Every project of any consequence is governed by trade-offs among scope, schedule, and cost. The classical formulation—the Iron Triangle of project management—holds that these three parameters are interlinked: change one and at least one of the others must move with it. Generations of project managers have been taught the popular shorthand: fast, good, cheap—pick two. It is intuitive, it is memorable, and it is wrong in an important way. Pick-two is an improvement over the fantasy that all three can be maximized at once, but it narrows the menu without picking from it, and it leaves the un-chosen parameter as a silent bill-payer. The mathematically sound approach is to fuse the parameters into a single composite objective and optimize against that. What follows is an argument for that progression, with NASA’s Faster, Better, Cheaper program as the cautionary tale of what happens when an organization refuses to choose at all.
Pick Three: The Incoherent Default
The most common failure in project framing is not picking two corners of the triangle. It is refusing to acknowledge that a triangle exists. Leaders announce that a project will be delivered faster than the last one, with higher quality, at lower cost—and they treat these as three independent dials that can all be turned up simultaneously with sufficient cleverness, motivation, or process reform. This is the pick-three posture. It is rhetorically appealing because it flatters everyone in the room: no one has to give anything up.
From an optimization standpoint, pick-three is incoherent. An optimization problem requires an objective function: a rule that lets you compare two candidate plans and decide which is better. That function must reduce to a single scalar quantity (Martins and Ning 2022). The moment you elevate two or more parameters to the status of co-equal objectives, you generate pairs of options where one beats the other on one axis and loses on the second, and the framework has nothing to say about which to pick. What you have is not an optimization problem. It is a values problem masquerading as one—and because it is unacknowledged, the values get assigned by whoever happens to be applying pressure at the moment of decision.
NASA’s Faster, Better, Cheaper program is the canonical example. Championed by Administrator Daniel Goldin in the early 1990s, FBC promised that missions would be delivered more quickly, on smaller budgets, without sacrificing quality (McCurdy 2001). Between 1992 and 2000 the agency launched dozens of FBC missions. Some succeeded—Mars Pathfinder is the most-cited—but the program collapsed amid high-profile failures, most famously the back-to-back losses of the Mars Climate Orbiter and Mars Polar Lander in 1999. The standard explanations cite organizational culture, inadequate testing, and political pressure. All real. But beneath them is a prior, structural problem: FBC named three parameters and elevated all of them to the status of objectives without ever defining what the agency was actually optimizing.
The Mars Climate Orbiter case is illustrative. The proximate cause was a unit-conversion error—pound-seconds versus newton-seconds—that sent the spacecraft on a fatal trajectory. The deeper question is why a problem this basic was not caught. Post-failure reviews documented that cross-team integration testing and independent review had been pared back as part of cost and schedule reductions. Had the agency been explicitly optimizing for, say, probability of mission success at a given budget, those reviews would have been recognized as cheap insurance against losing the entire investment. Instead they looked like budget items to trim, and they were trimmed. Mars Polar Lander, lost the same year, told the same story: another routine failure mode that integration testing was designed to catch, and another set of tests cut without countervailing pressure to preserve them.
FBC was not undone by any single bad decision. It was undone by the absence of a rule for making decisions. When three parameters are all called goals, the parameter hardest to measure—usually quality—becomes the silent bill-payer. That is the pick-three failure mode in compressed form.
Pick Two: An Improvement, but Incomplete
The standard corrective is the familiar shorthand: fast, good, cheap—pick two. This is genuine progress. It forces the organization to acknowledge that the parameters are coupled and that improvement on two of them will be paid for on the third. The trade-off is now explicit, which is the entire reason pick-two has survived as a teaching tool for decades.
But pick-two is only half of a method. It tells you which corner of the triangle to face; it does not tell you how far to go along the edge you have chosen. “Faster and better, accept higher cost” still leaves infinitely many points to choose from, ranging from a modest gain at modest extra cost to a dramatic gain at ruinous cost. Pick-two narrows the menu. It does not pick from it.
