Constraint relaxation is a problem-solving move: temporarily drop or loosen a constraint to make a hard problem tractable, then reintroduce the constraint and adapt the solution. In algorithms, the relaxed problem (e.g. allowing fractional solutions where integers are required) often yields structure — bounds, dual values, or a feasible rounding — that guides the full solution. In strategy and product, the same move breaks logjams: "What if we didn't need to be in every market?" "What if we could ship in six months instead of two years?" The relaxed problem is easier to solve; the answer often reveals what the binding constraint really was.
The power is in the ordering. Solve the relaxed problem first. Get a solution, or a proof that even the relaxed problem is impossible. Then tighten. If the relaxed solution happens to satisfy the original constraints, you're done. If not, you've learned which constraints bite and where to focus — redesign, negotiate, or accept a worse objective. The mistake is to treat all constraints as equally sacred from the start. Many "impossible" problems become possible when one constraint is relaxed; many "necessary" constraints turn out to be habits.
In practice, constraint relaxation shows up in resource allocation (relax the budget, see what you'd do), in design (drop "must work offline" and see what the product becomes), and in negotiation (what if the other side got X?). The discipline is to make relaxation explicit: name the constraint you're dropping, solve the simpler problem, then decide whether to reimpose the constraint or to change the real-world situation so the relaxed solution becomes feasible.
Section 2
How to See It
You're using constraint relaxation when you simplify a problem by ignoring one or more limits, solve that version, and then bring the limits back. The diagnostic: are we stuck because we're treating a constraint as fixed? What would we do if it weren't there?
Business
You're seeing Constraint Relaxation when a team says "we can't hit the deadline with this scope" and someone asks: what if the deadline moved? What if the scope did? Solving the relaxed problem (fixed scope, flexible time; or fixed time, flexible scope) reveals the true trade-off and often a feasible path.
Technology
You're seeing Constraint Relaxation when an optimisation problem is too hard in its exact form. You relax integrality (allow fractional allocations), or relax a non-convex constraint, solve the relaxed problem, and use the result to branch, round, or bound the original. LP relaxations of IPs are the classic example.
Investing
You're seeing Constraint Relaxation when you ask "what would this company do with unlimited capital?" or "what if regulation didn't exist?" The relaxed answer shows the unconstrained strategy; comparing it to the actual strategy reveals which constraints are binding and how much they cost.
Markets
You're seeing Constraint Relaxation when you model a market without a key friction (e.g. transaction costs, capacity limits). The relaxed equilibrium is a benchmark; the gap between it and reality is the value of the constraint. That gap drives trading and policy.
Section 3
How to Use It
Decision filter
"When stuck, list constraints. Pick one and relax it — solve the problem without that constraint. If the relaxed solution satisfies the original problem, you're done. If not, you've learned what's binding. Then reimpose, negotiate, or redesign."
As a founder
Roadmaps and strategy are full of constraints — time, money, headcount, compatibility. When you're blocked, relax one. "What if we shipped in 8 weeks with half the features?" The relaxed plan is a concrete alternative. Often the "must-have" constraint is negotiable (with customers, board, or yourself). Constraint relaxation turns "we can't" into "we could if X changed."
As an investor
When a company says "we're constrained by X," ask: what would you do if X weren't? The answer shows the real strategy and ambition. If the unconstrained strategy is compelling, the question is whether X can be relaxed (funding, regulation, talent). If the unconstrained strategy is weak, the constraint is an excuse.
As a decision-maker
Use constraint relaxation in planning and negotiation. Name the constraint that makes the problem hard. Solve without it. The relaxed solution is a reference point — sometimes achievable by changing the world (budget, timeline, scope), sometimes a bound that shows what you're giving up by accepting the constraint.
Common misapplication: Relaxing the wrong constraint. If you drop the one constraint that defines the problem, the relaxed problem is trivial or irrelevant. The skill is picking a constraint that, when relaxed, yields useful structure or a feasible path back.
Second misapplication: Forgetting to re-tighten. Relaxation is a tool, not the answer. You must bring the constraint back and either find a solution that satisfies it or explicitly decide to change the situation so the relaxed solution applies.
Musk repeatedly relaxes "industry" constraints — that rockets must be expendable, that cars must be assembled in one place, that timelines are fixed. By solving the relaxed problem ("what if we could reuse the booster?"), he gets a target and then works backward to make the constraint satisfiable or to change the constraint (e.g. regulatory, supply chain). Constraint relaxation is embedded in first-principles reasoning at Tesla and SpaceX.
