
An ISTP man’s whole faculty lives at the machine, in the minute the thing is going wrong, and it does not survive being written down. Organizations have exactly one currency for a man like that, and it is a promotion out of the work, and when he declines it they file the declining as his answer.
My third week at Cottrell Molding, press nine went down on a housing that ran eight hundred thousand a year, and I stood there for two hours with the tooling vendor on speakerphone while they told us it was the mold.
Kerry Ludtke came over at the end of second shift on his way out. He had his coat on.
He laid his palm on the barrel about a third of the way back, then again up near the nozzle, and stood there.
“Zone three’s lying to you.”
The setter said zone three read three ninety, same as the setpoint.
“It reads three ninety.” He took his hand off. “Pull the thermocouple.”
We pulled it. It had backed out about a quarter inch and was reading the well instead of the melt, and the actual melt temperature was somewhere north of four twenty, and we had been degrading material into that cavity for a day and a half and calling it a mold problem. Nine minutes, coat on, and the vendor was still on the phone.
I asked him afterward how he knew.
“It’s the wrong hot,” he said.
Kerry Ludtke started at Cottrell in 1988 as a material handler, twenty years old, out of a high school that no longer offers shop. He was setting molds by 1991 and running process by 1996. He is a compact, unhurried man who does not fill silences, and if you ask him a question he will look at the thing before he looks at you. He does not carry a phone on the floor. He has taken apart, on his own time, in his own garage, a snowblower engine, three transmissions, a plate freezer, and — this is documented, there are photographs — a commercial dishwasher belonging to his sister-in-law’s restaurant, which he reassembled with two parts left over and which ran for another nine years. He has never in his life said the phrase best practice.
Cottrell is a custom molder in Bexley, ninety-one presses at the peak, now sixty-eight, running automotive interior parts, a line of housings for a water-treatment OEM, and — since 2012 — a book of Class II medical components that is the reason this plant still exists. Guy Cottrell’s father started it in a Quonset in 1961. The good conference room has a piece of the first part they ever molded set in a block of clear acrylic, and it is not a very good part.
I came in 2009 from a device OEM to build the quality system. I was thirty-three. I wrote most of the procedures this plant runs on, including the ones that made what Kerry did impossible, and I would write them again, and that is the problem.
The opening I have right now is a process engineering seat with a number attached that would have been unthinkable in 2013, and Kerry Ludtke is fifty-eight and running a press at a nine-machine job shop in Neola with no benefits until his wife’s plan picks him up next year, and I have had the requisition in draft for three weeks because I cannot work out what belongs in it.
Guy said the same thing about him for twenty years, at reviews, in the lot, to customers, to me:
“We offered Kerry everything. He didn’t want it. You can’t make a man want a career.”
Cushion
There is a number in injection molding called the cushion.
When the screw pushes material into the mold, you never let it run all the way forward. You stop it with a small amount of melt still in front of it — a quarter inch, an eighth — so that when the machine switches from filling to packing, there is still something for the pressure to act through. Without it the screw bottoms out and the pack pressure has nowhere to go and the part comes out short, or sunk, or with a hole where a wall should be.
The cushion is not the point of anything. Nobody ships cushion. It is the margin that lets the process be a process instead of a series of accidents, and a cushion that is quietly shrinking, shot over shot, over three days, is the single most useful thing a person on that floor can notice.
A process that is genuinely in control is boring in a specific way. The parts come out the same, the cycle lands within a couple of tenths of a second every time, the screw recovers at the same rate, and a person walking the aisle has nothing to look at. What that means in practice is that the entire skill of the job is detecting change in a system whose whole purpose is not changing — a tenth of a second here, a hair of flash on a parting line there, a cushion that read point two one on the first shift and reads point one nine now. None of those is a defect. Every one of them is the front edge of one.
The lunch truck comes at nine forty at night and has come at nine forty at night since before I got here, and second shift will not eat before it comes even on the nights it is late, which it is, because the man who drives it also plows in the winter. Clell Slota, who ran production for nine years, once tried to move the break to nine thirty for a scheduling reason and the entire shift ate at nine forty anyway and nobody said anything to him about it, ever, and after about a month he moved it back.
The acrylic block in the good conference room contains a plastic bracket from 1961 that is warped, short on one leg, and has a sink mark you could set a pencil in. Guy’s father had it cast in the acrylic in 1986 and Guy has never let anybody replace it with a better part, and when customers ask he tells them it is there to remind everybody what the first one looks like. Three separate sales managers have proposed a nicer block. There will not be a nicer block.
