Why Your People Keep Bringing You Fires: What the Father of the Theory of Constraints Taught Us About Management

Near the end of a seminar session on managing people, having spent hours teaching a room of executives how to build respect, dissolve conflicts and turn sceptics into teams, Eliyahu Goldratt offered his credentials. They were not the credentials the room expected. Twenty years earlier, he told them, his own people had claimed he had "about half a split of a second attention." That he was a beast. Their considered assessment of the situation was that they needed twenty people against him, and even then it wasn't fair. He summarised the record himself: a regular nasty manager. Then came the sentence that justifies everything that follows in this article. "I am not anymore."

Notice what the confession does not claim. He did not say he mellowed, or that fatherhood softened him, or that he came to see the light about people. He remained, to the end, impatient, theatrical and constitutionally incapable of letting a weak argument pass. What changed was that he built a set of tools and submitted to them, and the tools changed the behaviour. He was open about the other half of this, too: some people are born managing well, and when you watch those naturals at work you see them running exactly these mechanics, intuitively and gracefully, without being able to name them. Most of us are not those people. Most of us, he said, have to learn it the hard way, and the hard way can be written down.

That is the claim of the workshop, and it is a bigger claim than it first appears. Managing people, he insisted, is not just an art. It is work, and he was precise about which kind: "the hardest work of them all, thinking." He offered a test for whether anyone actually believes this. Give an average manager the choice between spending half a million dollars and thinking hard for three hours, and watch what happens. "I didn't say anything," he added, and the room laughed the laugh of people who had approved the half million.

The two previous articles in this series followed Goldratt outward: first his marketing method, which builds an offer a market cannot refuse, then his selling method, which walks a buyer through the six layers of every no. This one turns the same machinery inward, toward the people you manage, and it was the session closest to his own scar tissue. It is the shortest journey in the series and the least comfortable, because every diagnostic in it points first at the manager. The interruptions you resent are readings from an instrument. The rule your best person hates is protecting something neither of you has named. And the target your team calls impossible is, on the evidence of a written list, merely very difficult. The tools are the same clouds, branches and trees the series has used twice already. What changes is the direction they are pointed.

Respect is the raw material

His summary began where most management books end, with a sentence that could hang in any break room: the key to managing people is mutual respect. On its own the sentence is furniture. What made it load-bearing was the operational corollary he attached. To gain respect you must give respect, and giving respect requires work, which is a different substance from warmth or an open-door policy. Work, on paper, in cause and effect: "You have to build the trees, the small trees, I call it bushes. You have to deal with the clouds, you have to do your work. Shooting from the hip doesn't work." Respect, in Goldratt's hands, stopped being a sentiment and became a product, something a manager manufactures through specific mechanisms or fails to manufacture at all.

The session took the manufacturing failures in order. The first, which opened the session, was the everyday way managers receive their people's ideas. The method he taught there is one this series has already covered in full: take the idea seriously enough to write it down, credit what it actually achieves, and then connect it, by written cause and effect, to the negatives you fear it would create, handing that branch to its author to trim. Criticism delivered as a documented negative branch tells the thinker their thinking was worth engaging; criticism delivered from the hip tells them it was worth swatting. Readers who want the full machinery will find it in the previous article, where the same negative branch turns a buyer's yes-but into co-authorship. The point here is only its position in his architecture: how you criticise is the first place your people learn how much you respect them.

The second failure is the one the bulk of the session attacked, and it hides under one of the most exhausted words in management. A person who is held responsible for outcomes they lack the authority to secure learns something precise from every episode, and learns it again each time. Goldratt described the inner state of that person without mercy: he feels lousy. He feels the system is crazy. He feels the system does not allow him to do his job. "He doesn't feel that the system respects him as an individual." Whatever the company's values slide says, the structure has already spoken. Lack of empowerment, in this reading, belongs in a harsher category than culture problems or confidence problems: it is disrespect, expressed in rules, renewed with every episode.

And the third failure lives inside teams, in the sideways direction: the quiet, experience-backed assumption that colleagues will not deliver. That one gets the session's most surprising remedy, and it comes last in this article as it came last in the session. First, the fires.

The fires are the diagnostic

Ask managers where their time goes and the answer has been stable for as long as anyone has asked the question. Goldratt cited the standard figure, more than half of managerial time devoted to chasing fires, and then the image he preferred, supplied by the managers themselves: the job feels like standing in a swimming pool full of ping-pong balls, trying to hold all of them underwater at once. The image is better than it first sounds, and worth dwelling on for a moment. Ping-pong balls do not stay down. The effort spent submerging one contributes nothing to submerging the next, the number of balls never decreases, and the only visible skill in the pool is speed. Every reader is currently holding some number of balls. The industry that has grown up around this condition sells prioritisation systems, delegation frameworks and calendar hygiene, all of which accept the fires as weather and offer better umbrellas.

