Answers: why don't your fixes stick?
Payload: a system's output is set by one constraint; optimising every part is not
optimising the whole; local efficiency can actively harm global output.
Hands off with: "So — find the constraint. But how? That's a thinking problem."
Title candidates: "Everybody's Busy. Nothing Moves." · "Where It's Actually Stuck" ·
"The Weakest Link" · "One Thing Is Stuck"
Interactive beats (3):INT-1 the hire · INT-2 the busy question · INT-3 the migrating constraint.
ACT 0 · The clinic (cold open — no theory, no vocabulary)
#
Beat
Source
0.1
A community clinic, 8:00am. By noon the waiting room holds sixty people. Average wait: three hours.
—
0.2
Everyone agrees where the problem is: the front desk. The phone never stops. The receptionist eats lunch standing up. She is, visibly, the hardest-working person in the building. Every complaint letter names her desk.
—
0.3
The clinic has money for exactly one new hire.
—
0.4
INT-1 — Where do you put them? Reader picks one of five: check-in · triage · doctor · pharmacy · follow-up booking. The page will not answer until they commit.
[RP] prediction-first
0.5
The reveal — daily capacities: check-in 60 · triage 45 · doctor 20 · pharmacy 50 · follow-up 40. The clinic treats 20 people a day. It has treated 20 people a day for four years.
after [D ch1 p9] production example
0.6
The violation. Whatever the reader chose, output moves to: still 20 — unless they chose the doctor.
[D ch1 p9]
0.7
And the front desk? Hire there and check-in speeds up — so patients now reach the waiting room faster. You have spent the entire budget, and the only measurable change is that people are frustrated sitting down instead of standing up.
—
0.8
The receptionist was never the constraint. She was busy because she was absorbing the anger the constraint generated. Busy is not the same as stuck.
—
[FIGURE 1]The clinic as five stations. Five boxes L→R, each with a capacity number. Patients flow as dots. Dots pool visibly at the doctor. On scroll, the counter at the far right climbs to 20 and stops dead. This figure is the whole piece in one image — build it first.
Before you can ask what's limiting a system, you have to answer something more basic and more often skipped: what is it for?
[D ch1 p5-7][syn P2]
1.2
A system: a collection of interdependent parts that turn inputs into outputs in pursuit of a goal. The last phrase is the one people skip.
[D ch1 p4]
1.3
The clinic's goal is not "see patients" — that's activity. It's "people leave healthier." Change the goal and the constraint moves with it. A constraint only exists relative to a goal.
[D ch1 p5]
1.4
Ask ten people in one organisation what its goal is and you get ten answers, each defensible. That is not a trivia problem. Until it's settled, every argument about what's broken is really an argument about what "working" would mean.
[syn P2]
ACT 2 · The chain
#
Beat
Source
2.1
Picture a chain. Its goal: transmit force. Pull harder and harder. It breaks — where?
[D ch1 p8]
2.2
At its weakest link. Obviously. Stay with the obvious, because it's about to turn subversive: how many weakest links does a chain have? One. The strength of the whole chain — dozens of links, tons of steel — is set by that single link.
[D ch1 p8-9][syn P2]
2.3
Now reinforce any other link. How much stronger is the chain? Not one bit. You've added weight, cost and effort, and it breaks at exactly the same force, at exactly the same link. All that improvement, and zero improvement.
[D ch1 p9]
2.4
This is the whole theory in one sentence: at any moment a system's output is limited by one thing. Everything else is a non-constraint, and improving a non-constraint does nothing — no matter how skilfully it's done.
[D ch1 p9]
[FIGURE 2]The chain. Links of visibly varying thickness. Scroll drives a load gauge upward; one link reddens and snaps. Second state: three other links thickened, load re-applied, snaps at the same link at the same number. The number is the point — show it twice, unchanged.
ACT 3 · The busy trap (the piece's centre of gravity)
#
Beat
Source
3.1
INT-2 — "Every single person in an organisation is working flat out, all day. Is the organisation running at its best?" Two buttons. Commit before scrolling.
[RP]
3.2
Almost everyone says yes. Here is Dettmer, flatly: "If a system is performing as well as it can, not more than one of its component parts will be performing as well as they can. If all parts are performing as well as they can, the system as a whole will not be."
[D ch1 p13]
3.3
Read it twice. It doesn't say full effort is unnecessary. It says full effort everywhere is positive evidence that the system is not running well.
—
3.4
Why: a non-constraint running flat out produces work the constraint cannot absorb. In a factory, that's inventory clogging the floor. In the clinic, it's a waiting room filled faster. On a team, it's a backlog of "finished" work nobody can review. Local efficiency converted into system-wide waste.
[D ch1 p13][syn P2]
3.5
So a well-run system requires most of its parts to have slack — to deliberately run below their maximum — so that the one part that limits everything can run flat out.
[syn P2]
3.6
The corollary is brutal. Picture an organisation with a dozen energetic improvement teams, all working hard on things everybody knows need improving. How many are working on the constraint? At most one.
[D ch1 p11]
[FIGURE 3]Twelve teams, one constraint. Twelve small squares; one lights. The other eleven stay grey. Caption carries the number, not the metaphor.
ACT 4 · Exploit before you spend
#
Beat
Source
4.1
Back to the clinic. We know the constraint is the doctor. Intuition now says: hire another doctor.
—
4.2
Not yet. Watch her day. Roughly two hours of it is spent on things that do not require a medical degree — re-taking histories a nurse already took, chasing notes, printing forms.
after [D ch1 p14] exploit
4.3
Exploit means: wring every bit of capacity out of the constraint as it currently exists, before spending anything. Move those two hours off her desk and the clinic goes from 20 patients a day to 28. Cost: nothing.
