Somebody wrote it down. When an audit comes apart years later, the point that mattered is nearly always already in the file, raised at the time, by somebody junior enough to still be surprised by things. Then it got closed. You have known that since your second year.

Sampling is going, and it deserves to. Pulling a set of items out of a population and reasoning from them was always a workaround for not being able to look at everything, and now you can look at everything. I’d take that promise slowly, because tools that test a whole population have been sitting in firms for years and sampling is still here. The direction isn’t in doubt though. Every journal entry, tested against every rule you can put into words. Every invoice checked against the order and the delivery note, for the whole year, not for the handful you picked. Recalculation. Confirmations. Checking that the statements agree with the ledger. The checklist that says which notes the accounts have to contain. The analytical work that used to eat a week of somebody’s summer. Take all of it, and most of the file assembly with it, because assembling a file has never been anybody’s favorite part.

That gives you more findings, better evidenced, sooner. Not one of them closes itself.

The half that comes next has no software in it. A point gets raised. It goes into a note. Somebody senior reads the note and decides whether it lives. The words that end it aren’t dramatic, and you have written every one of them yourself, correctly, on points that genuinely were nothing. Explanation obtained and considered reasonable. Discussed with management. Point closed, no further work required.

That is what makes the exception costly. Closing a point is quick, cheap and usually right. Keeping one open is a conversation, and the conversation is not with the client.

Watch where a point actually dies. The client isn’t in the room for it. Your manager comes to find you in the space the client lent you for the job, the one with the dead plant, the extension that won’t dial out and the file in a box under the table. He sits on the edge of that table and asks a set of completely reasonable questions. How much more work do you want to do on this. What would we actually do differently if you turned out to be right. Is it big enough on its own to change the accounts, and if it isn’t, why are we still here. What is the deadline. You would ask every one of those from his chair, and you have.

Then comes the kind version, which is the one that works. He offers to take it off your hands. He’ll pick it up with the finance director himself, informally, quicker that way, no need to make anything of it at this stage. Most of the time that is exactly what happens and it gets dealt with properly. It is also one of the ways a point leaves your desk and never comes back, without a single person in the building refusing anything. Plenty more die a duller death than that, on materiality, swept onto the schedule of unadjusted differences, where a column of things too small to matter on their own sits together and stays too small.

I’m guessing at the balance from outside, and an auditor may well tell me that conversation rescues a weak point far more often than it kills a good one.

The sharp end of all this is the paragraph about whether the company can keep going. You are not saying it will fail. You are saying there is real doubt, in a document its bank reads, its suppliers read, and its competitors read by lunchtime. Everyone in that room knows, and the finance director will say it out loud with his voice quite level, that the paragraph doesn’t only describe the danger. It is a piece of the danger. Nobody has to accuse you of anything. They only have to ask whether you are sure enough to be part of it.

None of that is an analytical difficulty. You can be completely settled on the numbers and still have everything left to decide. What you are weighing is whether to be the reason a client of many years goes out to tender. Whether your name sits on the difficult year. Whether the partners come to think of you as somebody who can’t land a job. The technical question was finished a week ago. What is left is a career decision in technical costume, and every auditor knows it, which is precisely why nobody in the room will name it.

Your profession did not leave this to individual nerve, and you are already composing the reply. There is a technical department, and once you consult it the answer is written down and binds everybody. There is an independent review partner on the jobs that matter. The audit committee has private sessions with no executives in them. There is rotation, so no relationship gets to ripen indefinitely. All of that was built for the exact failure I’m describing, all of it helps, and audit today is far better than the version your predecessors were running.

Every one of those mechanisms begins in the same place. Somebody has to raise a hand. A consultation binds after you ask for it and is worth nothing before. A review partner reads what he is handed. Private sessions help if you carry something into them. The whole apparatus sits downstream of one person choosing to spend something, and that person is employed, appraised and promoted inside the same firm as everyone they’d be going past.

So think about what better tools do to that shape. Findings get cheaper to produce and no cheaper at all to pursue. Read the whole ledger and you get far more exceptions than any sample threw up, most of them noise, every one now written down in a file where somebody has to dispose of it. Each needs a closing sentence. Disposal stays free. Pursuit costs what it always cost, and the gap between those two prices has just got wider.

One last thing, and it sits off to the side of all of that. Write down the names of the two people you would have to get past to put a modified paragraph into an opinion, and then look at what you’ve written. Both of them work where you work, both of their doors are visible from your desk, and the client’s name is nowhere on that short list. For a profession trained from its first week to worry about independence from the client, that is a strange place for the pressure to live.

The airplane flies itself and it has done for most of your career. On an ordinary day it holds a heading and a height better than you can by hand, and pretending otherwise in public is how your trade loses arguments it ought to win. Nobody who does this for a living has thought the job was the flying for a very long time.

So the honest list is long and you’ll agree with most of it. The flight plan, the route, the fuel computation, the performance numbers, the load sheet, the weather packet, the paperwork that used to arrive as a brick. All of that is calculated now, and it’ll be calculated better. Systems watch themselves and report their own faults before anybody notices a problem. Somebody will build a thing that reads the destination forecast and the traffic and the fuel state and tells you what it thinks. Fine. Take it.

Your trade has a flat, bureaucratic name for what’s left over. Threat and error management. In plain words, you spend the whole duty noticing things that could turn into problems and dealing with them while they are still small. There is no idle in it anywhere.

The runway changes late and the approach you briefed is no longer the approach you’re flying. The first officer is new and you are quietly doing more of the watching. There is ice, or the taxi instructions come fast at an airport where they always do, or the hold appears with no end time attached, or it’s the fourth sector and you can feel where you are in the day. None of that is an emergency. Every bit of it is work.

Then comes the one that costs something, and it never announces itself.

