Letter vs intent: prescriptive & performance-based compliance

There are two ways to be compliant — follow the recipe, or prove you met its purpose — and both are only as good as what you wrote down.

The idea

A code or standard has a letter (do exactly this: these dimensions, this count, this distance) and an intent (achieve this outcome: everyone gets out safely in time). The prescriptive route follows the recipe. The performance route demonstrates, usually by modeling, that an alternative means meets the same intent — an equivalency or alternative-means-and-methods case.

Both are legitimate. But whichever you choose, the decision that actually protects you is the write-up: the interpretation you made, the option you chose, and why. An undocumented compliance decision is indefensible years later, when a reviewer or a court asks how you got here.

Fire-egress puzzle · two routes to compliant

Scenario (illustrative numbers). A 300-seat community hall in a heritage building. Its far corner sits 250 ft from the nearest of its two exits — past the recipe’s 200 ft travel-distance limit. A third exit would mean cutting a load-bearing historic wall. Make it compliant.

Check the fixed dimensions against the recipe

With only two exits, one box fails. The recipe has a fix — but it costs you a load-bearing wall.

Ready to run the evacuation model.

Design levers — tune until egress beats the intent target

occupants: 300

each exit: 32 in

Reduce required time (RSET)

Extend available time (ASET)

detection pre-movement travel + flow | = time to untenable (ASET)

RSET (required)

—

ASET (available)

—

Margin

—

Demonstration

—

Press run evacuation to model egress. The intent target: everyone out before conditions become untenable, with margin to spare.

Documentation tray

Nothing recorded yet. Turn on “write it down as you go” to capture the interpretation, the option, and why.

How it works

The performance route replaces a fixed dimension with a modeled outcome. The standard framing is two clocks: ASET, the available safe egress time before conditions become untenable, and RSET, the required safe egress time for everyone to get out. You’re compliant with the intent when RSET clears ASET with margin.

RSET = detection + pre-movement + travel   (travel = walking + flow through exits)

worked (illustrative):
  detection      fast alarm ............. 0.5 min
  pre-movement   voice + trained staff .. 1.0 min
  walking        250 ft / 200 ft-per-min  1.25 min
  flow           220 occ / (2 x 48in x 1.2/in-min) = 220 / 115  = 1.91 min
  ------------------------------------------------------------
  RSET = 0.5 + 1.0 + 1.25 + 1.91 ......... = 4.66 min

ASET = base + sprinklers + smoke control
  ASET = 4.0 + 2.5 + 1.5 ................. = 8.00 min

  margin = ASET - RSET = 8.00 - 4.66 = 3.34 min  ->  demonstration passes.

Notice: pre-movement, not walking, is the big controllable chunk.
A faster recipe-beating design is real engineering — but the file is
what makes it defensible five years on.

When to use it

Reach for…	…and the trade-off
Prescriptive when the recipe fits and is cheap to meet.	Fast, low-argument, self-evidencing — but rigid; it can force changes a building can’t take (a heritage wall).
Performance when the recipe can’t be met or is wasteful for this case.	Flexible and often better-engineered — but it needs modeling, assumptions, review, and a defensible write-up.
Either way, always record the interpretation, option, and rationale.	Costs a little time now; it is the only thing that survives to defend the decision later.

Watch out for

Treating “performance-based” as “we can skip the requirement.” You’re not skipping the intent — you’re proving you met it. No demonstration, no equivalency.
Undocumented assumptions. The fire size, the occupant load, the pre-movement time, the walking speed — if the inputs aren’t written down, the model can’t be reproduced or defended.
No margin. A demonstration that scrapes past with RSET just under ASET leaves nothing for the day reality is worse than the model. Carry a safety margin, and record it.
Silent interpretation. Deciding what a clause “means” and never noting it. Years later no one can tell an interpretation from an oversight.
Letting the recipe win by default. Meeting the letter but obviously missing the intent (a compliant-on-paper exit that opens onto a hazard) is a failure the checklist won’t catch.

Worked example

An interviewer asks: “The prescriptive travel-distance limit can’t be met in a heritage building without demolishing a load-bearing wall. What do you do?” A strong answer names both routes and then the record. You’d say: the recipe has a fix — add an exit — but it’s infeasible here, so you pursue a performance-based alternative-means case: model the evacuation, show RSET beats ASET with margin using sprinklers, fast detection, a voice alarm to cut pre-movement, and a capped occupant load. Crucially, you’d add: “and I’d document the interpretation of the requirement, the alternative I chose, the model inputs and their sources, the margin, and the reviewer sign-off — because in five years that file is the only thing that defends the decision.” The interviewer is listening for both the engineering and the paper trail; strong candidates never separate them.

Check yourself

Your performance model cleanly shows RSET well under ASET, the reviewer verbally agreed, and the building opened. Three years later a claim arrives and you find nothing was written down. Where do you stand?

What does the performance route actually let you replace?