Goldratt's move was to stop accepting the fires as weather. He asked two questions about the standard scene, a person arriving at the boss's desk demanding an immediate action or an immediate decision, and the two questions dismantle it completely.

First: why did this person come at all? Because the burning thing is under their responsibility. If it weren't, why would they care? ("Maybe because he's a pain in the neck. Okay? But I'm talking about the usual case.") Second: why did they come to you? Because they think their boss is a genius? No. They came because the action that has to be taken is not under their authority. They own the outcome; they do not own the lever. Put the two answers together and you have what he presented as close to a theorem: every time a person comes to their boss demanding immediate action or an immediate decision, it is a clear indication that, for that person, on that subject, responsibility and authority are misaligned. The fire is not noise in the system. The fire is the system reporting, with an address attached, exactly where it is broken.

There is a second way to read the same scene, and Goldratt read it aloud: the person at your desk is standing in a conflict. His responsibility demands an action; a rule withholds the authority to take it. And what does anyone caught in a conflict do? Compromise. The escalation is the lieutenant's compromise, and it is worth naming exactly what he purchased with it. He came to you, and by coming he covered himself. Whatever happens to the shipment now, he told you in time; it is officially your problem. The compromise protects the lieutenant and abandons the outcome, which is the signature of compromises everywhere in this series.

He then held up the mirror that makes this section uncomfortable. When the lieutenant keeps arriving with the same class of fire, week after week, what does the manager feel? "We feel very important." The arrangement is stable because it pays both parties in different currencies: the boss collects significance, the lieutenant collects cover. Meanwhile the actual cost accrues invisibly, in the lieutenant's slow education about how much the system trusts him, which is to say the empowerment being eaten alive, one escalation at a time. Only one of the two people in this arrangement can end it, and it is not the lieutenant.

Ending it does not mean refusing the fire. He was practical about the moment itself: the person at your desk has no patience for anything except dealing with it now, so deal with it now. The discipline comes after. The misalignment that produced this fire has existed for years; in Goldratt's phrasing, it can exist another day, the sky will not fall. What must not happen is the usual thing, where the fire is doused and the arrangement survives untouched, guaranteeing the sequel. So: put out the fire, and then, later, alone, take five minutes and write down the conflict the fire just exposed. Not on the fly, he repeated, more than once. Never on the fly.

Writing down a conflict has a form. It is the same form this series has now drawn twice, the cloud, and this session showed it at its smallest and most useful scale, buildable in five minutes by answering five questions in a fixed order.

Five questions to find the broken rule

The story he used deserves to be kept, because it is every operational fire ever escalated. The lieutenant runs the shipping dock. An order must ship today; the client has been waiting, and today is the last day shipment can arrive on time. The complication is that this client has two warehouses, at opposite ends of the map, and nobody on the dock knows which one this order goes to. The one person entitled to ask, the account manager, has been unreachable for three days. The lieutenant appears at your desk: now what do I do? What usually happens next is exactly what you'd expect. The boss hunts down the account manager through back channels, the answer is extracted, the shipment goes, and everyone returns to their day. The fire is out. The arrangement that produced it is intact, and Goldratt described the sequel with the weariness of a man who had watched it from both chairs: whenever reality clashes with the misalignment again, and reality specialises in exactly this, he will be coming to us, and coming to us, and coming to us. The incident that could have removed the misalignment once and for all is spent instead on removing the symptom once, at retail prices, forever.

The alternative is the five questions, and the order matters; answered out of sequence, he warned, they lead you astray. The first: what need of the system does this fire jeopardise? Not the lieutenant's need, the system's. Here the answer is plain: a client's order fulfilled on time. That goes on one side of the cloud, and it is worth noticing that this need is precisely the lieutenant's responsibility, which is why he is the one sweating.

The second question: what rule prevents the lieutenant from putting out the fire himself? There is always a rule, and Goldratt was indifferent to whether it appears in any handbook: "I don't care if this rule is written anywhere. But it's still a rule nevertheless." Here, the rule is that only the account manager calls the client. Nobody else. (His aside to the room: do you know about such rules? Many companies with such a rule.)

The third step is a manoeuvre rather than a question: write down the lieutenant breaking the rule, in this case calling the client himself, and check that breaking it would actually satisfy the jeopardised need. Would the order ship on time if he just called? Yes. The check looks trivial and is not; it confirms that this rule, and not something else, is what stands between this person and their responsibility.