Nothing has failed. Nothing needs declaring. The destination is forecast marginal and legal, the alternate is a long way off, and the fuel is enough but the slack has gone. Or there’s an item deferred on the aircraft, properly signed off, entirely legal to fly with, and it is the middle of the night with weather ahead. Or the hold has been going on long enough that continuing and diverting are both defensible. A passenger is unwell in a way that might be nothing and might not be.

In every one of those, the obviously bad version is easy. If it’s clearly an emergency you say so, the procedure exists, everybody backs you and nobody asks a single question afterward. The hard case is the one where a reasonable pilot could go either way, and the only thing separating the two answers is money.

Everyone in the trade knows what a diversion costs, and no one has to be told. Fuel, handling, a crew running out of legal hours, a planeload of people in a city they did not choose, hotels, a broken schedule the next morning, an aircraft sitting in the wrong country for a day. You know that at the moment you decide. It is in the room with you.

The next bit gets described wrong every time it gets described at all. Nobody tells you to go. There is no voice on the radio saying press on, and if there were, the whole thing would be simple. What there is instead is a message asking for your estimate. A polite question about the fuel figure you took. A note about how many people are connecting. A crewing department that would like to know your plan. The knowledge that this aircraft has more sectors after yours, and that your numbers sit in a report alongside everybody else’s. Every one of those people is doing their job properly. Not one of them has done anything wrong. Together they are a hand on your shoulder that never has to say anything.

Command authority is the answer to that, and it’s worth being unromantic about what it is. It is a legal power held by one person to overrule the entire commercial arrangement, without asking, on the spot. Not to advise against it, and not to ask for it to be looked at again. To overrule it, alone, while it is happening.

Mostly it gets spent in small amounts. Extra fuel above the plan, which costs money to carry. A go-around because the approach wasn’t right, which costs an approach. Declining an intersection departure. Refusing an aircraft that is technically fine. Each of those is a charge against somebody’s budget, made by you, and the airline pays it and files it.

Look at what that looks like on a real morning. The plan comes up with a fuel figure somebody has worked out carefully and honestly. You write a bigger number beside it in pen, in a flight deck that smells of coffee and warm plastic, before a single passenger has boarded. Nobody countersigns it. Nobody is asked. The gap between the two numbers is money, and it will be sitting in a report with your name against it by the end of the week. Somebody may well ask you about it later, politely, in writing, having been asleep at the time.

So the power doesn’t arrive as one dramatic refusal in bad weather. It arrives as small charges laid against your own company, by an employee of that company, and somebody can always see who laid them.

Aviation got safe by taking discretion away. Standard procedures, callouts, stabilized approach criteria, fuel rules, an operations manual that has an answer for nearly everything. Your industry’s greatest achievement is that it does not rely on brilliant individuals. So why would the last scrap of judgment be the thing that stays?

Because procedure tells you what to do once you know which situation you’re in. The ambiguous case is exactly the one that hasn’t declared itself yet, and the criteria for the approach only help if somebody calls it. Your trade knows perfectly well that go-arounds don’t get flown as often as the criteria say they should. A power that gets used less often than the rules assume is already halfway to being a formality. So a human with command authority is not automatically the safe option, and anybody arguing this from the outside should say so.

How much of the ambiguous case a good system could take off you, I can’t say. I’ve never sat in the seat.

What I’m surer about is where the load-bearing part sits. Not the flying, and not even the judgment about weather and fuel. It’s whether the decision stays with somebody who can spend the company’s money against the company’s own interests, on the spot, without asking. Take that out, leave everything else where it is, and you have a different job in the same uniform.

Remember the last time you took the delay. Nobody agreed with you on the radio, nobody upstairs signed anything, and by that afternoon the cost had landed on the airline with your name beside it. That is the authority, and it’s the part people leave out when they describe your job.

A prescriber signs an instruction, and it does not reach a human body until a second professional, who never saw the patient, is satisfied by it. That is a stop, it is legal, and it belongs to whoever is holding the box. This morning it belongs to you. The prescription is signed, the dose sits inside what the book allows, and it’s wrong for the woman standing in front of you with her coat on. She has a bus to catch. There’s a queue behind her, a rack of baskets waiting, the phone is going, and the practice closes for lunch in a few minutes. You already know how this goes now, and so does she, because she can hear your side of the call.

Take the mechanical half first, because your trade has been living with it longer than most and has stopped being impressed. Labeling, picking, counting, the stock order, the reconciliation of one list against another. In the bigger operations a machine already fetches the box. Interaction checking has been software since long before any of this arrived, and the software flags so much that everybody in the building clicks past most of it, which is the oldest joke in the dispensary and also the point. Detection was never your bottleneck. Your screen has been detecting things at you for a generation.

What the screen cannot do is stop anything.

It’s worth saying in plain words what that stop actually is. If you are not satisfied, it doesn’t happen. Not delayed, not flagged for review. It doesn’t happen, and there is no appeal except finding a different pharmacist. Plenty of the rest of your working life runs on committees, formularies and guidelines written by people in rooms, and this isn’t one of them. It happens at the point of supply, and almost nothing else in medicine is built that way.

The awkward part is who holds it. You are, in the ordinary hierarchy, junior to the person you’re contradicting. You have to get past a receptionist, catch a doctor between patients, and say that their prescription is wrong, out loud, while your patient listens. You’d better be right, because you’ll need that practice to take your call again next month.

Some of it isn’t an error at all, and those are the harder ones. A prescription can be entirely correct and entirely wrong for this particular person, who is old, already on a long list, and has just been handed one more. Somebody is buying the same painkiller with codeine in it for the third time this week, and now he’s explaining about his back. A man wants antibiotics and has been rehearsing his reasons on the way over. The person collecting a supervised dose every day has turned up unsteady, and has to be told no in a shop, in front of other customers, without a scene and without losing him altogether. There is no mistake to point at in any of those. There’s just you, deciding against what somebody has come in wanting.

