Listen to the People on the Tools: When the Field Out-Engineers the Drawings

On a high-performance steel job, the drawing is a theory and the erection sequence is the proof. The people on the tools — ironworkers, bricklayers, abatement crews, the foremen who have raised more structure than any one designer has ever drawn — routinely carry knowledge that formal documents miss, under-specify, or get wrong outright. Across structural steel, masonry, demolition, and high-performance residential, Atlantic Canada operators describe the same pattern: drawings reach the field and fail on contact, and the workers executing them correct course through hard-won field intelligence. The fix is not better software or stricter specs in isolation. It is closing the loop between the design room and the field before the paper goes out — treating experienced tradespeople as co-designers, not executors of paper.

Why is a drawing only ever a theory until the field touches it?

A set of stamped drawings looks like a finished answer. In practice it is a hypothesis about how steel, block, and concrete will behave once a crew starts assembling it in weather, on a real grade, against real tolerances. The gap between the two is not rare and it is not small. The Construction Industry Institute estimates that 30 to 50 percent of RFIs trace back to errors, omissions, or ambiguities in the construction documents — meaning the drawings were theoretically complete but not buildable as written. A separate study of more than a million RFIs put the average cost to answer just one at $1,080, with nearly a quarter never answered at all. When the question goes unanswered, the field does not stop; it improvises.

That improvisation is where field knowledge does its quiet work. Tim Houtsma of Marid Industries Ltd, a structural steel fabricator and design-build contractor, frames the relationship between the office and the field as a standing instruction to designers: "pay attention to what the guys outside have to say, because they've put up more steel than you have designed" (Tim Houtsma, EP 1). The point is not that engineers are wrong and tradespeople are right. It is that the two hold different halves of the same knowledge, and only one half is on the paper. The erector knows the sequence, the lift, the connection that looks clean in the model and binds in the field. The designer knows the loads and the code. When those halves never meet before tender, the project pays for the gap later.

The economics of that gap are not subtle. A peer-reviewed analysis found that design changes and planning errors together account for 56.5 percent of cost overruns and 40 percent of project delays on large projects — overwhelmingly a pre-construction document problem, not a field-execution one. The MacLeamy Curve, the design-process argument behind most integrated-delivery thinking, shows why a change is cheap in schematic design and brutal once steel is up: your ability to influence cost falls as the cost of changing anything climbs. Field intelligence applied before the drawings are issued does its work at the cheap end of that curve. The same intelligence applied after erection is a change order.

What does an over-specified drawing actually cost?

The most expensive lines in a drawing set are often the ones that look most conservative. A spec that calls for more material, more coating, or a heavier assembly than the structure needs reads as caution. To a contractor who builds the thing, it reads as money spent for no performance. Andrew Smith, a masonry engineer working with the Atlantic Masonry Institute and the Canada Masonry Design Centre, describes the everyday version: a wall drawn in solid block where the same structural result is available another way. "if you used hollow and grouted solid it'd be cheaper, it'd be easier on the labor" (Andrew Smith, EP 5). Same wall, same standard, lower cost and less strain on a crew — but only if someone who has laid block reviews the assembly before it is locked.

Houtsma's design-build vantage shows why the savings are usually invisible from inside a single trade's drawings. Because his shop fabricates the steel and sees the building as a whole, he can trade cost across the envelope: "it cost me an extra 500 bucks in the roof structure but it saves them a thousand bucks in the walls" (Tim Houtsma, EP 1). A siloed designer optimizing the roof in isolation would never spend the extra $500, because the $1,000 saving lands in a different scope, on a different drawing, in a different budget line. Only a builder looking at the whole envelope sees the net win. This is the structural argument for early trade involvement stated in dollars: the optimization that matters most lives in the seams between disciplines, exactly where a discipline-by-discipline drawing set is blindest.

Marid's prefabrication model pushes the same logic upstream into the shop. "we try to prefab everything as much as we possibly can in a controlled environment in our shop, which reduces… the risks of person-hours on site," Houtsma says (Tim Houtsma, EP 1). When the model and the machine are tied together, the data discipline is total: "if it fits in the model it fits outside," and out of that 3D model the shop produces the CNC files, so "nobody touches any of that data and it goes straight to the machine" (Tim Houtsma, EP 1). That is the design-to-field loop closed by hardware. But it only works because the model already carries the field's knowledge of what fits — knowledge that had to be put there by someone who builds, before the first cut.

