The Cost of Your Own Hands

You haven’t cooked a plate of food this week. If you have, it’s because something went wrong, and that turns out to be most of what this post is about.

That only holds in a kitchen with sections and a pass. In a small place you’re on the stove every service and running the business at the same time, which is harder, and the running still happens. It just gets done between orders.

Your trade has wanted rid of the admin for a century and it’s finally going. Ordering against a forecast of how many people are booked, with the weather in it. Costing, which means the margin on every dish, recalculated on Tuesday when the supplier moves his prices again. Menu engineering, telling you which plate makes money and which one is a hobby. Allergen records, temperature logs that log themselves, waste tracking, stocktake, the whole due diligence file. Rotas built against forecast covers and a labor budget, holiday, hours, the lot. Prep lists printed for each section from the same forecast. All of it can go, and it should, and any chef who says otherwise has forgotten what Sunday nights used to be.

The screens have taken more than that. In a lot of kitchens the tickets are already sequenced, courses fired on a timer, each section told when to start so a table of eight lands together. How far that’s got in yours I couldn’t say, because it varies wildly from one building to the next and nobody publishes it. Where it works, it does the timing better than a tired person shouting.

There’s one word your trade puts on every job advert and never explains. Consistency.

It doesn’t mean the recipe. The recipe has been written down since your first week, and writing it down never made anything consistent. It means that the same dish leaves your pass four different nights, cooked by four different people, and nobody eating it can tell which of them cooked it. That is not a food problem. It’s a problem about hands, and the hands belong to other people, most of whom you didn’t choose and some of whom will be gone before the year turns.

Notice what that does to the clever rota. An optimizer will give you a schedule that hits your labor budget exactly, and it will do it by treating your people as hours. They are not hours. Two of them can run fish on a Saturday and the rest will produce something that leaves the pass and shouldn’t. One of them is quick and drops standards under pressure. Another is slow, never wrong, and can’t be put on a section that gets hammered. The nineteen year old is either the best hire you’ve made or gone by spring. That knowledge is the actual rota, and it lives in your head because it would be unkind and possibly unlawful to write most of it down.

Then Saturday, and Saturday is the whole job compressed.

It goes wrong in three places at once, which it always does. Somebody hasn’t come in. The fish section is behind and getting worse, which you can hear before you can see. A table has sent something back, a large party has arrived early, and a delivery was short so one dish is running out whether you like it or not. Nothing about this is unusual. It’s a Saturday.

Every move you can make is priced. Pull somebody off larder and fish comes back, and now larder goes down instead, ten minutes later, quietly. Take the dish off and the fish section can breathe, and front of house has to tell a full restaurant no, which they will do badly, and you’ve lost the margin. Slow the tickets and the kitchen recovers while the whole room’s evening stretches by half an hour and the second sitting stops existing. Send it out anyway and you’ve decided what your food is. You pick between those in a couple of seconds and then again, all night, and by morning you couldn’t reconstruct one of them.

There’s one more option and it’s the one that feels best. Step onto the section yourself. Your hands are the fastest in the building and everybody knows it, and inside a few minutes fish is level again.

That one is priced too, and the price is the pass. While you’re on the section you aren’t reading it. The tickets stop being one picture and become a pile, timings drift, and the sauce section starts making decisions on its own. You come off having rescued one section into a kitchen that’s behind everywhere, and the person who was drowning has learned that when it gets bad you take over. Some nights that’s still the cheapest thing on the list. If the room is small, or the night is far enough gone, or the section that’s down is the one the menu hangs off, your hands are worth more on the stove than your eyes are on the pass. Nobody can hand you the rule for which night is which, because there isn’t one.

Underneath every one of those choices is money you’re spending in real time. Comp the table or don’t. Keep the extra body on past ten or send them home and risk it. Run the special that uses the trim or bin the trim. None of that is cooking, all of it is yours, and it’s why the good ones are so tired.

Large chains solved this a long time ago, and not with technology. They took the decisions out. Components produced centrally, portioned and bagged. A spec with a photograph. A screen telling somebody on the lowest wage in the building exactly what to do and when. The result is more consistent than most independent restaurants manage, and it isn’t a scandal, it’s well engineered. People eat that food and go back for more of it. Anybody in your trade who calls that beneath discussion is losing an argument they haven’t understood.

What that system did was move the job rather than delete it. Somebody wrote the spec. Somebody tasted a new crop of tomatoes and decided what this batch means for the sauce. Somebody chose which corners a nineteen year old is allowed to cut. Those people exist, and they sit a long way from any stove. What sits in the actual building is a manager who can report a problem and cannot fix one, because fixing it would mean changing something and nothing there is theirs to change. That’s the trade you’re being offered. Not the disappearance of your judgment, the export of it, and a job in a kitchen where the answer to a bad delivery is a phone call to somebody in an office.

Next Saturday, when a section goes down and your hands start moving on their own, stay where you are and send somebody else. Watch what it costs, because it will cost something. Then look down the line and count the people you could have sent instead. What would this year have had to look like for that number to be two?

Whatever You Think

Whatever you think. You’ll hear it in a front room this week, from a woman who met you this morning, about a decision she would have chewed over for a fortnight in any other month of her life.

She isn’t being lazy and she isn’t being polite. She’s been awake since the phone went, she’s already signed something she didn’t read, and she has arrived at the end of what she can do. So she hands the rest of it to a stranger holding a folder. The stranger is you, and you’ll take it, because somebody has to and there’s nobody else in the room.

Most of your hours are not that. Most of your hours are a scheduling problem with a hearse in it, and I’m not going to pretend otherwise, because you know what your week looks like. Everything hangs off the slot at the crematorium, which is the one thing you can’t move. Then a minister or a celebrant, an organist, a florist, a printer, a notice, a caterer, a gravedigger, and a mason who turns up months later. Somebody has to get the certificate, deal with the coroner’s office when it’s one of those, book the registrar, fill in the cremation forms. Cars, bearers, and the order they walk in. Deposits, the disbursements you pay out of your own account and chase afterward, and the family that never pays at all. A dozen suppliers, none of whom work for you, all of whom have to arrive somewhere at the same minute on a date somebody else set.