Netflix relaxed the constraint that content had to be licensed from incumbents. The relaxed strategy: produce our own. That led to originals and a different cost and control structure. The constraint "we need studios' content" was re-examined; relaxing it defined the next phase of the company.
Section 6
Visual Explanation
Constraint Relaxation — Original feasible set (small) is hard to optimise over. Relax one constraint; feasible set grows. Solve the relaxed problem. If the optimum falls in the original set, done. If not, use the relaxed solution to guide search or to renegotiate the constraint.
Section 7
Connected Models
Constraint relaxation connects to how we simplify problems, trade off objectives, and find binding limits.
Reinforces
First Principles Thinking
First principles thinking strips away assumptions to rebuild from fundamentals. Constraint relaxation is the same move applied to constraints: strip one away, see what the problem becomes, then add it back with clarity about its cost.
Reinforces
Trade-offs
Trade-offs are the cost of satisfying one constraint at the expense of another. Relaxation makes trade-offs explicit: when you relax constraint A, you see what you gain; when you reimpose it, you see what you give up. The two models work together to prioritise which constraints to challenge.
Tension
Necessity & Sufficiency
Necessity: you need X for the outcome. Sufficiency: X is enough. Relaxation tests necessity — if we relax X and still get a good solution, X wasn't necessary. If the relaxed problem has no good solution, the constraint may be necessary. The tension: some constraints are necessary for the definition of the problem; relaxing those changes the problem, not just the difficulty.
Tension
Theory of Constraints
Theory of Constraints says: find the bottleneck and fix it. Relaxation says: remove a constraint and see what happens. The tension: TOC assumes the constraint stays; relaxation temporarily removes it. Used together, relaxation identifies which constraint is the bottleneck by seeing which one, when relaxed, changes the solution most.
Section 8
One Key Quote
"The great watershed in optimization is the transition from the impossible to the possible — and the key is often to relax the problem until it becomes tractable, then to tighten back."
— George Dantzig, on linear programming
Dantzig's work on linear programming and duality formalised the idea that relaxed problems give bounds and guidance. The practitioner's move: relax to get a solution or a proof of impossibility, then tighten. The strategic move: relax constraints in the world (scope, timeline, budget) when the relaxed solution is worth fighting for.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
Most "we can't" is "we won't relax a constraint." Time, scope, budget, or "how we've always done it" — one of these is usually the blocker. Make the constraint explicit. Solve the problem without it. The relaxed solution is either feasible (so the constraint was optional) or it shows what you're sacrificing. That clarity is worth more than another round of grinding on the constrained problem.
Use it in negotiation. When the other side says "we need X," ask: what would you do if you didn't get X? The relaxed answer reveals their priorities. When you're stuck, offer to relax something you control (timeline, scope) in exchange for something they control. Constraint relaxation is a shared tool for finding feasible deals.
Relax the right constraint. Relaxing the core objective (e.g. "what if we didn't need revenue?") is useless. Relax a constraint that might be habit, convention, or negotiable. The best candidates are ones the organisation treats as fixed but that aren't physically or legally fixed.
Re-tighten explicitly. Relaxation is a step, not the end. Either the relaxed solution satisfies the original (adopt it), or you reimpose the constraint and use the relaxed solution as a bound or a guide. Document what you relaxed and why you're reimposing or not. Otherwise the move is forgotten and the same blockage returns.
Section 10
Summary
Constraint relaxation is solving a problem by temporarily dropping or loosening a constraint, then re-tightening or changing the situation. It makes hard problems tractable and reveals which constraints are binding. Use it when stuck: name the constraint, solve without it, then adopt the relaxed solution or use it to guide the constrained solution. Relax negotiable or conventional constraints; re-tighten with intention.
Thiel's "what valuable company is nobody building?" is a form of constraint relaxation: relax the constraint that you must compete in existing categories.
Leads-to
[Optimization](/mental-models/optimization)
Optimization is the discipline of finding the best solution in a feasible set. Constraint relaxation is a standard technique: relax to get bounds and structure, then use that to solve or approximate the original. Relaxation is a tool inside the optimizer's toolkit.
Leads-to
[Inversion](/mental-models/inversion)
Inversion: consider the opposite. "What would make this fail?" Relaxation inverts a constraint: "What if this constraint didn't hold?" Both are ways to see the problem from a different angle and to discover what's really binding.