I don’t want to oversell this. Most of your day is not refusing anything. The overwhelming majority of what crosses your counter is fine, you dispense it, and that is the job. Any pharmacist who described their working life as a run of brave interventions would be laughed at by their own counter staff before the sentence finished. The refusal matters for what it says about the position, not for how often it gets used.

And the position has a second half that nobody puts in the job description.

Anyone can walk in on you. No appointment, no referral, no gatekeeper, no charge for the conversation, in the evening and on a Sunday, in towns where every other door with a clinician behind it is shut. The emergency room is the other one, and everyone knows what it costs to take a rash in there. So you get the man who has been buying indigestion tablets for a month and mentions in passing that his clothes are looser. A mother arrives after the practices have closed with a baby who isn’t feeding. Somebody asks about vitamins for a while and then asks the real question on the way out. Then there’s the one who cannot get an appointment for weeks, who is frightened, and wants somebody to tell them whether frightened is the right response.

You sort that standing up, in public, with no notes in front of you, between two other things. Then you decide which of them gets sent somewhere and how urgently you say it, knowing that if you send everybody you’re useless and if you send nobody you’re dangerous.

Look at what those two halves have in common. Coming in needs no permission from anybody. Stopping something needs no permission from anybody. Both run through one person standing behind a counter, and the license is what makes both of them work.

Your trade knows the uncomfortable version of this better than I do, so here it is. The clinical check on a repeat that hasn’t changed in years is not deep thought. It’s a glance. You have done a great many of them and you know exactly how thin that check can get on a bad afternoon, and anybody claiming every one of them is an act of judgment is not being straight. A signature can absolutely rot into a formality. Yours does, some days, and pretending otherwise would insult you.

The property that matters isn’t the depth of the average check. It’s that a stop is available, held by somebody who is right there, and used the few times it needs to be. That gets taken away by a business decision rather than by anything technical. Move the dispensing to a central operation, have the pharmacist check remotely from another building, and turn the shop into a place where a box is collected. Every legal power I’ve described survives that on paper. The door does not. Separate the two halves and you get a check with nobody behind it and a counter with nobody at it, and the arithmetic will look good for a while.

You are also the last place any of it is cheap to stop. Everything upstream of your counter is still a plan on a screen. Once the woman with the coat has the box in her bag and the bus has gone, undoing it takes a phone call, a recall, an apology and most of somebody’s afternoon, and that is if anybody notices at all.

Tomorrow, watch the people who come in and leave again without buying anything. Not the ones collecting, the other ones, the ones who came in to ask you something. None of them made an appointment. None of them was referred by anybody, and nothing being built to replace the dispensing has a door in it at all.

There’s a beam inside that wall and nobody knows what it is. You’ll design onto it anyway, probably this week, and nothing about that is unusual. The footing is under the floor, somebody put it there before you were born, and either nobody drew it or somebody drew it and the builder did something else on the day.

The analysis is going, and honestly most of it went years ago. Frames have been sized by software since before you qualified, and your trade has been complaining about graduates who can drive a package and not sketch a bending moment for at least that long. What is arriving now takes the next layer off. A model built from the drawings. The load takedown, the sizing, the connections, the schedules, the clause by clause check against the standard, the calculation package written up so a stranger can follow it. Specification and report as well. Nobody sane enjoys the write-up, and I’d hand the lot over without an argument.

The usual reply is that the calculation was never the job, which is true, and then comes the sentence I want to argue with. The job is deciding what to worry about. I don’t think it is, and the reason is a compliment to your profession. Your worry list is published. A century of failures got turned into load combinations, factors, deflection limits, tying requirements, accidental loads, fire, wind, the lot. That is what a design standard is. It is a written list of what to worry about, paid for in collapses, and it is the most successful piece of knowledge transfer any trade has ever managed. Anything written down that plainly is available to whatever reads it next.

The judgment is somewhere else, and there are two of them, and they turn out to be one.

Most of what’s published assumes new work. New steel of a known grade, concrete somebody will test, timber with a stamp on it. There is guidance for assessing what is already standing, it is better than it was, and it runs out early. Then you get handed a building with cast iron in it, or wrought iron, or lime mortar, or a floor of shallow brick arches sitting on iron beams, or timber that nobody graded because grading did not exist. There is no clause for a beam a plumber notched to get a pipe through. There is also no clause for the fact that this floor has carried a working machine shop for generations without a murmur, which is evidence, and rather good evidence, and there is nowhere to put it in the calculation. You use it anyway. Every engineer does. Nobody has written down how.

The other one eats your week. How much to open up, and where. Every hole costs money, needs access, wrecks a finish, and has to be argued for with a client who cannot see why they are paying you to make holes in their own building. So you ask for a few in the right places instead of a lot in the wrong ones. Most of the job then rests on that choice, and you made it by guessing what was likely to be in there.

The guess is the trade. Not construction in general, which is published, taught and examined. What you are running on is a private catalogue of buildings you have had open. The era, the town, what the builders around here did when they were behind, which decade put a steel in properly and which one took the wall out and hoped. A floor that moves under you as you cross it has told you the span before you measure anything. A crack has categories and the categories are printed, but whether it is still moving is settled by paint, by dust, by whether the last repair has opened again, and by whether the door at that end binds in winter. That is where the two judgments turn out to be one. Where to open up, and what to make of what you find when you do, come out of the same catalogue, and there is nothing else to answer either of them with.

Then the ceiling comes down and there is a steel in there nobody drew, sitting on half a brick, painted the same color as everything else, and the floor above has been fine since somebody’s grandfather put it in. You have seen it now. There is no clause for that either.