The fourth question is the one people skip, and skipping it wrecks everything that follows: what need of the system does the rule protect? Because the rule is not idiotic. Nobody instituted "only the account manager calls the client" out of spite. Let seven different people from your company talk to one client, each telling him something slightly different, and you have what Goldratt called a recipe for disaster: a confused client, contradictory promises, blame arriving at your door. The rule protects a real need, one contact point for the client. Continuity. Write it down opposite the first need, with the same seriousness.

The fifth question closes the figure: what is the lowest common objective the two needs serve? Lowest is the operative word. "Our company is prosperous" is true of every cloud ever drawn and useful in none of them. The lowest objective that requires both on-time orders and one contact point is good client service. Now the cloud is complete, and reading it back tells you exactly what the fire had been trying to say. The lieutenant carries direct responsibility for one need. The action that responsibility demands is blocked by a rule guarding the other need. He is pinned, structurally, between a responsibility and a missing authority, and every time reality clashes with that pin, he will be back at your door, feeling a little less like a person the system respects.

If five questions sounds like homework, the session offered evidence on effort. Goldratt set the room a fresh fire, a foreman needing overtime authorised because a vendor delivered four hours late, and had them write the cloud cold, unaided, in their seats. It was, for nearly everyone present, the first cloud of their lives. Almost all of them finished. It took under four minutes, and when he showed his own version and asked how many had landed somewhere close, more than half the hands went up. The learning curve on this tool, he said, is very steep, and four minutes is a defensible investment in ending a fire that has been recurring for years. The exercise carried a quieter lesson too: his own cloud was offered with the caveat that it was how he perceived it, "it doesn't mean right, but at least it has some chances." The tool does not require the boss's version of the conflict to be the true one. It requires a version precise enough to argue with, which is exactly what the next conversation needs.

Fix it with him, not for him

A completed cloud creates a temptation: you can now see the misalignment, so you can now fix it, tonight, alone, with a better rule of your own design. Goldratt's instruction was to resist, and his reason cuts to what this whole exercise is for. A correction you devise alone arrives, from the lieutenant's side of the desk, as one more decree from the same system that has been disrespecting him all along. It may even be the right correction. It will still be his manager's correction, and the empowerment it was meant to restore will die in the delivery. Find the correction together. The cloud is not just your diagnostic; it is the script for a conversation, and the conversation has choreography.

The first move is counterintuitive: you present the cloud in nearly the opposite order from how you built it. Call the lieutenant in the next day and start from the common objective. What we want in our company is good client service. Goldratt's prediction of the reaction is one of the workshop's finest moments, because every reader has worn this face: fine, he has heard it seven hundred times, "mother with an apple pie," leave me alone, management talk. Do not be discouraged; keep walking the figure. In order to have good client service, we must have clients' orders fulfilled on time. And here the aloofness evaporates, because you have arrived at his responsibility, his piece of the puzzle. He is no longer bored. He is on guard.

Now the move that springs the whole conversation: so yesterday, when you came to me, if we want orders fulfilled on time, you should have simply called the client. And the lieutenant detonates. Of course! BUT. Only the account manager is allowed to call the client. Stupid! Watch what just happened, because it is the entire method in one exchange: the lieutenant supplied the blocking rule himself. He is not being shown a conflict; he is building it, half of it in his own words, and from this moment he will never argue that the conflict isn't real. And then Goldratt's observation, so regular he stated it as a law: the lieutenant always knows the rule that blocks him, and very rarely knows why it exists. To him it has never been a rule protecting anything. It has been the system's stupidity, apparently designed for the specific purpose of making his life miserable.

Which defines your one real job in this conversation: complete his picture. The rule is not from nothing. In order to have good client service we also need one contact point for the client, and that is why the rule exists. Do not expect gratitude; he will hate the rule exactly as much as before. But something structural has changed. The entire argument has narrowed to a single arrow of the cloud, the claim that one-contact-point requires that only the account manager ever calls, and he is now fighting the logic instead of the system. Fighting means you must defend, and defending an arrow means saying the word because out loud. One contact point requires the rule because otherwise the client will get different promises from different people. Whether you intend it or not, the assumptions holding up the rule are now on the table, spoken, inspectable.

And the lieutenant, who has been waiting years for this, goes straight at them. What do you mean, promises? I don't want to promise the client anything. I'm not going to tell him anything. I'm going to ask him something. What do promises have to do with it? Goldratt's commentary, verbatim: "Hello, he's right." The assumption under the rule concerned people making commitments to the client. Asking which warehouse wants the delivery commits nothing. The rule, as written, is broader than the need it protects, and the gap between them is exactly the authority the lieutenant has been missing.