That is coordination, and coordination is exactly what’s being taken off people everywhere. It should be. Nobody joined this trade to hold on a telephone. Your price list is standardized now and mostly written for you. The forms are getting filled from the record. Scheduling can go, chasing can go, invoicing can go, and the job will be better for losing all of it.

The arrangement meeting is going the same way, and this is the part your trade is quieter about. Every choice in it is a field. Burial or cremation, coffin, cars, time, music, flowers, notice, catering. Somebody can sit at a kitchen table at midnight and answer the whole thing in a browser without speaking to a human being, and a great many people now do exactly that, and they’re satisfied.

Then, on the Thursday, her daughter asks whether they can see her mother.

Nothing in the folder answers that. You’ve seen the body and they haven’t. What a post-mortem leaves behind, what four days do, what can be put right and what can’t, all of that sits with you and none of it with her. People who don’t look sometimes carry it for years, and people who look at the wrong one carry that instead. So you say yes, and here’s what you’ll see, or you say no, and here’s what we can do instead. Either way you say it in a voice that settles the question rather than opening it.

They will never be able to check a word of it. Not that day, not later. The care you take with somebody’s mother happens in a room the family will never enter, done by people they’ll never meet, and if you were careless there is no version of events in which they find out. All of it runs on their belief that you weren’t.

That belief gets decided early, and I’d guess it’s decided before you’ve said anything useful. Something about how you came into the house, whether you sat down, whether you used her name, whether you looked at the photographs. I’m guessing, because I’ve never done your job and nobody writes this part down. If I’m right about that, it isn’t something you can go back and earn later. Turn up without it and the whole week is an argument.

With it, you get to run the family, which is the real work. The list of names comes off the widow, because she can’t make those calls and will sit with it in her lap for two days pretending she’s about to. You tell the son that his sister walks at the front. The daughter with the bad back is not carrying a coffin down a step, and that has to be decided in a way that doesn’t sound like you think she’s frail. Somebody has to ring the brother nobody’s spoken to since the wedding and get him to the church without it becoming an event. There’s a second family, who nobody mentioned, arriving separately. You say the ashes won’t be back for the date they wanted, and you say the same thing three times across four days, because none of it is going in.

On day one you also have to talk to somebody about money, in that state, about a number that’s large. No other trade gets handed that conversation on those terms.

Plenty of people don’t want to be run. They want the cheapest disposal available, no cars, no viewing, no arrangement meeting, a form completed online and the ashes posted back. They want it because the alternative costs real money for a performance they never asked for. My impression is that more people choose it every year, and that most of them are fine about it afterward. I’m going on nothing sturdier than what people say when you ask them. The usual answer from your trade is that they don’t know what they’re missing, which is what every trade says while its market walks out of the door.

I think that’s real and I think it takes most of the volume. What it doesn’t touch is the family that can’t be run remotely because it’s already coming apart, and those don’t announce themselves at the point of sale. So the position holds while the part of the work anybody actually buys stops being the part that teaches you to hold it. You learn to read a front room by sitting in hundreds of them, and the coordination is what paid for the sitting.

Think back to the last family who said whatever you think, and work out what you actually decided for them in the four days afterward. Not the coffin. The order they walked in, who carried, who got told what, which sentence you said three times. Now remember that nobody in that house will ever bury the same person twice. They have no second one to set yours against, no other firm’s version, nothing to compare it with. What you do in a week like that becomes the only funeral she ever had, and not one person there will ever know whether it was any good.

The List Was Wrong by Morning

There are always more jobs than there is day. Fencing. A gate somebody left open, the drill to get ready, a heifer that isn’t right. The sprayer wants a part and the scheme paperwork has a date on it. That trough in the far field which has been wrong for a fortnight. Forty of them if you sat and counted, and you never sit and count.

What you do instead is put them in an order, and then change the order all day. It rained in the night, so the list you made yesterday is already wrong. Not all of it. Two things have dropped off the top, two things that weren’t on it at all have arrived, and you sorted that out in the time it took to look out of the window and put the kettle on.

One admission before I go further. Farms are not one job. A dairy week and an arable week have almost nothing in common, and somebody with a hill flock would read both of them as office work. I’m writing about the ordering, because that’s the part I think they share, and I’d rather say that than pretend I can write to all of you at once.

Now the part that’s already gone, and you paid for most of it yourself, early, before anybody wrote an article about it. Guidance that drives straighter than you can and doesn’t overlap. Rates varied across a field according to what that bit of ground actually is. Pictures from above that flag the patch going backwards before you’d have seen it from the gate. Disease models that read the weather and tell you the risk is building. Moisture at the drier, weights, growth rates, the whole set of records that used to be a shoebox. Collars that spot a cow bulling more reliably than a good stockman on his best day. The compliance paperwork, the maps, the returns. The market information that your grandfather would have had to drive somewhere to get.

Every bit of that is real and most of it earns its money. Notice what all of it produces. It gives you better and better information about fields. Sorted by field, scored by field, mapped by field.

Nobody has ever produced the other thing, which is a ranking of your week.

The reason isn’t computing power. It’s that ranking needs the cost side, and the cost side isn’t in any of that data. A model can tell you the far field wants a fungicide and be exactly right. What it doesn’t hold is the rest of it. You have one sprayer and one operator. Tomorrow is the only day this week the wind is right. That near field wants doing as well and is worth more an acre. The heifer needs the vet, and the vet comes when the vet comes. Getting the far field done means not getting something else done. That is the decision, all day, every day, and it is made against your money and not against the crop.