What lives behind the drywall that no drawing shows?

Some field knowledge is not about improving a drawing. It is about pricing the things a drawing cannot see at all. Demolition and abatement is the purest case, because the work is defined by what is hidden. Dan Chisholm of Inflector Environmental Services describes estimating a demolition as a gamble — you don't always know how hard a given material is going to come down, or whether a floor is glued rather than nailed, until the crew is in it (Dan Chisholm, EP 34). Encased steel beams that turn up where the drawings showed none, a heritage wooden structure that cannot be touched, conditions sealed inside a wall for fifty years — the document describes the building as designed, not the building as it has aged, been renovated, and been patched.

Inflector's position in the build gives Chisholm an unusual cross-section of where drawings drift from reality. "We're the only sub-trade that touches every scope," he says — roofs, ceilings, mechanical, electrical, concrete, drywall, flooring, all of it removed at some phase (Dan Chisholm, EP 34). A trade that opens up every system sees, in real time, the cumulative gap between every set of drawings and the actual building. That vantage is precisely the kind of tacit knowledge that never makes it back upstream. As one analysis of the model-to-field gap puts it, field innovations and discoveries routinely disappear from the record: the crew solves the immediate problem, but the solution never improves the next drawing set, because no workflow exists to carry it back. The knowledge is used once and lost.

When the unknown cannot be resolved before the bid, the field prices the risk into the number. That is not padding; it is the honest cost of a drawing that cannot describe what it cannot see — and the owner pays it either way, in the bid or in the change order.

In theory the wall performs. In reality?

High-performance residential is where the gap between modelled and actual gets quantified, because the building is supposed to hit a number — an energy target, an airtightness result, an R-value. Casey Grey of The Conscious Builder Inc. frames the entire job as the distance between the model and the lived result: "in theory what's going to happen but in reality this is likely what's going to happen based on our experience" (Casey Grey, EP 22). The qualifier — based on our experience — is the whole argument. The model produces a prediction. Experience produces the correction.

Grey is blunt about manufacturer performance claims that arrive as settled fact. "don't take the word from the ICF company about the fact that they have an R-50 wall assembly, because it doesn't actually calculate that way" (Casey Grey, EP 22). A number printed on a product sheet is a theory like any drawing; the builder who has measured the in-use behaviour of the assembly is the one who knows whether it holds. This is why an integrated design process drags contractors, energy advisors, and architects into the same room before a single wall assembly is finalized. "you need the whole team on board" from the beginning, Grey says, because "everybody brings a different skill to the table and all of those skills are important" (Casey Grey, EP 22). The IDP is the residential-scale version of the same fix the steel and masonry trades are reaching for: get the field's knowledge into the design before the design is frozen.

The formal version of that loop — Integrated Project Delivery, with its multi-party agreements and shared risk — is still rare in Canada and effectively absent in the Atlantic provinces. Only about 60 IPD projects have been initiated nationally, concentrated in Alberta and Ontario, with the Edmonton Mosaic Centre delivered 12 percent below market cost and 29 percent ahead of schedule. Early trade involvement is a core IPD condition. The model proves the principle works; the region has barely tested it.

How much of the real knowledge lives in a foreman's hands?

Ask a masonry contractor where the buildability judgment actually resides, and the answer is not the drawing and not the engineer. It is the foreman. Darrell Jerrett of Darim Masonry Limited is direct about it: "we lean on those guys heavily, and the collaboration from our experience" (Darrell Jerrett, EP 5). The foreman is the knowledge node — the person who has laid the pattern, hit the tolerance problem, and learned which detail the drawing got wrong before it cost anyone money. That knowledge is not written down anywhere. It lives in hands and memory, surfaced only when an experienced person looks at a drawing and feels what will not work.