Pause also on what did not happen in this conversation, because the omissions are the craft. Nobody was told to show more initiative. Nobody received feedback about their attitude toward escalation. The manager never once defended the system in general or apologised for it; the two of them examined one rule, found the one assumption under it that had expired without anyone noticing, and left everything else standing. This is what taking a person's frustration seriously looks like when it is done with paper instead of empathy alone, and it is why the lieutenant leaves the room feeling something no recognition scheme produces: listened to, correct, and trusted with a wider lane he helped draw.

The rest is etiquette, and he treated the etiquette as seriously as the logic, because this is where the empowerment either gets manufactured or spoiled. You do not grant the fix on the spot; you say, you have a point, let me check. The check is with the account manager, and it is not a courtesy: cut him out of the loop and you have not dissolved the conflict, only relocated it to his desk. A day or two later you come back with the sentence the whole procedure has been engineered to produce: we checked, and we decided to accept your idea. Nobody, Goldratt noted, files a grievance against their own idea. Two conditions keep the fix alive afterward. Whatever the lieutenant learns from the client gets reported to the account manager, every time, so the information flows and the next occasion may need no call at all. And the following morning, when the same lieutenant appears with the same fire, wearing an innocent expression, you must not touch it. But we agreed: you handle that now. He is not confused, and he has not forgotten. "He's checking whether or not you were serious." In most places, Goldratt said, that is exactly what the first repeat is, a sincerity test, and passing it "is working miracles."

He put a number on the compounding effect, and it is the most cheerful threat in the whole workshop. Handle your fires this way, consistently, for three weeks, and you will have a major problem: what to do with your time. Decline to, and his verdict was equally compact. Don't do it, and empowerment becomes lip service.

You can only empower people your procedures respect

The overtime exercise the room had solved on paper carried a second lesson, and Goldratt extracted it with a question aimed at every manager present. The cloud was clean enough: the foreman needs overtime to meet the daily quota after a vendor delivered late; the rule says only the production director authorises overtime; the quota serves one need, the operating budget serves another, and both serve the operations targets. So far, a standard misalignment. But before you stride into the correction conversation armed with "I'm the only one who can see the whole picture," he suggested a hard look at what you, the director, actually do when the overtime request lands. You check how much inventory is in the warehouse. That is the whole picture. That is the entire executive judgment the rule exists to protect: enough stock, no overtime; nearly depleted, go ahead.

Sit with what that means about the foreman's working life. He runs a production line without the foggiest idea whether what he is producing today is needed by anyone, because the one fact that decides the question lives in a warehouse report he never sees. "And that's why you don't empower them." The block was never his judgment; the procedure starves him of the single input the judgment requires, and then the organisation reads his resulting escalations as evidence he needs supervising. The correction, once you see it this way, is almost embarrassing: tell him the rule you actually use. Check with the warehouse. Enough stock, don't authorise it. Not enough, do. The overtime fire never reaches your desk again, and nothing about it required a leadership offsite. What it required was a manager willing to admit that the whole picture he alone could see fit comfortably into one sentence.

From there Goldratt generalised, and the generalisation is the hinge of the whole session. Many times, he said, the quality of the procedures we are using determines whether or not we can empower. It is not only a matter of wanting to. When the operating rules are rotten, daily quotas ground out regardless of what anyone needs, then somebody really does have to hold the overall picture and decide, "usually wrongly," and the concentration of authority is rational even as it corrodes everyone beneath it. When the rules make sense, authority can travel down to where the responsibility already lives, and the transfer is easy. His exhibit came from the production sessions earlier in the program: a plant running drum-buffer-rope, TOC's scheduling method, can hand the foreman an overtime rule one sentence long. A hole in region one of the buffer, the red zone that means the constraint is about to starve, go to overtime. No hole, tomorrow is fine. "All of a sudden, we don't need all this mystique." People can be responsible for their work, which was his definition of the exhausted word: that is the meaning of empowerment.