Then the bit that gives the whole job its weight. Most of these windows don’t come round again. Drill three weeks late into cold wet ground and you don’t get a slightly worse crop, you get a worse crop for the entire year and there is no operation available to you afterwards that fixes it. Cut for silage on the wrong afternoon and you’ve decided what your cows eat until spring, and therefore what they give, and therefore what the milk cheque looks like in months you can already name. Your work holds two kinds of mistake, the ones that cost an afternoon and the ones that cost a year, and they feel identical at the moment you make them.

Sitting under that is a contract, and this is where farming stops resembling any other outdoor job. You sold some of this crop before it existed. You did it because the price was decent and because the bank likes to see it. The contract has a tonnage on it and a spec. If the moisture or the protein or the specific weight misses, the price doesn’t drop politely. It falls through the floor, or the load comes back. If the year goes badly enough you buy grain from somebody else, at the price it costs on the day, to hand over grain you promised when you were feeling optimistic in the spring. A good deal of what you decide in July has that document somewhere in it, and none of the software knows the document exists. That’s my reading from outside the gate, and I’ve probably made too much of it. Sell everything off the field on the day and that document doesn’t exist.

The cost side could be put into a system, and I don’t think anything stops it. The machinery is on a screen already, the labor is on a screen, and the contract is a document with numbers in it. Give it all of that and it will rank your week. It won’t forget the trough in the far field. It has no favorite field and no history with the neighbor. On plenty of weeks I’d expect it to order the jobs better than a tired man doing it at a kitchen table at ten at night, and I’d want to see its list before I wrote mine.

What the walking is for sits underneath all of that, and nobody counts the walking. Not the round itself, which is stock work and jobs and gates. The thing the walk produces is a reordering. You go out with a list ranked by fields and come back with one ranked by the farm. Standing in that gateway you could see this one can wait a week and that one can’t. You could also see that doing the far end today commits you to a route through the rest of the week, one that ends with the sprayer parked in the yard on the only dry morning. Nobody photographs that. It isn’t a measurement of anything.

There’s an awkward fact underneath all of it. You get one go a year at most of these decisions, and the ground is different next time because of what you did to it last time. So your evidence is a small pile of years, all of them yours, and none of them quite about the year you’re standing in. That’s why it worked last year is such thin evidence and why every one of us reaches for it anyway. It’s also why a confident stranger with a model can’t be argued with using anything except a year, and a year is expensive.

In the morning, before the kettle, watch which job you actually go to first and put it beside the one you settled on last night. They won’t always be the same, and the gap is the only record anybody keeps of what the walk was for. There’s one question no ranked list has ever asked you, and it isn’t about the far field. What is this year supposed to leave you with by the end of it?

Nothing Is Happening

Waiting is the decision. It doesn’t feel like one, because nothing happens at the moment you make it, and you’ll make it again at the next check, and again after that. Each time, she’s been at it a bit longer and the decision costs a little more.

Hand over the writing first, all of it, and your specialty produces more of it than any other. The booking history retyped every time she moves from one team to the next, and the timings plotted and plotted again. The summary that goes back to the community midwife, the returns that go off to whoever counts these things, the file that follows a birth around for weeks afterward. Obstetrics documents more heavily than anywhere else in the hospital because it gets sued more than anywhere else in the hospital, and none of that defensive typing was ever midwifery. The monitoring goes the same way, and it should. Nothing about a heart rate needs a person to be the one reading it, and the trace doesn’t get thinner because there’s another woman down the corridor. Your trade’s long argument with monitoring was never that the machine reads badly. It was that watching harder led to more being done, and that more being done was not the same as more babies going home. That’s a real argument and I can’t settle it, but notice it’s an argument about what to do with a reading rather than about the reading.

That distinction is the whole of the quiet hour. Every measurable thing in the room is normal. The question in front of you is not answered by any of them.

The question is whether this is a labor that needs time or a labor that has stopped, and those two look identical to anything that measures. What separates them is a set of things you’d struggle to write down. Whether the contractions are changing her or just happening to her. Quiet and inward isn’t the same as quiet and frightened, and the two don’t want the same response from you. She has asked you that question twice, and it matters whether she didn’t hear the answer or has gone somewhere else. How long since she ate. Whether the head has moved down since you last had your hands on her. What the liquor looks like, and whether the smell of the room has changed.

And whether she’s frightened, which your trade treats as clinical and mostly doesn’t write down. Fear slows a labor. That’s a claim and it’s mine, I can’t give you the mechanism behind it, and I’d rather say so than dress a working assumption up as physiology. What follows from it is that a good part of that hour goes on the room instead of on her. You turn the light down. You get the third person out, or you get her mother in. You stop the stream of people putting their heads round the door to ask how she’s getting on. You sit down, because standing up means you’re about to do something and she’ll read that off you before you’ve said a word. You’re treating her by moving the furniture, and it counts as treatment even though it looks like tidying up.

Everything else in the building is pushing the other way, and none of it is malice. There’s a guideline with rates of progress in it, written by sensible people for good reasons. There’s a board somewhere with your room on it and a woman in triage who needs one. A round will come through at some point wanting a plan, and carry on as we are has never once been written under the heading plan. Then there’s the handover. You go off at eight and somebody who wasn’t there picks her up, with the notes but not the hours behind them, and the safest move available to a person who has just walked in is to act.

Then the second half of the job, which starts the moment anybody has to be told anything.

She wrote a plan at her kitchen table in a calm week. It’s on the side, and it’s the calmest document in the room. It says things about lighting and about not being offered anything she hasn’t asked for. Not a stupid document, and not written by a stupid person. It was written by somebody who wasn’t in labor, about a labor that turned out to be a different one. Somebody has to take that document apart with her, line by line, without it landing as failure. The system’s version of this conversation is a form with a signature at the bottom. Yours is finding the sentence that lets a frightened person make a real decision in the gap between two contractions, and then making sure it was hers and not just yours arriving in her mouth.