What is happening with that knowledge is not the familiar story about wisdom being too deep for words. It is smaller and more annoying than that. Every piece of it is worth almost nothing on its own. One street, one decade, one builder’s habit. Write it down and you have produced a sentence nobody will ever go looking for. It pays only as an accumulation, held in one head, sorted by resemblance, and handed up without being asked for. Firms have tried to keep it. Defect libraries, lessons learned, the folder on the server that somebody set up with real enthusiasm. They fail, and not because engineers are lazy. The unit of this knowledge is too small to be worth recording and there is too much of it to record all of.

Scanners are getting better faster than the argument I’ve just made. Radar through concrete, a camera down a small hole, meters that find the reinforcement, monitoring left on a crack through a season. Feed a point cloud into a model and a lot of what I have been calling unknown stops being unknown. I think that is true and I think it takes a real bite out of my argument. What a scan gives you is shape. It does not give you the grade of the steel, or whether the connection was ever finished. It won’t tell you what the mortar has turned into inside the joint, or whether the bearing under that beam is brick or rubble with a face on it. Somebody still has to decide what to scan and get it paid for. I don’t know how far those tools go and I wouldn’t trust my own guess about it.

Which leaves the awkward fact about the catalogue. It is built out of whatever you happened to have open, so it is local, lopsided and full of holes you cannot see from the inside. Send somebody to a town they have never worked in and they are confident in exactly the same way, right up until they aren’t. Nobody in the office can tell the difference, and on a bad job neither can they.

The next time a ceiling comes down in front of you, stand there for a minute before anybody starts sweeping up. Everything you can see was invisible on Monday, and the drawing you designed from is still in the car with none of it on it. That drawing was wrong long before you got there. So is the next one.

Any fool can cut. Your trade has been saying that to itself for generations, usually with the other half attached, the part about knowing when not to. It is the one thing surgeons say about their own work that outsiders never repeat back. It is also the one thing nothing in your training, your figures or your pay has ever behaved as though it believed.

The operating room is the most rehearsed room in the building, and that is why it’s the part of your work with the clearest path to being taken apart. I’m not telling you a machine is about to do a bowel resection on its own, and you’d close the tab if I did. The point is smaller and harder to argue with. Cutting is the end of a long process, and it’s the end that has been written down. Steps, approaches, checklists, planning off the imaging, navigation, instruments that hold a line better than a tired hand at the end of a list. Your trade did most of that to itself, on purpose, because standardizing the technical act is how outcomes got better. Anything a profession has already specified is available to be assisted, and then to be assisted more.

Meanwhile the operating room is a minority of your hours and every surgeon knows it. Clinic. The meeting where a room full of specialties looks at one patient. The ward round, the on-call phone, the letters, the complaints, the consent conversations, the family who want a word outside.

One appointment out of all that is what I actually want to talk about. It’s short, it happens in clinic, and nothing in the way your work gets counted describes it properly.

An old man is sitting in front of you with something you could take out. The scan says resectable. His heart is not going to enjoy the attempt, he lives alone, he was already slowing down before any of this, and his daughter has read about the operation and would like him to have it. You could do it. You are good at it. You have done it many times and the technical part does not frighten you at all.

You are going to tell him no, and the appointment is going to run long.

Notice what you did not use to decide. Not the scan, which everyone in the room can see. Not the staging, which is a category. You used how far he walks now, whether he gets up in the night, and how he did after the last small thing that went wrong. Whether there is anybody at home for the weeks afterward. What he actually wants, which took a while to get out of him with his daughter sitting there. Then you weighed a recovery against the life it would be a recovery into.

You have a script for the other conversation. Consent has been trained into you, shaped by law, written into a form, argued over in court. Somebody has thought hard about how to tell a person what you are going to do to them and what could go wrong. There is no equivalent for this one. Nothing ever taught you how to say I could operate and I’m not going to, to a man who came in hoping, with his daughter’s face changing while you speak. You made your own words up years ago and you’ve been using them since.

You don’t decline alone. The case goes to the meeting, a room full of specialties looks at it, and that is exactly what the meeting is for. It’s a good arrangement. It has saved a great many patients from one person’s bad afternoon, and any post that skipped past it would deserve the eye roll. I might be splitting a hair that doesn’t exist. Plenty of surgeons hold that the meeting genuinely decides.

It still doesn’t move what I’m describing. A meeting recommends. A named surgeon accepts the case onto a list or doesn’t, and the one holding the knife is the one who can refuse it. Everyone has sat in the version of that meeting where the room agrees with whoever spoke first, the form gets completed, and nobody in it would say afterward that the decision was theirs. Everyone has seen a committee turn into paperwork. What keeps your own acceptance from going the same way isn’t a rule. It’s that you’re the one they call in the middle of the night when it goes wrong, and there is no version of this where that lands on the meeting.

Then there’s the part that stays open. Operate and it goes badly, and you find out. It gets discussed at the meeting where deaths and complications are gone through in front of your colleagues, it goes into your figures, and everyone can see it. Decline, and the decision is written down too, in the notes and in the letter back to the referrer, and none of that is an answer. If he dies at home in the spring, nobody can tell you whether you were right. Not the family, not the referring physician, not you, not ever. There is no version of him who had the operation, so there is nothing to hold the decision up against.

You carry that anyway. That is the job I’m trying to describe. Somebody has to hold a decision that will never be settled, in a career where everything else gets checked, and go back to clinic on Thursday and do it again.

A model will produce a risk score for that man, and it will be a decent score. What that score can and can’t be is worth pinning down. It’s built from people who were operated on, because those are the ones with outcomes attached. The man you turn down leaves no trace in that data of what happened to him next. So the score can tell you what tends to happen when somebody like him goes on the table. It cannot tell you what happens if he doesn’t, and it has never once had to say so to a daughter.