Then he reached for the hardest case he knew, flagged as a huge problem in its own right during his project-management session: the multi-project manager. Full responsibility for the deadline, the budget and the content, while almost nobody who actually does the work reports to him. "All the responsibility, no authority. How lovely." The instinctive repair is organisational, redraw the chart, give him the people. Goldratt's repair left the matrix untouched and rebuilt the procedures instead. With project buffers installed the way the earlier sessions prescribed, the project manager's responsibility gets redefined into something his authority can actually carry: raise the red flag when his project is eating into its completion buffer at a dangerous rate, and raise it on the specific task causing the damage, not the all-purpose "we are going to be late!" that every project manager shouts and no resource manager can act on. A correctly raised flag gets an immediate response, with exactly one exception: another project, even later, needs the same resource right now, and in that case the flag-raiser can see precisely why he is waiting. The accountability follows the redefinition. You may blame a project manager for failing to raise the flag, or for raising it when he shouldn't. That is all, and that is enough: "We shouldn't blame a person for something that he cannot do a thing about it."

The question-and-answer exchange that followed pushed the idea to its limit, and it belongs here because it completes the argument. A participant raised the worst kind of fire, the kind that arrives through several links of the chain, from people suffering someone else's inaction, carrying no request for a decision at all, only the news that it is already too late. Goldratt answered first with a Yiddish curse, which he translated with evident pleasure: may you have three beautiful cars with three chauffeurs, and the first car races off to fetch the doctor, and the second races off to say it's urgent, and the third races off to say it's too late. Then the serious answer: no cloud repairs a fire that arrives post-mortem. The only defence against too-late is procedural, buffers placed at the right points in the flow, so that the warning that someone is dropping the ball arrives while there is still time to react. The same logic that makes empowerment possible is the logic that makes fires arrive early enough to matter.

And when the questioner proposed that every manager should therefore learn these techniques, Goldratt disagreed, in the direction nobody expects. Not every manager. People. Because the tools work as powerfully upward and sideways as they work down, and so a manager who refuses to write trees does not doom his department; his people can learn the techniques "and cause him to manage them well." He was equally clear about the arithmetic of adoption. Run this in your own department alone and you are doing a lot; run it across the board and the class of too-late fires collapses, because every link in the chain is clearing its own misalignments before they cascade into yours. The exchange ended on the line that summarises the whole position: managers are people, but not all the people are managers. The technology of respect was never a management perk. It belongs to whoever does the thinking.

A target nobody believes in

The last movement of the session turned to teams, and it opened with a diagnosis most team-building never states. Put a group in front of something they must achieve together and the big fear in every head is identical: the others will not deliver. Everyone privately trusts their own effort and discounts the rest, and that discount is the same commodity this whole session has been tracking, disrespect, this time flowing sideways. Which supplies the definition most team-building skips: team spirit is every individual's earned respect for the group as an operating unit, a substance quite distinct from fondness, and one that cannot be exhorted into existence.

The standard remedy got dispatched with the best credentials in the room. Some companies, he noted, take the whole management team into the desert for four days. As a veteran of many years of Israeli army reserve duty, he had spent far more than four days in the desert, and could report the results with authority: "It does not build a team spirit. It builds other things. Not all of them very good."

His alternative was a design principle: prove to the team, by the team, that they can achieve more than they think they can. Every word in the sentence is doing work. The proof must be real, since teams have finely calibrated detectors for manufactured wins, and it must be delivered by the team itself, since respect for the group can only be earned from evidence the group produced. And the instrument of proof is a very particular kind of goal, which he defined in two halves, both load-bearing. A target is something nobody doubts would be good if achieved: double the company's net profit, double everyone's bonuses, something whose value requires no meeting. Ambitious means nobody believes it can be done. Both halves matter, because a believable goal proves nothing when reached, and a goal of debatable value dissolves into arguments about whether it was worth reaching. What you want is the goal everyone wants and nobody believes, because a team that achieves that one has learned something about itself no offsite can teach: they are surprised by their own performance, and they now know they can rely on each other.

Which sounds like a recipe for demoralisation, and then comes the move that looks completely insane. Having agreed the ambitious target, the first working step is to solicit obstacles. Go around the room and ask everyone, deliberately and at length, why it cannot be done. Every instinct in modern management says the opposite: protect the energy, stay positive, don't let the cynics anchor the room. Goldratt's reasons for inverting this are the heart of the method. The first is respect: every person in that room is carrying concerns about feasibility, has usually carried them through years of initiatives, and has learned that nobody wants to hear them. Being unheard is humiliation. Start, for once, by listening to exactly the concerns. The second reason is talent: "People have a phenomenal talent to bitch and moan." He told the room the joke about the baby who never said a word until age three and then, at dinner, suddenly spoke a full sentence of complaint; asked why he had never talked before, the child explained that until now, he'd had no reason to complain. Use the talent. It is the most renewable resource in any organisation.