There is also a subtler problem. Pick-two encourages the fiction that the un-picked parameter is fixed, when in practice it is merely deprioritized. A project that picks fast and cheap has not nailed quality to a stake; it has declared quality the variable that will absorb whatever pressure the other two cannot. Without an explicit rule for how much erosion is acceptable, the deprioritized parameter becomes the silent bill-payer—the same failure mode as pick-three, just reached more slowly.
The Single Objective Function: The Mathematically Sound Choice
The rigorous answer is to consolidate what the organization cares about into a single composite objective and optimize against it. Return on investment fuses cost and outcome into a single ratio. Probability of successful launch fuses quality, schedule discipline, and risk into a single number that goes to zero the moment any one of them fails badly enough. Cost per kilogram delivered to orbit does the same work for a launch business. The trade-offs among scope, schedule, and cost still exist; they become internal to the objective function, weighted explicitly, rather than left to ad hoc judgment in the moment.
A single objective makes trade-offs comparable, exposes hidden disagreements early, and survives delegation. The project lead asking “should I cut two weeks of integration testing to hit the launch window?” has a clear question with a clear answer rather than a values question to escalate. The executive and the team who both say “we want all three” are forced to discover whether they actually agree, while it is still cheap to find out. And the hundreds of small decisions made by people who will never be in the room with senior leadership all resolve against the same yardstick.
Most importantly, a single objective defends against the silent bill-payer problem in a way pick-two cannot. With probability of successful launch as the objective, a cheap, fast mission that fails at Mars scores zero—the cost-cutting that destroyed it is mathematically visible, not buried in a parameter no one was tracking. With ROI as the objective, a product shipped on time at low cost that nobody will buy scores zero in the same way. You cannot quietly trade away something that sits inside the function you are trying to maximize.
The honest limitation is that this approach forces you to do something hard: actually decide what you care about and how to weight it. Pick-two lets you avoid that decision, which is why it remains popular despite being analytically weaker. The trade-offs do not disappear because you decline to define them. They get made anyway, by whoever has the most leverage in the moment, and the cost lands wherever it lands. Pick-three is a denial of the problem; pick-two is a description of it; a single objective function is an actual solution.
What Single-Objective Framing Looks Like in Practice
The Apollo program is the cleanest historical example. The composite objective was unmistakable: land a man on the Moon and return him safely before the end of the decade. Schedule and crew safety were built into the success criterion itself—a late landing or a dead astronaut both counted as failure. Cost was an input the program drew on as needed. There were certainly internal trade-offs and bitter arguments, but they all resolved against the same yardstick.
SpaceX’s reusable-rocket program has run on a similar principle, with cost-per-kilogram-to-orbit as the de facto objective. Schedule slips and early test failures were tolerated because they served the real goal: lower the unit cost of access to space, durably. Both organizations committed to a single objective they sought to maximize and used that commitment to make consistent trade-offs across thousands of decisions made by people who never had to ask permission. That is what project control actually looks like when it works.
Conclusion
The Iron Triangle is real. Scope, schedule, and cost are coupled, and pretending otherwise is the most common form of project mismanagement. But the popular response—pick two—is a half-step. It acknowledges the trade-off without resolving it, and under sustained pressure it tends to collapse back into the pick-three failure it was meant to correct. The mathematically sound move is to fuse the parameters of interest into a single composite objective and optimize against that, with the trade-offs explicitly weighted inside the function rather than left to whoever is in the room when the next hard call comes.
FBC’s failure was not that it picked the wrong objective; it was that it never picked one. Any organization that names multiple parameters as goals without consolidating them into a single objective function is setting itself up for the same failure mode: trade-offs made implicitly, by the path of least resistance, with the cost landing on whichever parameter is hardest to measure until it is too late. Pick three and you are lying to yourself. Pick two and you are partway home. Pick one composite objective and you have something the organization can actually be steered by.
References
Martins, J., and A. Ning. 2022. Engineering Design Optimization. Cambridge University Press. https://mdobook.github.io.
McCurdy, H. E. 2001. Faster, Better, Cheaper: Low-Cost Innovation in the U.S. Space Program. Baltimore: Johns Hopkins University Press.
First published on murraycantor.com, May 2026. Listed with the other articles on the writing page.