It runs the other way too. When she says no, you have to put that into the record as a competent refusal rather than as a midwife who couldn’t persuade her, because those two are written differently and read very differently later. When the team comes in they’ll talk about her in the third person while she’s lying there listening, not because they’re unkind but because that’s how a handover sounds. You translate that back into something she can hold. Same facts, different currency. The system speaks in risk and she is speaking in fear, and nothing converts one into the other except a person who has both.

That cuts both ways, and it has to. A unit can hold out for a normal birth so consistently that it stops being a judgment and turns into a house style. It can drift the other way just as easily, and reach for the theater list before anybody has had time to think. Once either one is the style, nobody in that room is deciding anything, and the cost of not deciding lands on somebody who never got a say in it. So the job isn’t waiting, and anybody who tells you the residue of your work is patience is selling you something. The job is holding two accounts of the same labor at once, hers and the guideline’s, and knowing at each point which of the two is currently wrong.

Picture that quiet room and put everybody in it who has a view about what should happen next. The woman, the man holding her hand who wants somebody to do something, the round that’s coming, and whoever wrote the guideline years ago in a different building about a woman who isn’t this one. Not one of them can hear any of the others. You’ll translate all of it, in both directions, as the only person there that both sides are certain belongs to them.

The Second Van

Somebody was in this building before you and stopped work a very long time ago. The sheathing says when. So does the color of the cores when the front comes off, the shape of the accessories, and the clips and how far apart he put them. All of it arrives before the bag is off your shoulder.

That isn’t a party trick. The date tells you what the rules were when that person stood here. The rules tell you what they were allowed to do, and what was normal to do, and so what else is likely to be in this building before you’ve found it.

There are two electricians in this trade and they hold the same card. One of them spends the week on a site with a drawing, putting in an installation that doesn’t exist yet. The other one turns up at something that already exists and has to work out what it is. I don’t know how the trade splits between them, and I’d guess the first is the bigger employer and always has been.

The first van’s work is the one that comes apart, though not all the way and not the way an outsider imagines. The design has been arithmetic for years, and it’s now arithmetic that software does properly. Cable sizes, volt drop, disconnection times, fault levels, the schedule of circuits, the labels. Prefabricated lengths delivered cut, modules built in a factory and craned in. Setting out done by a machine that puts a dot on the floor where the hole goes. Test instruments that keep their own results and write the certificate. What none of that touches is somebody on their knees drilling, clipping and terminating, and anybody who tells you a factory has taken that off the job hasn’t been on one lately. What has gone is the deciding. A good part of the first van’s day is already instructions being carried out by somebody who didn’t write them.

The second van has a problem no drawing has ever solved, and it starts with a number that’s wrong.

You test the ring in the kitchen and the readings don’t agree the way they should. That’s the whole of what the instrument gives you. It doesn’t tell you whether there’s a spur off a spur behind the units. It won’t say whether somebody broke the ring in a kitchen refit and joined it up again in a box under the floor, or crossed two circuits in a loft in a hurry. The number is a value. What you need is a story, and the story has a person in it.

So you start reasoning about somebody you’ll never meet. A kitchen went in around a certain year, and that year people did a particular thing when the cable wouldn’t reach. The house had an extension on the back, and whoever wired that extension took a feed from the nearest thing they could get at, which is the ring you’re standing on. That is the work. Not measuring, which the instrument does better than you, but building a picture of what a stranger was thinking while they solved a problem you can no longer see.

The trap that shows this best is two circuits sharing one neutral. Somebody did it to save a cable run through a solid wall, and nothing in the rules of the day said much against it. It was tidy, and cheaper for the customer. Then it sits there for decades doing nothing wrong, until somebody fits protection that watches both sides of a circuit and expects them to match. Now it trips, apparently at random, and the fault is not in any component. Worse, switch off the breaker for one of them and the wires are still live from the other. That is a dangerous thing to walk into, and no drawing anywhere shows it. What finds it is knowing that people used to do that, and roughly when, and why they thought it was fine. Nobody ever wrote that down. The regulations of the day got printed and taught and argued over, and what a man actually did on a Friday when the cable wouldn’t reach went into a wall instead.

Your trade doesn’t trust the paperwork, and has built a procedure out of not trusting it. You prove the circuit dead, then you prove your tester on something you know is live, then you prove the circuit dead again. Every electrician does it every time, because the label on the board is only a claim made by a stranger. Plenty of them are wrong, some from the day they were written and some from the day a circuit got moved and nobody said.

There’s a version of the future where the second van’s knowledge stops being scarce, and it has nothing to do with the records getting better. Your trade has started writing itself down, in public, for free. Forums, videos, and the photograph of somebody’s appalling find posted at eleven at night, with a comment underneath naming the decade and the reason. Then the long arguments about what the regulation says against what everybody around here actually did. All of that is the second van’s knowledge, put into words at last by the people who hold it, and anything that reads the lot ends up with a decent working model of what electricians used to do and why. Which is the thing I’ve just told you nobody writes down. That is the best case against me, and I don’t have a tidy answer to it.

What it can’t reach is the ordinary. The nail through a cable behind a skirting board. The insulation packed over a cable in a loft until it runs hot. The socket added in a kitchen refit, by a fitter who doesn’t consider himself to have touched the electrics. The son-in-law who ran a supply to the garage and did a reasonable job of it except for one thing. Nobody photographs any of that, because nobody involved thinks they’ve done anything worth mentioning. The careful ones and the careless ones leave exactly the same paperwork behind, which is none.

One thing about the man who was here before you. He wasn’t an idiot. Nearly every trap in that building was somebody’s sensible answer to a problem that has since disappeared, worked out under the rules of the day, in a hurry, for a customer who wouldn’t pay for the proper version. You did the same thing yourself this year and you can name the job.