The facilitation format he used, refined over dozens of runs, he called the bridge system, after the card game, and its rules repay close attention because each one guards a specific failure mode of rooms. Go around the table in strict turns, one obstacle per person per turn; the rotation exists because in every team some people take over and nobody else can talk, and the format, unlike the facilitator, cannot be talked over. Each obstacle gets compressed to a single sentence, and the facilitator's only licence is to negotiate that compression: I will work with you until we have one sentence, because a story is not an obstacle. Anyone may pass, and passing costs nothing, because like in bridge, a pass now does not surrender the right to bid later; the person who spends three turns silent and then produces the obstacle everyone else was avoiding is a fixture of these sessions. And one rule is absolute: nobody may declare anyone else's obstacle not a problem. The urge is universal, and the standard putdown, that's not the real issue, the real issue is mine, gets Goldratt's standing reply: true, you have a bigger one, in your eyes. This is still a problem. Write it down. He was explicit about why the rule matters beyond etiquette. This round is the point in the whole exercise where everyone can contribute, including the people who will contribute nothing but worry, and a worry accepted in one written sentence is a person recruited.

Then the empirical regularities, which he reported with the confidence of a man who had counted. The first twenty obstacles arrive at speed. Then the passes multiply. Somewhere past forty, the new entries turn thin, and the room itself starts groaning at them. He had never, in dozens of runs, seen a target generate more than about fifty obstacles, and this held regardless of scale, "even sometimes on a national level," and regardless of nationality. Even Israelis, he added, topped out around fifty, and we are good at this. We are really good at this.

The number matters because of what it does to the room, and he recommended testing it yourself. The moment the ambitious target is agreed, poll the group: consensus, total, one word, impossible. Now do nothing but write the obstacles down, all of them, on paper where everyone can see, and poll again. Nobody says impossible anymore. Very difficult. Almost impossible. Not big chances. But the infinity is gone, because impossibility was never a count of obstacles, it was the feeling of an uncounted swarm, and a swarm that turns out to number forty-seven has stopped being infinite. The devil, written down, is not so big.

From "impossible" to "why not?"

The list is only the raw material. The conversion step asks, of each obstacle in turn, one question: what must we achieve so that this obstacle is overcome? The answer is what TOC calls an intermediate objective, and the definition hides a quiet weapon. An intermediate objective earns its existence by overcoming an obstacle that blocks the target, and by nothing else. Anything the group feels it "has to do" that overcomes no listed obstacle has just lost its reason to exist, and Goldratt named where such orphans come from: "It has to do with last time we have done it this way." Whole layers of legacy activity die at this step, without anyone having to argue against them, because the definition declines to admit them. He held the same standard in the other direction: an intermediate objective floating in the list "out of the blue," connected to no obstacle anyone raised, is a symptom, and what it signals is missing obstacles, more reasons the target is a pie in the sky that nobody has voiced yet. Ask for them and add them to the list.

The mechanics stay humble. When no clever intermediate objective presents itself, there is always the degenerate one, the obstacle's simple negation: the obstacle is "our boss will never approve it," the objective is "our boss approves it." If that still feels impossible, it has obstacles of its own; ask for them, list them, derive their objectives, and descend until intuition changes its vote, until the group feels, we don't know exactly how, but yes, this we can make. And the etiquette from the obstacle round carries forward with a twist readers of the sales article will recognise as the inventor move wearing work clothes. Obstacle number five, who raised it? You did. Then you get first shot at what overcomes it. The person with the intuition to spot an obstacle almost always has the intuition to name its remedy; Goldratt's figure was ninety per cent. Watch what this does to the room's chief pessimist: obstacle by obstacle, the same person who proved it impossible is now authoring the plan, and "they don't say impossible anymore." Complaining has been converted, at scandalous efficiency, into planning.

Sequencing, the part that usually consumes a planning offsite, falls out of logic already on the table. This objective must precede that one only when some obstacle blocking the second is overcome by the first; there is no other legitimate reason for an ordering, which is itself clarifying, since most project sequences contain steps ordered by habit rather than by anything blocking anything. And since every objective sits on the page next to its parent obstacle, the dependencies mostly read themselves off. A group can sequence fifty pairs in about an hour, a number that sounds impossible only to people who have sequenced by argument instead of by obstacle. TOC calls the result the intermediate objectives map, and with it in hand the finish is ordinary project management, deliberately so: convert the map into a plan, who does what and how long, place the buffers properly, and run it with the same machinery the operations sessions taught. The exotic part is over; that is the point. What the team now holds is a project the team designed, and Goldratt's emphasis fell on the possessive: this is a project the team wants to succeed, because every step in it is somebody's answered worry.