There’s a nastier turn in this, and your trade found it first. Publish outcome figures by surgeon and declining the difficult case becomes the rational thing to do, for reasons that have nothing to do with the patient. Everybody in surgery knows a colleague who stopped taking the hard ones and got a better looking year for it. So refusing is not automatically the noble half of the story. It’s just where the weight sits, and if you change what gets measured, you change who is protected and who is left holding it. That is worth watching whenever somebody proposes to score this work.

Picture your figures for the year, the real ones, the ones your unit gets judged on. Now picture the second list beside them, the patients you decided against, with no outcomes on it at all. Nobody keeps that list. I can’t tell you what’s on it, and neither can you, and I’d want anybody planning to automate this work to explain that gap to me before they started.

The part of your work with the most paper under it gets checked hardest. The part with almost no paper under it is close to final, and nobody sat down and planned that. A stack of orders is waiting this morning, settled between two sets of lawyers and typed up by somebody else, and your name goes on the bottom of work you never argued about. Nobody calls this a scandal. It is how the work has always been done, and it is the honest place to begin, because it is already the thing people are afraid of. You have a list today, and you will get through the list.

Research goes. So does the first pass through a file handed over at short notice, the chronology, the fair summary of what each side says, the check that a case still means what it meant when somebody first cited it. The procedural history at the front of a judgment goes, and nobody will mourn it. A first draft of an uncontested ruling on facts nobody disputes goes. And a great deal of what people imagine your discretion to be left years ago, into guidelines and ranges and categories that tell you the shape of the answer before you open your mouth. Judges have been saying that out loud for a long time. It was not a machine that did it.

The rest is short, and none of it sounds impressive written down.

Two people have told you different stories about the same afternoon. The documents don’t settle it. One of them is lying, or both are half remembering something that mattered to them at the time, and you have to write down which. That finding is the least reviewable thing you do all year. A court above you will take your law apart at leisure and leave that alone, because you heard the witness and they didn’t.

I don’t think that deference is a compliment to your eyesight. Anyone who has sat through a long trial knows how ordinary the judgment of a liar really is, and how often confident people get it wrong. It is an allocation. A dispute has to stop somewhere, somebody has to be the one who stopped it, and the system has taken the person who listened to the whole thing and made them final. Not because they see through people. Because the finding has to belong to someone.

The same arrangement is at its plainest when you pass a sentence. Most of the number is often settled before you speak. Where there’s a guideline it has done the arithmetic, the categories have done the rest, and any judge who pretends otherwise is enjoying themselves too much. Then it gets said out loud, by a person, to another person, with that person’s mother sitting behind them. None of that changes the number. It changes who the number came from, and everybody in the room can answer that question without being told.

Liability is the wrong word for what you’re carrying, and the difference matters. Nobody is going to sue you. In most places you have immunity for what you do on the bench, and where you have it, it exists so you can decide against powerful people without pricing the cost to yourself. What you carry isn’t legal exposure. It’s authorship. Your reasons are published under your name. When you get it wrong, a court above you says so in a document anybody can read, other judges will read, and the lawyer who lost in front of you will read twice.

A signature rots, and the objection has been sitting in the first paragraph since you started reading. A person who approves a document they did not reason their way to has produced an audit line and not a decision, and everyone has watched that happen somewhere. You already sign agreed orders in a pile, so you know the feeling from the inside. Put a system in front of a judge that drafts a full ruling, with findings and reasons that read better than his own, on a list that is too long, where the draft is nearly always fine. The honest description of that morning’s work is that somebody checked it looked right.

I can’t point to a rule that stops this, because there isn’t one. The only thing holding is that reversal is personal and public. A machine’s draft, once you adopt it, is your draft. The error gets your name on it and the correction gets your name on it too, and that is an uncomfortable enough arrangement to keep most people reading properly.

I’d rather admit where that protection is thin than pretend it isn’t. It works while the number of decisions is small enough for each one to be attached to a person who can be embarrassed by it. Push the volume up and it wears through. At the small end of civil work, where the sums are modest and nobody is going to appeal anything, I’d expect the person to go first, and nothing in this post rescues that. It ends up as a service with a complaints procedure, which is not the same animal and everybody will pretend it is.

That leaves the part of the job most judges enjoy least. Ask one what the difficult bit is and many will start on the law. The law is difficult the way a puzzle is difficult, and it arrives with cover attached, because you can point upward and say the answer was never yours to choose. Deciding who is lying comes with nothing to point at. No authority, no clause, no table. As the rest of the work moves off your desk, that is what your day thickens with, and it is the part where you are most on your own and least able to show your working. That last bit is a guess. It comes off what judges write about the job rather than off doing it, and a week of lists, case management, remedies and sentencing may look nothing like the picture I’ve drawn.

There’s a version of the future where that gets described as efficiency. The reading was done for you, the draft arrived clean, and all that’s left of the office is the half nobody can help you with. Nobody planning this has said out loud that they are concentrating a job down to its most exposed component. It would be an honest thing to say.

When somebody describes a system that would decide these cases, don’t argue about whether it would get them right. Ask them something smaller. Who gets told they were wrong, where is that person standing when they’re told, and what happens to them afterward. If there’s no answer to that, they haven’t described a court.

Nothing about this job is hidden. Read the record, assign the codes, decide which one goes first because the order changes what the stay is worth. Query the doctor when the notes will not support what plainly happened. Keep it accurate, consistent and defensible long after anybody has forgotten the chart, and clear the queue, because the count is watched.

That is all of it. Unlike most job descriptions it leaves nothing out, and that is the problem, and it is most of what I have to say.

I have been going through these jobs looking for what is left when the logistics are taken away. Here I don’t think anything is left, and I would rather write that than perform an ending I don’t believe. This job ends. Not thins, not shifts, not keeps the title while the day changes underneath it. Ends. You have known that longer than I have, and I am not going to talk you out of your own eyes.