Write it down tonight. The job of yours you can name, the one you did in a hurry for a customer who wouldn’t pay for the proper version, and one line saying why. It’ll cost two minutes and get thrown away. Somebody opens that wall on a wet Tuesday in thirty years and starts reasoning about you, the way you were reasoning about a stranger this morning. He’ll date it right off the cable alone, which is the part he’ll find easy. What he’ll want is the reason, and the only place it has ever existed is your head.

Nobody Wants You to Stop

He nods. It isn’t quite a lie and it certainly isn’t laziness, because he nods for a reason that makes complete sense from where he’s sitting. He wants the lesson to move past him. He is right that it will.

You saw it. Every teacher reading this saw it in their first term and has seen it most days since, and seeing it has never been worth very much on its own.

Take the delivery first, because that argument was settled long before any of this arrived. A clear explanation of the thing you’re teaching today has been available to anybody who wanted one for a very long time. Books had it. Then the videos had it, made by somebody with better lighting and a second take. Any child with a phone has been able to get a cleaner explanation than the one you’re giving at half past nine on a Thursday. That was true before any of this and it didn’t close a single school. Now the explanation is free, endless, patient, and reworded on demand for whoever is asking. Around it goes the rest of the load. Marking. The resources you built at the kitchen table, and the differentiated version for the four who need it. Reports, comment banks, the cover work you set from home while you were ill. None of that was ever teaching, and the profession has been saying so, loudly, to anybody who would listen, for about as long as it has existed.

There’s a harder concession behind that one, and I think you should make it rather than fight it. Something that has seen every answer that boy has given this term will find the hole in his understanding more reliably than your tricks do. Not more kindly. More reliably. The whiteboards, the exit slips, the question you ask the class while watching one particular face. All of those are clever workarounds for not being able to see inside thirty heads at once. A system that never gets tired, and never wants the lesson to have gone well, beats every one of them. How far that goes I’m guessing at, from outside a classroom, which is the wrong place to guess from. Somebody who has watched one of these systems wave through a child who guessed well would tell me I’ve been generous here, and I’d want to hear it.

Knowing costs nothing. Acting on it costs you the room.

The arithmetic is brutal and every one of you does it in about a second. Stopping for one child spends the time of the twenty-nine who have it, and those twenty-nine are also your responsibility, and they are sitting right there, and three of them are already turning around. There’s a boy at the back who finished four minutes ago. There’s a girl whose mother has told you twice this year that she isn’t being stretched, and the mother is not wrong. The lesson has a shape and you can feel the shape starting to go. Everything you can see and hear in that room is telling you the same thing, which is to press on and pick it up later.

The turn in this is what makes the job hard rather than merely tiring. The child you would be stopping for is the one who least wants you to stop. He knows that if you stop, everybody finds out it was him. He nodded to get out of exactly that, and stopping now hands him the thing he was avoiding. So you’re refusing on behalf of somebody who is quietly voting against you, along with everybody else in the room, including the clock on the wall.

Whether that refusal is worth making depends on your subject and a history teacher would be entitled to roll their eyes if I pretended otherwise. Some things can be met again from another angle next month and land properly the second time. Others compound. If he can’t hold a fraction in his head today, everything you teach him afterwards about proportion and rates and algebra sits on ground that isn’t there. Two years on he looks like a boy who is bad at your subject, to everybody, and to himself first. Knowing which kind of hole you’re looking at is most of the judgment, and it’s the part that takes years to get right.

Refusing to move on almost never looks like stopping the lesson dead, whatever the phrase suggests. It’s the seating plan you rebuild on Sunday night. It’s the ten minutes at the start of Thursday spent on something you already taught. It’s giving up your own break, or handing the class something to be getting on with so you can crouch by one desk for a few minutes without it turning into an event. Every one of those is paid for out of your own time or your own control of the room. And none of it happens quietly. Carry that debt for three weeks and you are behind the other class, in a department where somebody puts coverage on a spreadsheet, before an assessment written on a date nobody consulted you about.

Give every child a system that goes at their own speed and refuses to advance until they’ve got it, and there is no room to hold up and no twenty-nine to spend. Nobody waits for anybody. The pressure I’ve described is a side effect of teaching thirty people at once, which is an accident of how schools are funded rather than anything to do with learning. That is a serious argument and it deserves better than the reflex it usually gets.

What it runs on is the boy being honest with the machine. He wasn’t honest with you, and he had good reasons, and the machine gives him more privacy in which to be dishonest rather than less. He will click until it lets him through, or he will sit there quietly not clicking at all while a screen waits patiently for a child who has decided to be somewhere else. So the refusing still has to happen somewhere, and I’d want to know where before anybody builds a school on the assumption that it doesn’t. It won’t necessarily be you. It could be one adult watching a screenful of children who have all gone quiet, paid less, with far less room to hold anything up, and the job I’ve described would still be getting done. Whether your trade survives that arrangement is a different question, and nobody has answered it.

Next lesson, when you get the nod, don’t take it. Give him something a step off the example you just worked, and stay beside him while he starts it. Watch the rest of the room rather than him while you wait. Then count how long you last before you move on, and be honest with yourself about what moved you.

The Word You Use Instead of Worried

Bed four is the same color he was at eight o’clock and he isn’t the same man. The chart doesn’t know that yet. Nobody wrote it down, and nobody thought it was clever. That part is real, and it has never been the difficult part of your job.

The difficult part happens in a corridor, on a phone, to somebody who is already walking somewhere else.

Start with what’s leaving, because a great deal of it is and nobody will grieve for it. The charting first. The admission assessment, the care plan, the fluid balance, the risk assessments repeated every shift, the same facts typed into two systems that were bought in different years and don’t speak. Then the observations, where the machines are already better witnesses than you are. They watch continuously instead of every few hours, they’re never late because the whole bay went off at once, and they don’t get pulled away to a relative in the corridor. Then the drug round. Cabinets that count, scanners that check the patient against the box, the arithmetic done before you ask for it. Almost all of that can go, and honestly it should have gone years ago. None of it was ever the work. It was the ward’s record keeping, and it landed on you because you were the one already standing at the bed.