Then the closing ritual, which Goldratt reported with his own astonishment still audible. After the target, the obstacles, the objectives, the map and the plan, you ask the team one question: should we try? He knew of no case, across all the times this had been run, where the answer was no. Not once. And notice how modest the yes is; nobody signs in blood to hit the number. What comes back is: why not? Let's give it a shot. We don't commit that we reach the target, but let's give it a shot. "That's what you need. Nothing more." A team that walks in certain something is impossible and walks out asking why not has crossed the only distance that matters.

He was equally specific about the costs and the range. The whole exercise, target through plan, takes about eight hours with a team; take your time, it does not take much more. It requires no expertise in cause and effect, because at every step the room does the work; of all the tools in the program, this is the one he called easiest to run without training. It has been used on huge targets and on volunteer community groups that had deteriorated to the point of barely tolerating one another, and it worked. And underneath the whole apparatus, he insisted, there is nothing mystical: necessary conditions, sufficiency conditions, "pure logic, used systematically." When the team then reaches the target nobody believed in, you do not have a workshop memory or a laminated charter. You have a team, because the fear that the others will not deliver has been disproven by the only evidence that counts.

The practice ladder

Goldratt closed the session with deployment instructions, and they amount to a principle: practise each tool where failure is cheapest, and only then aim it where missing is unaffordable.

He anchored the first ladder in the situation every listener already occupied. There are, right now, plenty of suggestions and initiatives circulating in your company that you are not entirely happy with; if there weren't, he asked, why are you bitching and moaning so much? That standing inventory of half-objections is the practice material, and the only question is where to spend it first.

For constructive criticism, the ladder starts in a surprising place: upward. Your boss's ideas are the first practice ground, partly because it is the least risky arena, whatever it feels like, since as he pointed out, whatever you are doing to your boss's ideas today is worse, and partly because criticising upward is the version that forces the most prudent preparation. Nobody walks into their boss's office with a sloppy negative branch. When it works there, move sideways, to the ideas of peers. Only when it has worked twice do you take it downward, to your own people, which he called the toughest one, the one that you cannot miss, because a manager mishandling criticism of a subordinate's idea does damage that outlasts the meeting.

For the fires, the ladder runs the other way. Use them first to find and remove misalignments for your lieutenants, three, four, five times, until the method demonstrably works in your hands. Only then turn it upward on the most tempting target of all, your own misalignments, the places where your responsibility outruns your authority and your boss is the keeper of the rule. His warning here was specific and worth engraving: on each such issue, you get one chance. Arrive unprepared, get shot down, and the second attempt on that issue is rework against a listener who no longer wants to hear it. The cheap reps come first; the expensive shot comes when you no longer miss.

The team exercise comes last: a quiet day away, the whole team, an ambitious target and the eight hours, once the smaller tools have made you fluent. And running under all three ladders, the same two disciplines the series has met before. Never improvise: "don't do it on the fly" is the session's most repeated sentence, and the sales article's warning about improvised Socratic technique, the sharpened boomerang, applies to every tool in this one. And trust the learning curve, which is steeper than anyone expects: after five real cases, he said, you will not need pencil and paper anymore, and he was careful about the order of that sentence. The one who reaches the stage of not needing the paper, that's the one who is good. The right to work without the paper is purchased with pages.

One audience question at the very end gave the session its domestic coda. Can you use this at home, with your kids? "It works beautifully. Beautifully." Series readers will recall that his marriage had already made an appearance in the sales session, as proof that conflicts between people are real. The tools, apparently, commute.

Management in 2026

The fires diagnosis has aged into something sharper than it was when he delivered it, because the fires have changed medium and multiplied. The modern lieutenant does not appear at the door; he appears in Slack, at 4:40pm, with "need a quick decision on something." The modern manager who spends the day resolving these is celebrated for it, in the contemporary vocabulary of unblocking, and ends the quarter praised for responsiveness and bewildered by the calendar. That manager is standing in the swimming pool. Goldratt's move translates without modification: log the escalations for a fortnight, pick the ones that have recurred, and spend five minutes writing the cloud behind each, the need at stake, the rule that forced the escalation, the need the rule protects. Some rules will survive the exercise. The recurring pings mostly will not, because most of them are the shipping dock again: a rule broader than the need it guards, pinning someone competent to your calendar.