Look at what the work is made of. A written record goes in. A set of codes from a published list comes out. The rules for getting from one to the other are published too, along with the guidance on the awkward cases, and the guidance gets updated and published again. Then the claim goes off, and it comes back paid, denied or adjusted, which means the whole thing has a scorekeeper attached. Text in, text out, rules on paper, a right answer that arrives in the post. There is no room to be in. There is nobody to be present for. Every property that has protected other trades is missing here, and none of that is a comment on how hard the work is.

It has also been happening for a long time, which your trade knows and mine keeps discovering. The encoder came, then the software that reads the note and suggests the codes, and a lot of coding quietly became checking. In the high volume, low variance corners, the ones where every chart looks like the last chart, claims already go out without a person seeing them at all. Nobody announced that. It arrived as a pilot, then as a default.

Now the parts you would defend, and you would be right to raise them. Sequencing is real judgment. Choosing the principal diagnosis out of a messy admission changes the group and the payment, and it takes somebody who knows both the rules and what the doctor meant. The query is real too, and it is the piece coders name first. You spot that the record says one thing while the care obviously was another. Then you write to a busy physician in a way that gets an answer without leading them, which is a real skill with a compliance edge on it.

I don’t think either of them survives. That is a guess, and it is the one in this post I would most like to be wrong about, because the query is the piece your trade defends hardest. The reason is the same for both. Both are judgments about text, made against published rules, with the answer checkable afterwards. That is the exact shape of the thing being built. The query has a further problem, which is that it is a gap between two documents, and finding gaps between documents is not the hard part of anything anymore. The relationship with the physician, the part that actually gets the answer back, mostly sits with the clinical documentation people, and that is a different job with a different badge and there are fewer of them.

In every other one of these, the thing that stayed was some exception the system couldn’t take. Yours are all written down. That is the whole difference, and it has nothing to do with skill, which is not what saves a job and never was.

There is a smaller version of the work that survives a while. Auditing the machine, working denials, the appeals that need somebody to write a paragraph a human will read. Those are real and they are fewer, and they are a different job wearing your certificate. I would not build the rest of a working life on them.

So let me say what this actually costs, because that is the part nobody writing about automation bothers with. Coding was one of the few genuine professional credentials a person could reach without a degree. You studied at a kitchen table after a shift somewhere else, you sat the exam, and you came out with letters after your name that meant something to strangers. It paid properly. It could be done from home, which meant it could be done while raising children or looking after a parent, and it could be done from a town with no other work of that kind in it at all. There was a ladder inside it, specialties and audit and management, and people climbed it. That is a rare object. It is not replaceable by a general observation about new jobs appearing elsewhere.

I am not going to tell you what to retrain into. Any guess I made would be worth nothing, and you would hear the hollowness in it straight away. I am also not going to hunt for a hidden core in your work that turns out to be sacred. I have looked and it isn’t there, and handing you one would be a worse insult than this paragraph.

What I think it looks like from inside is this. The targets rise, because the software has made you faster and the target follows the software. The routine work thins first, so what is left in your day is the ugly stuff with none of the easy charts to balance it. A colleague leaves and is not replaced. The postings change wording. Then somebody uses the phrase moving to a review model in a meeting, and everyone in the room understands.

Your job was never medicine and it was never quite administration. It was the friction of a payment system, and the friction had to be staffed by people who could read a chart. That is not a small thing to have been, and it is not a criticism of anybody who did it. It is only a description of where the work sat, and work that sits there does not get to stay.

Read the postings for your own specialty, the ones that are up right now. Put them beside the one you answered when you started. Nothing in this post will have been news to you. You have known for a while, and you have almost certainly known longer than the people who manage you.

Price is the smallest lever on your desk and the only one with a number on it. A submission lands in the last week of the renewal season. Three years of losses, a schedule of values with the same address typed three different ways, and a broker’s note saying it’s much like the account you wrote in the spring. You have an hour for it, and the pile behind it hasn’t stopped growing since the weekend.

If your book is motor or household, none of this is about you. That was settled by tables and screens long before any of the current noise. For everyone else, that hour goes on two different kinds of work and only one of them has a name. Pricing what has already happened is the named one. Deciding about what hasn’t is the other, and on most days it never announces itself, because the difference only shows up on the day it matters.

The first one is going, and I’m not going to be coy about it. Loss history is data, and a model is a machine for eating loss history. It will triage the submissions, price the ordinary risk better than you do, benchmark the terms against everything else in the market, and handle a renewal where nothing has changed. It will do it at the end of the renewal season without getting tired, which is when your book gets its worst decisions made. Anybody in your trade who insists a person prices small commercial better is going to lose that argument in front of the people who hold the pen on it.

The mistake about your job starts in the part of the lever that has no number on it. Where there is a technical rate it comes from the actuaries, and in the lines this post is about there often isn’t one, so you build the number yourself out of exposure and whatever the market is doing that month. Either way the number is the smaller part. Selection, structure and refusal are the rest of it. The deductible, the limit, the sub-limit, what sits inside the aggregate, the exclusion you insisted on, the wording you would not accept. Premium is one number among all of that. The trouble lives in a sentence somebody drafted years ago that a court will read in circumstances nobody had thought of.

The thing that actually kills insurers isn’t the enormous single loss. Everybody plans for the enormous single loss. That is what the reinsurance is for, and the whole market has rehearsed it. What ends companies is a book full of separate risks that turn out to be one risk. Every account sound on its own. Every one of them leaning, without anybody writing it down, on the same supplier, the same assumption inside the same standard clause, the same piece of infrastructure nobody thought to ask about. You wrote a portfolio. You owned a single position.

The machine’s confidence will be highest exactly there, because every one of those accounts resembles the history closely, and resemblance is what confidence is made of. Each price is defensible. The portfolio is a bomb. And this is the part I would put in front of any underwriter and ask them to shoot down: a loss with no precedent does not arrive looking strange. It arrives looking ordinary, many times over, one broker after another.

