Highway megaprojects are documented far better than they are described. A build large enough to make national news is almost certainly producing an annual financial plan, a risk register, a probabilistic cost review and a procurement record, most of it public. Very little of that reaches the coverage. What reaches the coverage is a single number and a completion year, usually stripped of the two things that make either one meaningful: the date the estimate was struck and the boundary drawn around what it counts.
This section works the other way round. Each case study starts from the paperwork the agency that owns the project has already filed, quotes the figure with its vintage attached, and says plainly which accounting the figure belongs to. Where an estimate has changed, the change is treated as the subject rather than as an embarrassment, because the reasons estimates move are the most transferable knowledge available about how these projects work.
The accounting problem comes first
Nothing in federal law standardises what a megaproject total contains. The statutory machinery requires large recipients of federal aid to file management and financial plans, and federal guidance sets useful conventions about confidence levels and escalation, but no rule fixes where a project ends and a program begins. The result is that four current builds publish four incompatible kinds of number, which is the argument taken up in the survey of the largest current builds. One is a contracted total precise to the dollar. One is a program name covering two decades of procurements not yet let. One is a most-likely value inside a stated probability range for a corridor that includes light rail. One is the first phase of an eight-mile scheme whose later phases are deliberately unpriced.
None of those is dishonest. Ranking them is.
Case studies, and what each one is chosen to show
The Brent Spence Bridge Corridor is the clearest current example of a project whose public case had to be rebuilt after its vocabulary expired. The two federal condition labels that once sorted American bridges into national repair categories have both been retired, and the two states now say directly that the existing structure is sound and will stay in service. The argument for spending billions there rests on geometry and capacity, and the finance rests on a political commitment to build without tolls, which converts the residual into two state capital programs and explains why the published price covers a first phase only.
The Hampton Roads Bridge-Tunnel expansion is the opposite kind of document. Virginia’s annual plan shows its working: a cost estimate review with a named date and a stated percentile, a bid that landed within one percent of it, a contingency and an incentive identified separately, and a risk register that prices the ground through a geotechnical baseline report. It also contains a paragraph in which the department records its concern that its own contractor is failing to hit production rates. Reading that alongside the tunnelling method, the sixteen years of environmental restarts and the eight-lane reconfiguration is a better education in megaproject delivery than any summary.
The Big Dig is here because it is the origin of both regimes the other two operate under. The annual financial plan requirement, the probabilistic review convention and the national tunnel inspection standards are all traceable to specific Boston failures with dates attached. It is also the case where folklore has done the most damage, because the cost comparison that travels furthest sets an unescalated 1982 figure against an escalated one from 2003. The constant-dollar version of that comparison is public, and it tells a different story.
The questions this section keeps returning to
Three questions recur across every project studied here, and the answers keep moving.
The first concerns when a number deserves belief. An estimate produced before a design exists is a different object from one produced after a risk-based review against a developed design, even though both are quoted the same way. The reforms of the last twenty-five years improved the second kind and left the first largely untouched.
The second is who carries the risk of the ground and of the market. Delivery method decides this, and the answer shows up in unglamorous contract features: baseline geotechnical reports, commodity price adjustment clauses, contingency ownership, completion incentives that only some funding sources will reimburse. These clauses move more money than most design decisions do.
The third is what happens to the parts nobody funded. Corridors are announced whole and built in pieces, and the piece that gets built first is the one with the strongest benefit case and the clearest money. Whether the rest is ever priced is a fair test of a program, and it is a test that takes a decade to run.
Case studies are added here as the documents warrant rather than as projects make news, since a project that has just published a financial plan update is worth more attention than one that has just published a rendering.