You can see the thing. Somebody else has to act on it, and that somebody is not you.

Your trade solved this once already, and the solution is hanging at the end of the bed. Take a handful of numbers you were taking anyway, add them up, and above a certain figure the response stops being a request and becomes a rule. Notice what that device is actually for. It’s assembled out of numbers you had already taken. What it hands you is a figure the building has agreed in advance to act on, which is a different sort of thing from knowing. That’s my reading of the score rather than the official account of it, and I’d hold it loosely. Watch one get a patient moved and you’d probably call it knowledge after all.

You have a word for this too, and it isn’t worried. Worried is about you, and it invites a reasonable person to manage you instead of the patient. Concerned is a position. Say you’re concerned about bed four and you’ve put something on the record that somebody now has to answer. Same nurse, same patient, same evidence, and a completely different obligation on the other end of the line.

Then the call, which is the actual work. He isn’t refusing you. He’s ranking. He has a bleep, several wards, and a list in his head where every item is somebody’s patient. You aren’t telling him bed four is unwell. You’re telling him bed four is more unwell than the person he’s already walking toward, in under a minute, to somebody who’s been awake since yesterday morning.

You were taught a format for it. The situation, the background, what you think is going on, what you want him to do. The first three describe something. Only the last one asks a tired man to change what he was about to do next, and it’s the part that goes missing when you’re new on the ward and he sounds busy. Nobody can teach you the tone for it either, because the tone is different for every person you’ll ever ring.

The same words don’t work the same for everybody. A sister who’s been on that ward for years says come and look at this, and the registrar comes. Somebody two years qualified says the identical sentence and gets asked to repeat the observations and call back. None of that difference is in the patient. It’s who’s speaking, how many times that person has been right in front of this particular listener, and whether the last one they escalated turned out to be something. You assemble it slowly, you can’t put it on an application form, and it doesn’t come with you when you move hospitals.

You carry the cost of being wrong in a way nothing else on that ward does. Get somebody out of bed at three for a patient who turns out to be fine and nobody will be unkind about it. It gets remembered anyway. Do it twice in a fortnight and you can hear it in the pause before he answers.

Put continuous monitoring on every bed, which is coming and in places has already arrived. The score keeps itself up to date and does it all night without being asked. On paper that lifts the whole problem off you. What it puts on a real ward is a stream of alerts, most of them a lead that’s come off, a patient who sat up, a probe on a cold finger. Staff learn the true rate of those inside a week, the way they learn everything else on a ward, which is from each other and without discussing it. Then somebody has to decide which of the ones going off right now is the real one, and that person is standing exactly where you’re standing, arguing for a device that’s been overruled all morning. The figure used to be scarce, and scarcity was the whole of its power. Make it continuous and it stops being an argument and starts being noise.

Give the machine its best case, because it has one and it isn’t small. A system that watches every bed and rings somebody with a number nobody can argue with may well get bed four seen sooner than you managed. That’s a better night for bed four, and a better night for bed four is the outcome the whole ward exists to produce. I’m not going to dress that up as a loss because the losing end of it is yours.

Ask the registrar you rang last night what actually made him come. Not what you said, which you can both recite, but whether it was the word you used, or the fact that it was you saying it, or the last time you rang him and turned out to be right. A screen at four in the morning can hold the first of those and nothing else. Everything except the believing is already being built.

Everything You Disturbed

Somewhere in the technical log of the aircraft you worked on last night there’s a line with your name in it and a number that belongs to you and to nobody else. Until that line exists the aircraft is a large object parked at a gate. Once it exists, people get on it. Nothing else about it changed. Somebody wrote in a book.

Most of the rest is going, and a good deal of it already has. Planning, scheduling, the work pack, the parts chase, the reliability figures, the manual you used to hunt through. The troubleshooting that made the reputation of every engineer you learned from, which now tends to arrive mostly answered. Images from inside an engine assessed by something that has seen more of them than you have. The check that the paperwork is internally consistent and complete. You won’t miss any of it, and the next round will take more.

In your trade the objection isn’t a forecast at all. It’s a history. Aircraft have come apart for reasons that traced back to a name in a box for work that was not done the way the box said, and each time it happened the rules got written again around the gap. So nobody has to persuade you that an approval can go hollow. Your industry keeps the case files on exactly that, and the trade already has its own quiet words for it, spoken about somebody else. Go through a heavy check and count the boxes if you have the patience. Every task card, every step, a name in a box, and nobody honest would claim each one of those is a considered act of judgment.

Three ordinary structural facts keep that box open, and not one of them is mystical.

The first is that your line is about an object rather than a document. An approval goes hollow most easily when nothing can ever contradict it. Nobody can go back and check that a reviewer really reviewed. But the aircraft persists, and everything you said about it is a claim that reality is in a position to argue with. The next engineer who opens that panel is looking at what you left. So is the one at the next check, and the one years from now, and anybody investigating anything. A statement about a physical thing that stays available for inspection is a different animal from a statement about a piece of paper.

The second is that the record exists to attribute. That’s not a side effect of good filing. The whole architecture of aircraft records was built so that when something is found, somebody can walk it back to the hands that did it and the person who said it was finished. Your approval is personal and can be taken from you personally, separately from whatever happens to your employer. That is unusual. Most professional consequence lands on a company and stops there.

The third one is the whole trade, and it’s the one I’d put the argument on. Your release isn’t a claim that a task was performed. The task cards already say that. What you are certifying is that the aircraft is in the condition the paperwork says it is, including every single thing you disturbed to get at the job. The panel you opened. The connector you pulled to reach past it. The blanking cap, the rag, the lamp you put down inside the bay while you were reaching for something, the part you moved and meant to come back to.