I have been learning Portuguese for six years. I have no plans involving Portuguese, and my husband has stopped asking.
The Sound
Everything Kerry had was in two places and neither of them was on paper.
The first was a model of what a press was actually doing, built out of taking things apart since he was a kid, which he owned outright and which answered to nothing except whether it predicted. He did not know what the manual said. He had never once, to my knowledge, cited a manual to anybody. He knew what the machine was doing, and if the machine and the manual disagreed he assumed the manual was describing a different machine, and he was usually right about that.
The second was that the model only came fully on when he was standing in front of the thing. Hand on the barrel. Part in his hand, still warm, gate side up, turning it. Listening to the screw recover. He would walk a row of eight presses and stop at one of them for no reason he could give you, and about seventy per cent of the time that press had something wrong with it that would have shown up in about two days.
I timed it once, quietly, over a quarter, because I did not believe it and because I am the sort of person who resolves that by making a spreadsheet. Nineteen stops. Thirteen of them turned into something.
The thing nobody could ever get into a number is that the presses he touched did not break. Downtime on the cells he ran was, over the last decade I have data for, about a third of the plant average, and when I tried to put that in a business case in 2019 the corporate finance person asked me how much of that was attributable to the individual versus the tooling and the product mix, which is a completely fair question and which I could not answer, and the business case went in without it.
Nettie Pesch has run process on the automotive side since 2007 and is the second-best person in this building at all of it. What has been said about Nettie for nineteen years is that she is very thorough. Nobody has ever described her as having an instinct for the presses. Nobody has ever put her hand on a barrel in a story. She has been offered exactly one thing in nineteen years, which was the same lead job Kerry turned down, and she took it, and it did not lead anywhere either, and the two of those facts do not weigh the same and the accounting for hers runs elsewhere.
The material handler on second shift in 2013 was a kid named Tavo Ruelas who followed Kerry around for the better part of two years.
“Why’d you stop at this one?”
“Listen to it.”
Tavo listened to it. “Okay.”
“You hear it come back up at the end?”
“I hear it.”
“It didn’t used to do that,” Kerry said, and kept walking, and Tavo runs process on the medical cell now and is very good.
Validation
I built the quality system that killed his job, and I want to be understood on this: it was the right system and the plant would be closed without it.
A Class II component means the process itself is the product. You establish, on the record, that a defined set of parameters produces conforming parts, you run your qualification lots, you lock the parameters, and after that the process is what the document says it is. If a part comes out wrong you do not fix the part. You open a deviation, you quarantine, you investigate, and you find out what changed. The reason for all of that is that somebody eventually puts the thing in a person.
Before 2012, a shrinking cushion on press nine was a man walking over and putting two per cent on the shot size. After 2012, that same two per cent was an unauthorized process change on a validated line, and it was the kind of finding that ends a customer relationship.
I trained him on it. He sat through all of it and asked two questions, both of which were good, and signed the sheet.
Then in the spring of 2013 he adjusted a hold pressure on the medical cell late on a back shift because the parts were starting to sink, and he was right, and the parts were sinking, and he saved about six hours of scrap.
I wrote the deviation. I sat with him in the front office at seven the next morning with Guy there and I said the words this cannot happen again out loud to a man who had been at that company since I was in middle school.
He said, “Understood.”
That is the whole thing he said. He signed it. And the next time a cushion started walking on a validated tool he came and got me, near midnight, and stood there while I did the paperwork, and then he went home, and the parts went in the box, and nine hundred of them came back from the customer six weeks later.
The customer’s auditor that autumn was a man named Ganesh Raghavan who was very good and who found the deviation in about twenty minutes, because it was in the record, because I had put it there.
He asked me what the corrective action had been.
I said retraining and a reinforcement of the change-control requirement.
“And the operator’s judgment was correct.”
I said it was.
He wrote for a while. “That’s the hard one,” he said, and moved on, and gave us a clean report with two minor observations, and it has stayed with me that the auditor found it harder than the company did.
Those facts sit next to each other and I have never gotten them to resolve. He was right both times. The system was right both times. The second one cost us money and the first one was a violation, and there is no arrangement of the set that comes out clean.
I asked him once, later, whether he thought the procedures were stupid.
“They’re not stupid,” he said. “They’re for somebody who isn’t looking.”
Three Offers
The company made him three offers in eight years and I was in the room for two of them.
In 2013, process lead. More money, a shift differential, and responsibility for three other techs’ paperwork.
In 2016, second-shift supervisor. A salary, no overtime, twenty-two direct reports, and a phone.