Somebody has to see it, and the only people who ever have are the ones who watched a book turn. You bind something today and find out later, sometimes at the end of a career, and it comes back as a reserve moving on a decision you can barely remember making. Not the big losses. The small steady wrongness, arriving late, about business you thought was fine. That is the slowest lesson in finance, nobody has found another way to teach it, and it is why the dull business matters. Ordinary risks are the only thing you get enough of to have a book of your own. If the model writes all of them, the person behind you sees referrals and nothing else, which sounds like a promotion and is nothing of the kind. An exception means nothing without a mass of ordinary underneath it to be exceptional against.

Some of this has nothing to do with data at all. You learn to decline by living through a soft market with your own name on the declinature. Everybody around you is writing the business, the broker is disappointed in you, your line is being filled by somebody down the street, and the numbers say you are wrong for a long stretch. Then it turns, and you find out. Reading about a cycle is not the same as having sat in one and held. If the ordinary book is machine written, that pressure never lands on a person early enough in their career to shape them, and the willingness to say no stops being a memory and becomes an opinion.

There is a serious objection to all of this, and your exposure management people will make it before you do. Aggregation is not unmanaged. There are catastrophe models, stress tests, disaster scenarios, accumulation limits by zone and class, and a whole function that exists to ask exactly the question I have just described. That is all true, and any post pretending otherwise has earned whatever it gets.

So let me keep my claim small. Everything in those models is a peril somebody already named. The scenario exists because a person once looked at a pile of paper and got an itch, and said out loud that these accounts might be the same account. That naming is the act. It has never come out of the data, because before the first loss there is no data, and after it there is no shortage of people willing to model it. What worries me is that the naming gets done by people who spent years being wrong slowly, and we are quietly closing the only place that ever happened.

And your trade is built so that none of this has to be noticed for years. Insurance is the one business that takes the money before it knows what the goods cost. A book written on borrowed confidence looks like a good year, then another good year, and the accounts agree with you the whole way through. By the time the reserves move, the way of pricing that did it has been the house standard for years. It is in the model everybody inherited, and the underwriters who might have argued with it are running somebody else’s division.

Take two accounts you have written this year, different brokers, different classes, and read the two wordings side by side until you find the assumption they are both resting on. If nothing turns up, read them again, because a real one is never in the same paragraph twice. Then work out whether anybody starting this year will ever hold enough business to run that test at all. I know what I would bet on that, and this isn’t the post where I write it down.

The work you did in your first years wasn’t design and everybody in the studio knew it. You resized other people’s layouts, you set price lists, you built the deck for somebody else’s meeting, and you redrew a logo from a photograph of a logo because a photograph of a logo was all the client had. Then you went home and did your own work, which was better, and which nobody paid for.

That production tier is going first and going fastest, and I doubt anybody in your trade will argue with me about it. My claim is narrower and less comfortable. Those jobs were the whole training scheme. There has never been another one, and nothing that is arriving now does what they did.

Start with type, because it is the clearest case. You set a headline, somebody senior looked at it for about a second, said it was loose, and walked off. You fixed it, or you thought you fixed it, and they said it again. This happened on jobs where the headline didn’t matter, week after week, until the day you saw it yourself before anybody told you. None of that arrived as a rule, because there isn’t one to arrive. What you were building was a threshold for wrongness, and a threshold only calibrates against attempts you made yourself and cared about.

Then production, which designers outside the trade forget and designers inside it never do. A file comes back from a printer with the type filled in, or a color that was gorgeous on screen turns to mud on uncoated stock, or a laminate eats the finish somebody sold to the client. You find out about bleed by not leaving any. You learn what a printer will phone you about by being phoned. All of that knowledge came from small jobs, because a small job is where you are allowed to be the reason something went wrong.

And then the clients. Somebody says make it pop, and every designer alive has heard that sentence and winced. What it usually means is that they can’t tell where to look first, which is a real problem and a fair complaint, and their proposed fix is wrong because fixing it is not their job. You learn that translation by getting it wrong. You do the obvious thing, you make everything louder, the client is still unhappy and can’t say why, and eventually you work out that the answer was to take something away. Nobody teaches that in an afternoon. You learn it on a leaflet nobody kept.

Set that against what a junior gets offered instead. Options, at speed, in quantity, on demand. And this is where I think the substitution fails, because being handed a screen of things you didn’t make isn’t repetition. It is browsing. The old loop had you commit to a choice, defend it out loud, and find out you were wrong about something you actually believed. That is what a crit is for, and it needs a piece of work you own. Pick one of forty and you have expressed a preference, and you can be made to defend that too. I think that is a thinner kind of being wrong than the one you get from having made the thing yourself. I’m guessing at the size of the gap, though, and an art director who trains people for a living might tell me there isn’t one.

The awkward part sits right there. Directing one of these tools well takes exactly the eye that the old jobs built. You have to look at four passable images and know that the third is closest, and that the type is off, and that the whole set is solving the wrong problem. That is evaluation, and evaluation was never free. It came from years of making the bad version yourself. So the tool works best in the hands of people who are already good, which is a lovely arrangement for them and a closed door for everybody behind them.

There is an economic version of this that studios will discover the hard way. Juniors were never charity. They were profitable, because production hours billed and cost little, and the training happened inside work that was already paying its way. Nobody sat down and decided to run a school. Take away the billable production and the junior stops being a small profit with a lesson attached and becomes a straight cost with a promise attached. Studios are not sentimental about costs and never have been.

Art direction is a taught craft, and none of the teaching has ever needed equipment. People are trained in it on purpose, by looking hard at work they did not make, saying what is wrong with it and why, and being told they are wrong about that. If an eye can be built that way, then a junior with forty options and a good director at their shoulder is not starved of anything. They are in a tighter loop than you ever had, with more attempts in a week than you used to get in a season.