There is no task card for any of that. There is no data source for it either. A machine can confirm a torque figure, a part number, a serial number, a test result, and it can confirm all of them better than you can. It cannot answer whether there is something in there that shouldn’t be, because that question has no record attached to it anywhere. It has a person who was there, at night, with the panel open, who knows what they did.

What tests any of this is never dramatic. The aircraft is due out, the fault is small, and there’s a list agreed with the regulator of things that may be broken and still fly, with conditions attached and a clock running. Defer it and everybody’s morning is fine. There’s no shouting. There is a delay code, and delay codes have owners, and yours is on the technical one. Nobody in operations will ask you to sign anything. They’ll ask how long. They ask because they can’t do it themselves, and neither can the duty manager, unless he holds the approval and is willing to put his own name and his own number in the box. He can go and find another engineer, which happens, and which is the honest weakness in all of this. What nobody can do is fill that box in on somebody else’s behalf.

Where this actually breaks has nothing to do with machines. You sign for work other people did. Contractors, agency staff on a heavy check, an apprentice, the shift before yours. You decide how much of it to look at and you cannot look at all of it. If that name ever becomes something people write without looking, it’ll happen there, and it’ll happen for reasons that have nothing to do with software.

Go out to the stand tonight and watch one leave. Fueled, loaded, crew aboard, doors about to close, and one box in the log still empty until somebody walks up to it. Nothing out there can fill that box except a person holding an approval. Then work out what it would take for that to stop being true, and notice that the answer isn’t a machine. It’s a rule that gets rewritten, a shortage of engineers, and one bad month.

Somebody Solvent

There was a time when a man crushed at work had to prove somebody had been careless, and mostly he couldn’t, so he got nothing. Then a scheme was invented that paid him on a schedule whatever had happened, took the question of fault out of it completely, and nobody has seriously proposed going back. The consequence didn’t disappear when the blame did. It got priced, loaded onto an employer and passed to an insurer. Finding a person at fault was thrown out as an expensive way of deciding who deserved help.

That is what liability does. It moves. So I want to argue the other way here, against the claim running through this block, which is that what’s left of your job is a place where consequence has to land. Call that the answerability argument. I don’t think the place is as fixed as it sounds.

Start with what your signature already isn’t. If you’re employed and you get something wrong, you don’t pay. Your employer pays, or its insurer does, and the arrangement is so ordinary that most professionals have never read the clause that does it. Your engagement letter probably caps what anyone can recover from your firm at some multiple of the fee. The consequence has already been sized down to something a business is comfortable carrying. Directors get indemnified as a matter of course. Clinicians get indemnified. Nobody thinks any of this is a scandal, and nobody thinks it has hollowed out the professions.

Then look at who actually runs your blame when it comes. The insurer does. It decides whether to defend or settle, and it will settle a case you’d have won if settling is cheaper than winning, over your objection, with a clause saying nobody admits anything. Professionals find that obscene the first time it happens to them. It is also completely normal, and it means the thing called your personal exposure is not even yours to conduct.

That leaves a license and a register. That is real, and it’s the strongest ground the answerability argument has, so be exact about what it covers. Look at what people actually get struck off for. Dishonesty. Covering up. Practicing while unfit. Working outside their competence and lying about it. Not, in the ordinary run, being wrong about a hard judgment made in good faith. The personal residue isn’t liability for your decisions. It’s liability for your honesty, and that is a much smaller thing than the profession likes to imply. That reading of the registers is a guess, and it’s the load-bearing one here. I’ve never sat on a disciplinary panel. If competence cases are commoner than they look, this whole paragraph goes.

Now the part that makes this hard to argue with, and it isn’t the part the industry uses.

The push to relocate blame doesn’t come mainly from employers who want to be rid of you. It comes from the people who get hurt. Put yourself on the wrong end of a bad decision and ask which defendant you’d rather have. An individual with a capped policy, a mortgage and a good story about how difficult the case was? Or a manufacturer with deep pockets, a documented process and a certified design, whose machine made the same decision everywhere else it was fitted, in a way anybody can go back and examine? The second one is easier to sue, easier to prove against, and far more likely to pay. Product liability exists because that preference won, slowly, against the fierce objection of everyone who preferred the old arrangement. Claimant lawyers will get to machine decisions before the vendors do.

There is also a model for the whole thing already running, at scale, in the most safety-obsessed corners of the economy. A design gets certified once, by a body, against a standard, using a process that is itself audited. Every unit built afterward is covered by that certificate and no individual signs for any of them. Nobody calls it a moral vacuum. The trades where one named person signs off each individual thing are not the general case. They are the exception, and they survive for historical reasons rather than principled ones.

And these are jobs, which is the part that gets forgotten in the philosophy. The named person exists because a rule says a named person must exist. Rules like that get rewritten, usually after an accident, by people who never asked the profession whether the signature felt meaningful. The day a regulation says an approved system may discharge the duty, that position doesn’t get argued away. It gets legislated away, and the answerability argument will still be true and will not matter. Anyone who has watched a statutory named-person role get filled by whoever satisfied the requirement, rather than by whoever had the judgment, knows how thin that anchor already is.

The reply I’d expect is that blame is not really about paying afterward, it’s about deterring beforehand, and a company can’t be frightened the way a person can. I think that gets it backwards. A premium is a continuous, precise, forward-looking deterrent that reprices every year on evidence. Fear of losing a license is blunt, rare and mostly imaginary, because the thing hardly ever happens and everyone knows it. Firms change behavior when the price of cover moves. That is deterrence with a dial on it, and it works on exactly the decisions the answerability argument says only a frightened individual can make.

One thing gives me pause, and I’d rather name it than hide it. Every case I’ve cited is one where the decision had already been standardized before the liability moved, and the jobs in this block are chosen precisely because theirs haven’t been. That isn’t nothing, but it’s a delay, not a defense.