The estimating table is where that judgment either shows up or bites. Jerrett describes the moment of catching a missing requirement on a stack-pattern veneer bid — the kind of reinforcing spec that, if missed, turns a clean number into an exposed one: "oh freak i didn't bring that i didn't carry that in my numbers" (Darrell Jerrett, EP 5). Catching it meant a call to the institute's technical line and a corrected bid; missing it would have meant eating the cost. When the unknown could not be fully resolved, the answer was the same one Inflector reaches for: "we had to take a strong number on it to make sure that we were covered" (Darrell Jerrett, EP 5). The pattern repeats across every trade in this essay. Where the drawing is silent or wrong, field experience either corrects it or prices it — and on a hard-bid, stipulated-price contract, getting that wrong is the contractor's loss alone.

Andrew Smith's account of arriving at a site closes the loop on why this matters before tender, not after. Crews "get on a job site, look at the drawing, see that it's just not constructable" (Andrew Smith, EP 5). At that point the only paths are an RFI, a change order, or a field fix proceeding at risk — and Canadian practice makes the last one genuinely risky. A field solution that improves on a stamped drawing but was never formally approved leaves the contractor in a legal grey zone: the engineer's stamp does not transfer to a field variation, and the contractor typically cannot recover the cost of the improvement without a prior approved RFI. The field carries both the knowledge and the liability for using it.

What gets depreciated when the trades retire?

There is a clock on all of this. The field knowledge that corrects the drawings is concentrated in a generation that is leaving. The average bricklayer is 53. Nationally, 24.6 percent of Red Seal trades workers are 55 or older, and the youth cohort shrank while the 55-plus cohort grew between 2016 and 2021 — a structural aging, not a cyclical one. In the Atlantic region specifically, roughly 23 percent of the construction workforce is expected to retire over the next decade, with New Brunswick alone projecting 6,500 retirements against a need to hire up to 8,400 by 2034.

The transfer mechanism is fragile. Roughly 80 percent of a Canadian apprenticeship is on-the-job training under a journeyperson, which means the knowledge moves person to person or not at all. When the journeyperson retires without a debrief, the only record of their field solutions is the muscle memory of whoever stood next to them. The productivity numbers already register the loss: labour productivity fell in every Atlantic province between 2020 and 2024, down as much as 16.4 percent in Newfoundland and Labrador, with the Job Bank explicitly attributing part of it to experienced workers leaving and taking their experience with them. The field intelligence that out-engineers the drawings is the real asset being depreciated, and unlike steel or equipment it does not show up on a balance sheet until it is gone.

What does it look like when the loop actually closes?

The fix is not exotic. It is design-build, integrated design processes, trade advisory committees rewriting college programs, and industry institutes standing up technical lines that a foreman can call mid-bid — every one of them a mechanism for moving field knowledge into the design before the drawings are frozen. Marid's model-to-machine pipeline shows the loop closed by hardware. Grey's IDP shows it closed by process. The masonry institute's technical line shows it closed by a phone call. The common denominator is that the field is in the room before the paper is final, not after the steel is up.

The region is beginning to build the scaffolding for it. The Atlantic Construction Alliance, formed in 2026 by eight provincial associations, has named procurement modernization as an early mandate, and relationship-based contracting is now a live procurement-reform conversation in Canada. Atlantic Canada's repeat-client culture — the same crew, the same GC, the same owner — is exactly the substrate on which tacit knowledge has always traveled. Formalizing it is less an invention than a recognition of how the good work already gets done.

The template for the close is the field proving the number itself. When contractors fund the research on a load-bearing masonry approach, prove the figures, and build the demonstration project, the field is no longer correcting the drawings after the fact — it is generating the engineering that future drawings will cite. That is the whole argument in one move: the people on the tools are not executing the theory, they are writing it. The on-the-record takeaway from the operators across these episodes is consistent and unglamorous. Get the field into the room first. The crew has put up more structure than the drawing has ever described — and the cheapest place to learn what they know is before the paper goes out, not after it fails on contact.

Related: the specialty trades and field operations hub, and the structural-systems guides on tilt-up vs wood vs mass timber vs steel in Atlantic Canada and hurricane-resistant homes and tilt-up construction.