizations made anywhere other than the active constraint generate excess work-in-progress inventory without increasing finished system output.[2]

Core axioms and accounting definitions

Goldratt redefined operational performance around three fundamental metrics:[1]

• Throughput (\(T\)): The rate at which the system generates money or finished value through sales and completed deliveries. Value generated but unreleased is not throughput.
• Investment / Inventory (\(I\)): All the money and resources currently tied up inside the system (including work-in-progress code, pending pull requests, and un-merged branches).
• Operating Expense (\(OE\)): All the money the system spends to turn inventory into throughput.

The Five Focusing Steps

TOC outlines a systematic cyclical method to continually expand system capacity:[2]

1. Identify the constraint: Locate the specific stage or resource with the lowest capacity that paces the entire system.
2. Exploit the constraint: Ensure the constraint operates at 100% efficiency without idle time, waste, or performing low-value tasks.
3. Subordinate everything else: Align the operating pace of all non-constraint stages to the capacity of the constraint. Producing faster than the bottleneck creates harmful inventory build-up.
4. Elevate the constraint: Invest capital, tooling, or personnel to expand the physical capacity of the bottleneck stage.
5. Prevent inertia (Repeat): Once a bottleneck is elevated, the constraint inevitably shifts elsewhere. Return to Step 1 and avoid allowing organizational habit to dictate the new operating model.[2]

Drum-Buffer-Rope synchronization

To control flow and prevent queue explosion, TOC implements the Drum-Buffer-Rope (DBR) architecture:[3]

• The Drum: The constraint stage sets the production beat (tempo) for the entire organization.
• The Buffer: A bounded queue of work maintained immediately upstream of the constraint to prevent it from ever starving due to upstream volatility.
• The Rope: A signaling mechanism that limits the release of new work into the start of the pipeline strictly to the rate at which the constraint finishes work.[3]

Bottleneck migration in software engineering and AI

In software delivery pipelines, generative AI coding agents drastically increase authoring speed (upstream production). However, without expanding automated verification and maintainer review capacity (downstream constraint), code generation simply accumulates unreviewed pull requests, increasing cycle times and production error rates.[4] Real delivery throughput improves only when automated testing and verification tools elevate the review bottleneck.

See also

• Throughput
• Amdahl's Law
• Plan-Approve-Execute

References

1. ↑ Goldratt, Eliyahu M., and Jeff Cox. The Goal: A Process of Ongoing Improvement. North River Press, 1984.
2. ↑ Goldratt, Eliyahu M. Theory of Constraints. North River Press, 1990.
3. ↑ Dettmer, H. William. Goldratt's Theory of Constraints: A Systems Approach to Continuous Improvement. ASQ Quality Press, 1997.
4. ↑ Faros AI. "The 2026 Engineering Productivity and AI Acceleration Report," 2026.