[D ch1 p14]
4.4
Subordinate means: everyone else now runs at whatever pace keeps her running flat out. Reception deliberately slows. That will feel wrong — it will look like under-performance on every local metric you have. It is the correct decision.
[D ch1 p14]
4.5
Only now, if you still need more, do you elevate — spend the money. And when you do, you hire for the doctor, not for the loudest queue.
[D ch1 p15]
4.6
Dettmer's warning: organisations that don't think this way "jump straight from Step 1 to Step 4" — from identify to spend. They incur costs they didn't need to, and skip the free improvement entirely.
[D ch1 p15]
4.7
The five steps, named once, briefly: identify · exploit · subordinate · elevate · go back — and beware inertia.
[D ch1 p14-15]
ACT 5 · Three fast jumps (~30 seconds each; three panels, one scroll each)
#
Beat
Source
5.1
A marketing funnel. 10,000 visitors. 40 sign-ups. Three people work on awareness, conversion and loyalty — and the team optimises all three equally. Rufus's line: if you're improving all three equally, you already know you're being inefficient. The constraint is conversion. Until it moves, more traffic is worth approximately nothing.
[RP]
5.2
A team's week. Every calendar full. Nothing ships. Everything is waiting on the one person who reviews. Adding work upstream adds only to the pile in front of them.
[RP]
5.3
Caring for an ageing parent. A dozen worries, all real, all urgent-feeling. The thing actually holding everything up is one unmade phone call. Not the biggest problem. The stuck one.
[RP]
5.4
Same shape every time. The question is never how hard are we pushing? It's are we pushing on the thing that's actually stuck?
[syn P2]
[FIGURE 4]Three systems, one shape. Funnel, week-calendar, and a life-admin list — drawn in the same visual grammar as Figure 1 so the reader sees the isomorphism before reading a word.
ACT 6 · It moves
#
Beat
Source
6.1
Fix the doctor and the clinic doesn't become "fixed." The constraint moves — to pharmacy, then to follow-up booking.
[D ch1 p10]
6.2
INT-3 — reader drags the doctor's capacity slider up and watches the red band jump to a different station. The system's output rises, then flattens again somewhere new.
[RP]
6.3
Push far enough and every station outruns the population it serves. Constraint gone? "Well, not really." It has moved outside the walls — to the referral hospital, or the catchment itself. Still a constraint. It just now needs a completely different set of skills to attack.
[D ch1 p10]
6.4
Which is why this never finishes. "Yesterday's solution becomes today's historical curiosity." Inertia is the enemy: we solved that, no need to revisit it is how a solved problem becomes the next constraint.
[D ch1 p13]
ACT 7 · The turn — most constraints are invisible
#
Beat
Source
7.1
Everything so far has quietly cheated. In the clinic, the constraint was a person you could point at, with a number attached. In real organisations it almost never is.
—
7.2
Most constraints are not machines or people. They're policies — a rule, a practice, a habit, a way-we-do-things-here. Every prescription needs the doctor's countersignature. No appointment over ten minutes. Nothing ships without sign-off. We don't discuss money in this family.
[D ch1 p13][syn P2]
7.3
Every one was installed for a good reason, usually to solve a problem nobody now remembers. John Gardner: an established organisation is "a coral reef of procedures that were laid down to achieve some long-forgotten objective."
[D ch1 p13]
7.4
Two properties that should give you a small chill. First: breaking a policy constraint is nearly free — and yields more improvement than breaking a physical one. Second: they are invisible, because everyone experiences them as normal.
[D ch1 p13]
7.5
A bottleneck machine announces itself with a pile of work in front of it. A bottleneck policy hides inside "that's just how it's done" — defended, sincerely, by good people whose own numbers depend on it.
[syn P2]
ACT 8 · The handoff
#
Beat
Source
8.1
So the instruction is simple: find the constraint, and put nearly everything you have into it.
—
8.2
But if the constraint is usually a policy — invisible, normal-seeming, defended by sincere people, and connected to the symptoms you can see by long chains of cause and effect that cross every boundary on the org chart — then how would you ever find one?
[syn P2]
8.3
You can't see it. You can't touch it. And your intuition, trained entirely on the visible, keeps pointing at the receptionist.
—
8.4
Ask five knowledgeable colleagues and you'll get five confident, different answers. On what basis would you decide who's right? Seniority? Confidence? Volume? Most organisations settle it with some blend of authority, anecdote, and whoever tells the most vivid story — which is to say they don't settle it. They guess.
[syn P3]
8.5
That is not a management problem. It's a thinking problem. And it's the reason a theory about factories ended up producing something much stranger: a set of tools for thinking itself.
[syn P2]
8.6
→ Next: why you'd ever draw your reasoning as a tree.
—
Notes for drafting
The method is not named in this piece. No "Logical Thinking Process", no "LTP", no tree vocabulary. "Theory of Constraints" appears once, late (Act 4 or 6), and only after the reader has felt it.
Jargon ledger — each term's first use must follow its felt example: constraint (2.4, after 0.6) · system (1.2, after 0.1–0.8) · throughput/output (0.5) · local optimum (3.4, never named — described) · policy constraint (7.2).
Withholding discipline.INT-1, INT-2, INT-3 must be unanswerable by scrolling past. The answer renders only after the reader commits.
Numbers must stay consistent across figures and prose: 60/45/20/50/40, market ≈ 60/day, exploit → 28.
Cut candidates if long: Act 1 can compress to two beats; Act 5 can drop to two jumps (keep the funnel and the parent).
