B Team thesis / 001

Our
thesis.

Why construction has better tools, more information, and the same persistent inability to predict what a project will cost—or when it will finish.

00 / The beginning

Construction often begins when someone decides they need a building or space created.

They seek out general contractors, who offer price predictions that are often more persuasive than precise. The client typically chooses the contractor whose estimate best aligns with their own mental accounting—what they believe the project should cost—regardless of how grounded that estimate may be in actual scope or risk.

01 / Construction history

The agreeable estimate

Design quality can vary. The pattern at bid time usually does not.

Owners can hire a design team independently or through a design-build approach. They can pursue top-tier architects and engineers or choose more economical options. Regardless of the design team’s quality, the selected general contractor will almost always underestimate final cost and construction duration—often because of optimism, competitive pricing pressure, or flawed forecasting methods.

THE SELECTION LOOP
  1. Owner forms a target
  2. Contractors shape a story
  3. The most agreeable estimate wins
  4. Reality arrives later

02 / Technology

Better tools. Same system.

Innovation has always been part of construction. Prediction has not.

Power tools, cranes, scheduling software, and paper-backed drywall each promised greater efficiency and control. Yet a century of new tools has not solved the deeper systemic problems in planning and management.

Robotics may be the next impressive tool that changes how work is performed without materially changing our ability to predict cost, duration, or outcomes. Better power tools are useful. They are not a better operating model.

Hand toolsPower toolsSoftwareRobotics?

03 / General contractors

The marketplace without a feedback loop

General contractors function like marketplaces—but without the intelligence that makes modern marketplaces improve.

They match owners with subcontractors, coordinating specialized labor, material, and services. Unlike modern marketplaces driven by data science, matching algorithms, and feedback loops, construction is commonly managed through heuristics, intuition, and manual oversight.

The network does not learn. More projects do not reliably make the system smarter. Fragmentation persists, productivity stagnates, and general contractors operate on razor-thin margins.

04 / Project pricing

A stack of agreeable guesses

A $100 million estimate is not one precise calculation. It is an assembly of interpretations.

General contractors compile and select from subcontractor estimates, add supervision, fee, taxes, insurance, and contingency, then present the result as a confident prediction. The owner often selects the contractor using similar logic: choosing the story that best fits an internal model of what the building should cost.

When the true cost, scope, or duration is misrepresented during contracting, the delivery network begins with broken nodes already built in. Once reality diverges from the assumptions, predictable breakdowns are treated like unpredictable anomalies.

Trade interpretationsSupervision assumptionsRisk allowancesFee + insurance + tax“The Price”

05 / Failure to estimate

Failure moves downstream

A deductive estimating mistake wins work. An additive mistake quietly loses it.

Subcontractors often rely on a single person’s interpretation of the drawings. If that interpretation misses scope and produces a low number, the quote can look attractive and be accepted. An additive mistake is usually filtered out when the bidder loses—so only one kind of error survives the selection process.

When the field teams discover the mistake, scope debates, change-order conflict, urgent accountability meetings, and pressure follow. The people doing the work inherit the consequences of a decision they did not make.

“Accountability” applied downstream cannot repair a selection decision made upstream.

Evidence register / External records

Failures are not unknowable.

Public records show recurring problems involving estimates, expected costs, financial reporting, and executive response. These links lead to the original reporting or government record; allegations and outcomes should be read in their source context.

ER-01

Fluor

Optimistic cost and timing estimates

SEC ↗ER-02

Granite

Financial reporting and expected-cost controls

SEC ↗ER-03

AECOM

Claims involving disaster-recovery estimates

DOJ ↗ER-04

AECOM unit

Whistleblower litigation and billing practices

ENR ↗ER-05

Tutor Perini

Executive response following a quarterly loss

ENR ↗ER-06

KBR

Whistleblower confidentiality agreements

SEC ↗

The unresolved measure

More capability.
Less certainty.

Authoritative sources—including the U.S. Bureau of Labor Statistics and Project Production Institute—document the industry’s uneven productivity performance. The exact comparison changes with the dataset and period. The larger point remains: decades of improved tools have not created a dependable system for predicting project outcomes.

BLS construction labor productivity ↗Project Production Institute ↗

06 / Paradigm shift

Leave the cave

Construction keeps funding the shadows.

Like the prisoners in Plato’s Allegory of the Cave, the industry can mistake familiar sayings, inherited practices, and confident estimates for reality. Even construction technology investment often reinforces legacy behavior instead of changing the system that produces the outcome.

B Team exists to build a different operating model—using data science, predictive modeling, and cognitive workflow design to replace intuition with insight and guesswork with measurable evidence.

HeuristicsEvidencePrediction
Build what comes next ↗