So here is the test, and it’s cheap enough that somebody should run it. If the human in the loop is doing load-bearing work, taking them out should raise the cost of covering that work. Watch the market for cover on a machine-certified process against the same process with a person signing. If the machine version is dearer, the signature was carrying something. If it prices the same or cheaper, it was decoration, and everything else is a trade defending its own dignity. The second test is what happens after the first serious harm caused by a system that replaced a signer. If the response is a new named human role, the answerability argument was right. If it’s a standard, an audit regime and a recall, it wasn’t.

Go and look at the certificates in whatever building you’re sitting in. The lift has one, the boiler has one, and somewhere there’s a plate on a pressure vessel nobody has read since it was fitted. Some of those approvals belong to a person and some belong only to a body, and nobody in the building could tell you why the line falls where it does. It wasn’t drawn by anybody thinking hard about responsibility, but one trade at a time, by whatever turned out to be cheaper to cover.

You Passed This Last Time

Every inspection you do happens on somebody else’s ground, in somebody else’s hours, with somebody else’s money running. That isn’t a detail about where you park. It is the shape of the whole job, and it’s why the checklist was never the hard part.

The checklist half is going, all of it, and good. Plan checking against code was always a matching exercise and machines match. Permit paperwork, scheduling, the notice, the report, the photographic record indexed by date and location. A scanner that finds the steel in a slab and reports the spacing. Sensors in the pour. A model checked against the code overnight, with every clash listed by morning. Every bit of that can go and the trade will wave it off. The paperwork is the part of your week that has grown while the inspections stayed the same length.

Some of the visit is going too, and it already has in places. Video inspections work, and for more things than the trade wanted to admit. They also come with a fact worth keeping in front of you. You see what the camera is pointed at, and the person holding it is the person being inspected.

What survives isn’t in any code, and the first piece of it is where to look. An hour, and a building, and most of it will still be covered up when you drive away. The rule every experienced inspector runs on isn’t written anywhere, because writing it down would be an insult. The standard of the visible work tells you about the work underneath it. Straight lines, clean cuts, a tidy site, materials stored properly. That is a judgment about a crew and about the man running it, not about a building, and no clause in any code will ever contain it.

The second piece is the one that puts your stomach in it.

You’re standing over a slab before a pour. The steel is in and it’s wrong. Not dangerous, not catastrophic, because that almost never happens. The bars are sitting too close to the face, so there won’t be enough concrete over them to keep the weather off for the life of the building, and where two bars overlap the overlap is short. It will pass any casual look. It will also be under the pour before lunch.

The truck is booked. The pump is on hire for the day. The crew is standing in the cold with nothing to do. Weather is right today and might not be tomorrow, the owner has a loan with dates on it, and the site manager has a schedule he’s already been shouted at about. Every single person in front of you wants the same answer, and it is not the one you’re about to give.

You are the only person on that slab with nothing to lose today. That is the sentence people reach for about your trade, and it isn’t true. You’ll be back on Thursday for the drainage. You’ll be on this site until it tops out, then on the next one this builder does, for as long as you both work in this town. Most people who have to tell a stranger no get to leave afterward. You’ll see him at the supply yard on Saturday.

Then he says the sentence, and my guess is that most of you have heard it in the same flat tone. You passed this last time. Sometimes it’s a bluff. Often enough it’s true, because he built it that way last time and it went in and nobody said anything. Now your own consistency is the thing being used against you, and the choice he’s offering is that you were wrong then or you’re wrong now, and he genuinely doesn’t mind which.

The pressure that follows doesn’t come from him. It comes from your side of the fence, and it arrives as a phone call you are not on. It is never that you were wrong. It’s that you’re obstructive, inconsistent, holding up a job, difficult to deal with compared to the others. Somebody in your own office takes that call. Somebody in your own office passed this same detail across town last month, and that colleague is the comparison being drawn. Where the developer chooses who inspects, you’re a supplier, and difficult is a characteristic that can be swapped out.

Inspection is a sample, and every one of you knows it. You looked at part of a slab for part of an hour on one morning. Things get covered between visits, work gets changed after you’ve gone, and a camera left running plus a scan of every pour would find more than your eyes ever will. All of that is true. It is also not what the argument was ever about.

What a finding can do depends entirely on who is holding it. A report saying the laps are short at eight in the morning is worth what the person carrying it is willing to do at nine. A stop notice is a piece of paper too, and it works because somebody signs it, stands on the slab and takes what comes back. Remove that person and the finding still gets written. It gets logged, answered and closed, the pour goes ahead on time, and the building carries on being what it is.

There’s a smaller version of the same thing that never shows up anywhere. You point at something, three men fix it while you stand there, and you pass it. No fail, no notice, no dispute, no record that anything happened. Nobody has ever costed that, because it produces no document at all.

And the reason it stays rare has nothing to do with what you can see. It’s that saying no is something you do to a stranger, on his ground, in front of his crew, knowing you are due back on Thursday. Then the complaint travels, and not to him. It goes to your own office, where somebody who wasn’t on the slab decides whether you were right or whether you were being difficult. That is the load the job actually carries. Not the finding. The willingness to make a finding cost somebody something, when the argument about it lands on your own side of the fence.

Here is what I’ll give away, and it’s not small. Most of your week isn’t any of this. It’s driving, notices, phone calls and inspections where everything is fine and everyone is decent about it. A job whose defensible core is one wet morning out of many is a job that employs fewer people, covering more sites, arriving with the report mostly written, and no argument about what’s left will change that arithmetic.

For a month, count the arguments that ended up in your own office rather than on the slab. Then look at who was standing in front of you each time. Every one of them was worse off that day and could tell you exactly by how much. Not one of them can point at whoever is better off, and neither can you, and no trade has ever won an argument standing on that.