Working copy
Draft text, rebuilt from the manuscript on every change. Stubs and source passages are still here, and none of it is final. Back to the working copy contents
Drafted The chapter is written and still being revised.
The bidders: racing for the light
The bidder's question is not the buyer's question. The buyer asks what the work should cost. The bidder asks what it will cost them, and what it will take to win. Most bids lose. Public work runs at about one win in six to ten bids, and estimating is overhead paid out of the jobs that are won. The stake on each is sharp. On a typical public highway job the gap between the lowest and the second-lowest bid is about 5 to 8 percent, and an American contractor's profit after all costs averages about 5 to 6 percent, in the same range. An estimating error the size of the winner's lead is the size of the whole year's profit on that job. Bid 5 percent high and the job goes to someone else, and the cost of bidding is lost. Bid 5 percent low and the firm wins a job that pays nothing.
Who is racing for the light?
The small firm, first, because it is the one the project is not for. When the whole picture, the crews, the equipment, the costs and the appetite for risk, fits in one head, an off-the-shelf AI already gives a bid to review, and a tool with insurance on top would turn profit into overhead. The small firm needs nothing from a standard except to be left alone by it, and not pressed by buyers to keep records it does not need.
The ambitious smaller firm stretching to a bigger job is the exception worth naming. Firms cannot discuss prices with other bidders under collusion legislation, so a smaller ambitious firm would either need to take a massive amount of risk when bidding a bigger project, or use AI to confirm its hand calculation. The check is the second option made respectable. But better estimation and bidding strategies are only part of a firm's capacity to deliver projects. Front-line, supervision, administrative, management and accounting skill all need to grow in order to increase capacity. Bonding companies know this and are unlikely to change their winning strategies.
The medium and large firm is the market. It has several estimating offices, middle managers between the estimators and the people who sign, and a bonding line in the tens of millions. It bids the largest jobs, where a losing bid wastes the most work and where no single person can check everything. What it needs from a check is trust between its own people: a senior manager relying on a bid built in another office, with a record of what was checked.
The largest firms have a different ailment. For them, velocity has been replaced by speed. They bid on a lot, they work on a lot, but their overall company size and the value of work on the books usually plateau. There is simply not much market share left to share, and the velocity of younger firms makes margin growth difficult to sustain.
The surgical team, in Fred Brooks's sense, is the bidder the project is for in the long run. I have mapped out the documentation requirements needed to go from a single small crew, self-employment with assistants, to an actual business, and was discouraged to see how much paperwork is needed to grow. The issue is not so much that there are no intermediate solutions to match organizational sizes. It is that every shortcut adopted while growing is an expensive change once the company starts scaling. AI fixes this. Part of the motivation for the project comes from the need for tools to help surgical teams work together while maintaining their status as principals on larger projects.
And then the new species: firms that had never tendered, now feeding their sales material to a bid-writing tool and submitting to everything. They are the thicket. What they need is a hearing on what they can build, not on how well their tool writes.
What they need
Put plainly, a bidder wants six things from the tender's door.
A check before the bid goes in, because the money is in the gap. A way for managers to trust bids they did not build. Fewer wasted bids, which means clean documents, questions answered before the bids close, and the five losing bids in six costing less to prepare. A fair and predictable hearing, where the same bid gets the same answer and the reasons are given. Room to spread the price their own way, so a check flags a line for the firm's own review and never sets the price. And their own data kept their own: a firm's bid history is its memory, and it will share the market's record, not its own.
A seventh is a bonding company that sees the difference between a firm that checked its bid and one that did not. The bonding market will be the slowest to move, for the reason given above; chapter 13 says where a check could still help a bonding company.
What they fear
The record proves what they knew. A record that proves a clerical slip also proves that the firm's bot saw the error in the drawings and the firm said nothing. Bidders will keep a record only if it is theirs, sealed, and the buyer keeps the same. Chapter 3 is about that.
Everyone leaning on the same bot. The winner's curse works because each bidder's estimate has its own error, and the lowest bidder is often the one whose error ran lowest. If many bidders check their prices against the same reference, and the reference leans one way, their errors stop being independent and the whole market is pulled the same way at once. The standard creates a risk for both sides of the tender: if the reference runs too low, bidders lose as a group; if it runs too high, buyers lose. The answer is a market where several independent choices diffuse the bias, rather than one reference everyone leans on. The lean is chosen by the firm using the tool, as a setting of its own risk appetite, with insurance on the margin as the backstop for the tool's error. A shared reference that moves prices together can also look like coordination when nobody agreed to anything, which is one more reason the check flags and never prices.
Margins squeezed, and judgment made to look like error. A firm that spreads its price for good reasons does not want a buyer's tool calling the spread a mistake.
What a standard does for them
Especially with margin coverage for the software usage, the cost to bid will drop. This would reduce the overhead needs of firms while improving their competition in their niche. The pitch to large companies is a growing pie: faster tenders, faster life cycles, more projects built, and the capacity to hold a multi-billion dollar project in more than one head.
Not every bidder gains, and the chapter should say so. Firms that win on skill and good cost control gain from clean tenders and a fair check. Firms whose margin comes from unclear documents and the claims that follow lose, and they will say so too.
For the surgical team the gain is the status it came for. Cheap paperwork gets a team in the door. A bid the owner can check on the same terms as a large firm's is what makes it a principal rather than a subcontractor, and a public check leaves no shortcut to unwind later when the team grows.
In the thicket, a polished bid is a sign of life, and the buyer has stopped reading signs. A bid that read the same documents, priced the same quantities and pointed its claims at evidence is the bidder showing that it is alive. That is what the race for the light turns into once somebody can measure who reached it.