My answer, and you can weigh it against your own years, is that the loop which taught you had a client in it. Somebody outside the studio was unhappy, or the printer rang, or the thing went out into the world and was wrong in a way nobody in the room had predicted. A director’s verdict arrives before the work has met anybody, and it is one person’s. I don’t think the tighter loop replaces that, and I can’t prove it either. Nobody has run that experiment long enough to know, because everybody teaching it now learned the old way.

Replacing the loop deliberately would mean studios funding work that earns nothing so that somebody learns to see, and that is exactly the sort of spending that dies the first time a year goes badly. A few good places will do it, mostly because one stubborn person insists. Most won’t, and will complain later about the standard of the people applying to them, and will not put the two things together.

Look at the smallest job in the studio this week, the one nobody wants and nobody will ever see. Before you feed it to anything, hand it to whoever joined most recently, and let them carry it all the way out to the printer with their own name on it. It will cost you an afternoon and a slice of the margin. Nobody is going to make you do that, and nothing in the way a studio gets run will ever suggest it.

Elongate member. It’s a rod, the attorney who wrote it knows it’s a rod, and they know that you know. It went in that way to keep the claim broad and your search off the shelf where the rods live. Most of what is hard about your work is already in those two words. The office would describe the job as one thing. Search the prior art, meaning everything the world had published before this application was filed, apply the law to the claims, write an action, read the reply, allow it or don’t. Every word of that is true, and it is two different jobs, counted and paid as one.

The first job is retrieval, and I think it goes completely. Prior art is the best behaved body of documents anywhere. It is published, it is dated, it is classified by people who spent whole careers on the classification, and the families are linked across every office that has ever handled the same invention. Art in other languages stopped being a wall some time ago. Searching a corpus like that by meaning rather than by words is precisely what these systems are good at, and you already work with tools that are a long way past keyword lists. Some of you will be glad.

There is one honest complication in that, and it matters later. Searching well depends on reading the claim properly, which is where the elongate member comes back in. A good part of what looks like search skill is really claim construction, and I would put that in the second job rather than the first. That is somebody who has never examined an application telling you what your own craft is made of, and I know how that reads. Examiners would say that building a strategy, and knowing when you have searched enough to stop, is the skill, and that it doesn’t come apart into halves the way I’ve just cut it. Which of us is right isn’t something I can settle. I’m guessing, and I’d rather say so. What I’ll stand behind is narrower: the part that really is finding documents gets cheap fast.

Novelty gets cheap with it, and I don’t think that one is controversial. Does one document, on its own, already contain every element of this claim. That is a comparison. It is checkable, it has a right answer, and a machine that is good at retrieval is good at the comparison too.

Then there is the second job, which is a peculiar thing to explain to anybody outside your trade. You have to decide whether the claimed thing would have been obvious. Not obvious to you, sitting there with the application open in front of you and the whole idea explained on page four. Obvious to somebody the law invents for the purpose. That person knows everything published in the field up to a particular date and nothing after it. They are thoroughly competent and have no imagination at all. They have never seen this application, because at the moment the question is asked it has not been written. The entire value of that construction sits in what it is denied.

Which is why hindsight is the permanent hazard of your trade. Every examiner is warned about it early and still catches themselves at it years later, because once you have read the disclosure you know how the pieces go together. The application has just taught you. Two documents that sat in different classes and never met now look like an obvious pair, and they look that way because somebody showed you the join.

So look carefully at what is being offered to you as help. A model has read everything, including a great deal published after the date that matters. Its whole competence is putting together things that had not been put together. It has read the application, and an application is, among other things, a set of instructions for how to combine the references. Ask it whether a claim is obvious and it will answer, fluently, every time, with a reason that reads beautifully. What you have been handed is the most productive source of hindsight anybody has built, and hindsight is the one error this part of the law exists to prevent. A date filter keeps late documents out of a search. Nothing filters what a judgment is made of.

You have to say why a person in that invented position would have picked up those two documents, put them together, and had a reasonable expectation that the result would work. Your reason then gets read by an attorney whose client is paying to take it apart, and it may be read again by people above you both. A plausible reason produced after the fact is not a weaker version of that. It is the specific thing the requirement was written to keep out.

I don’t want to leave you with the comfortable reading, because I don’t think it holds. Your trade is not squeezed by machines first. It is squeezed by counts. Output is measured, everybody knows their own numbers, and every tool that has ever arrived in that building was absorbed as a raised expectation rather than as breathing room. If a system drafts the action and you review it, the real question is whether reviewing a fluent draft is the same activity as forming a judgment. I don’t believe it is. A convincing document pulls the reader toward agreeing with it, and it pulls hardest when the reader is behind.

The pressure from the other direction is worse. Applicants have the same tools you do. Drafting is cheap now, filings rise, and any rejection you write can be answered at length by the same afternoon, at a length nobody would have paid for before. The load grows at both ends while the part of your work that used to absorb it is the part being taken away. How that resolves is beyond me, and I am not going to pretend otherwise. It is the part of the problem that never turns up in anybody’s plan to speed things up.

There is a version of your office where the second job becomes the whole job, and the counts fall to match, because a decision that has to be defensible in writing is worth more than three that are not. There is another version where the drafts arrive faster and the counts rise, and the judgment quietly becomes a checkbox on somebody else’s document. Nothing technical decides which of those happens. It gets decided by whoever sets the numbers, and the numbers have never once been set to favor the slower half of the work.

Watch what your own hand does while you read the reason to combine in the last action you signed, out loud, to yourself. If it goes to the application, the thing you were judging has just taught you the answer. Then be honest about how much of your own pile gets decided that way. I have no idea what the answer is, and I notice that everybody outside your office is certain they do.