Two later bodies of work make instructive comparisons, mostly for what Goldratt adds to them. Amy Edmondson's research on psychological safety established that teams perform better when people can voice concerns without fear, and the finding now anchors a thousand culture decks. The obstacle round is psychological safety implemented as procedure rather than aspiration: everyone speaks in strict turn, every concern is written as a contribution, and dismissing a colleague's worry is against the rules of the game. Nobody in the room is asked to feel safe; the format simply makes the concern-voicing happen, after which the feeling follows the fact. Similarly, Gary Klein's premortem, in which a team imagines the project has already failed and writes down why, has become the best-known tool for licensed pessimism. The obstacle harvest is a premortem with a second act: where the premortem ends at the list, the intermediate objectives engine converts each entry into a step of the plan and hands the authorship to the person who raised it, which is precisely the part that turns pessimists into planners rather than merely into prophets.

The deepest transfer, though, is the procedures argument, and it lands harder now than it did then. Three decades of empowerment discourse have treated the question as cultural: trust, autonomy, flat hierarchies, permission. Goldratt located it in the plumbing. A company whose rules route every exception through a director's judgment has decided, structurally, that nobody closer to the work can be trusted with the deciding fact, and no values workshop overrides a decision the procedures re-announce daily. The modern version of "check the warehouse" is putting the actual decision rule, the one the director really uses, into the tools the team already works in, so the authority finally lives at the same address as the responsibility. Companies that do this discover what the foreman's plant discovered: the mystique was never judgment. It was an unshared report.

For the founders and managing directors this blog mostly serves, the fires diagnosis carries one more twist, and it closes a loop the series opened. In most mid-market firms, the binding constraint on growth is no longer the product or the delivery capacity; it is the founder's attention, and the first article in this series established what the Theory of Constraints says you must do with a constraint: exploit it, and subordinate everything else to it. Now look at the founder's calendar through that lens. Every fire that reaches it, every exception, approval and quick decision that a written rule could have made, is constraint capacity spent on work the constraint should never touch. The escalation log is therefore an inventory of leaks in the scarcest resource the company has, and the five questions are the repair kit. A founder who spends three weeks clearing recurring fires the Goldratt way is doing constraint management on themselves, which may be the most profitable application of the method in this entire series.

Trying it this week

The entry point costs one week of noticing. Keep a note of every fire that reaches you, every request for an immediate decision or action, however it arrives. At the end of the week, pick the one you recognise, the fire that has visited before wearing different details. Handle the next occurrence as you always do; the person in front of you has no patience for methodology, and shouldn't. Then take five minutes alone and answer the five questions in their order: the need of the system the fire jeopardised; the rule, written or not, that kept your person from acting; the check that breaking the rule would have met the need; the need the rule protects, taken seriously; and the lowest objective both needs serve. Book twenty minutes with the person the next day. Walk the cloud from the objective forward, let them erupt with the rule and the outrage, complete the half of the picture they have never seen, and then defend the arrow until one of you finds the assumption that gives way, remembering that they have been rehearsing their side of this argument for years. Take their fix, check it with whoever the old rule protected, and deliver the sentence the whole method exists to produce: we decided to accept your idea. Then pass the test that arrives the next morning.

Calibration, in keeping with this series' habit of honesty about costs: your first cloud will take longer than four minutes, because the seminar room had Goldratt pacing behind it. The correction conversation will wobble the first time, which is why the practice ladder starts where failure is cheap. The ambitious-target day is earned, not scheduled; run it after the smaller tools work in your hands, not before. And every tool in this article fails at exactly the speed it is improvised, which is why the man who taught them, a physicist who could improvise rings around any room he entered, kept repeating the least glamorous instruction in his repertoire: sit down and write.

Which returns, finally, to the beast. Set aside every technique in this workshop and one instructive fact remains: the person teaching them had been, by his own cheerful admission and his team's arithmetic, a manager so bad it took twenty people to make the fight fair, and what changed him was a handful of tools simple enough to fit on one page, applied on paper, five times, until the paper was no longer needed. No insight, no epiphany about human dignity, no mellowing with age; the man in the workshop is as sharp-tongued as ever, running better machinery. The fires you resent are coordinates, each one marking the exact spot where someone you hired carries responsibility you never matched with authority. The rule your best person hates is protecting something real, and the two of you, on opposite sides of a drawn cloud, can find the assumption that no longer holds. The target your team calls impossible is about fifty written sentences away from being called very difficult. Managing people, he told his room at the close, is not shooting in all directions. It is building, and building, and building, and it runs on the one resource no budget line captures: the three hours of hard thinking that most managers, offered the choice, would still pay half a million dollars to avoid.

This article draws on recordings of Eliyahu Goldratt's workshop session on managing people, part of the same program behind the earlier articles in this series. The tools it describes, the conflict cloud, the negative branch, the intermediate objectives map, are set out at book length in his novels The Goal and It's Not Luck, and the people-management application survives in the TOC self-learning program on managing people.