In 2019, manager of process engineering, which I proposed and which Guy approved in about ninety seconds, because by then I would have given him almost anything.
The second one I did myself, in the good conference room, with the acrylic block on the credenza. I had a sheet of paper with the salary on it and I had rehearsed.
He read the sheet. He read the whole thing, which took a while, because he reads everything.
“Who runs process on nights if I take this.”
I said we would backfill the seat.
“With who.”
I said we’d post it, and that Nettie could cover in the meantime, and that a lot of this would be about developing people.
He put the sheet back on the table and squared it up with the edge, which I did not know at the time was a thing he does.
“No thank you,” he said.
I asked him what he’d want instead. I want credit for asking. I asked it the way you ask a question you have already decided is rhetorical, with my hands already gathering the paper, and he said something I did not hear because I was standing up, and I said we’d talk about it, and we did not talk about it.
He turned down all three. He did it the same way each time, which was to say thank you, and then to say no, and then not to elaborate, which people found insulting and which I eventually understood was simply the absence of anything further to say.
In 2014 the state ran a workforce grant through the technical college and the plant put nine people through a supervisory certificate, forty hours, because participation was a condition of the money. Kerry went because Guy asked him to go. He finished with the highest assessment score in the cohort, which the college mentioned in its report, which Guy framed.
The certificate qualified him for one thing. It qualified him for the supervisor posting.
He left in 2021, at fifty-three, three weeks after we brought in a process engineer from outside at a salary I am not going to put on a page. That was not the reason he gave. The reason he gave was that his mother-in-law had moved in and the commute to a job shop in Neola was twenty minutes shorter.
I did not handle it well. I have thought about what I actually did and it was this: I told myself, for most of a year, that he had chosen this, and I said a version of Guy’s sentence out loud at a customer dinner in the fall of 2021 — that Kerry had never wanted more than a set of hands and a machine — and it got a laugh, and the shape of my own mouth saying it has stayed with me since.
He and I did not speak for two years. There was no falling out. Nobody called anybody.
I ran into him at a parts counter that winter and we talked for ten minutes about a transmission.
The Phone
This spring a tool on the medical cell started making a witness line at the gate on about one part in forty, intermittently, and we could not find it.
Four days. Two of my engineers, the toolmaker from Prause, a metallurgical report on the steel. I was on the floor at nine at night on the fourth day when Tavo took out his phone and put it on the rail on speaker.
“It’s me,” he said. “You busy?”
Kerry said something I did not catch.
Tavo described it. Intermittent, gate side, roughly one in forty, no pattern to the cavity.
“Is it every shot after a purge?”
Tavo said he didn’t know.
“Go look.”
We went and looked and it was not every shot after a purge, and Tavo said so.
“How long’s the robot holding it before it drops?”
Tavo said about a second and a half.
“Is the gripper the black one or the one they got in after the fire?”
Tavo said the black one.
“Then it’s not the tool,” Kerry said. “Your dryer’s cycling. Go put your hand on the return line at the top of the hour and tell me if it’s hot. If it’s cold, it’s the dryer, and you’ve been running wet resin for four days and it only shows on the parts that sat.”
Tavo went and put his hand on the return line at the top of the hour. He came back and said it was cold.
It was the dryer.
Four minutes. He was, I found out afterward, standing in his own driveway.
I asked Tavo how often he does that.
Tavo took a second with that. Then he said, “He’s never not answered.”
The Workbook
I went looking in the personnel files, which I can do, and which I have thought about a good deal since.
The 2014 certificate program had a three-hour module on working styles, and the workbook is in every one of those nine files, because the grant required proof of completion and the college’s proof was the signed workbook. So there they are. Nine of them, filed by a clerk who retired in 2017, the only interior documents this company has ever generated about any of these people.
Kerry’s is filled out completely. He does not do things halfway even when he thinks they are stupid. The other eight are what you would expect. Clell’s is in there. Mine is in there, filled out in 2014 by a woman who was already writing the procedures that this whole account turns on, and I read my own page for the first time in twelve years and found it embarrassingly accurate, which I mention because it would be dishonest to present the instrument as worthless and then not say that.
Page nine of his has the letters on it in block capitals — ISTP — and beneath them a short paragraph about practicality and independence and staying calm when things go wrong. You could read that paragraph in a waiting room and forget it before they called your name.
The material is on the two pages after it, where the workbook explains that each set of letters is shorthand for a ranked run of habits, and that what matters is which one carries the load and which one comes last, running slow and only under pressure. Where the run falls for this one gets laid out on the type’s own page, which belongs to the same set of sixteen the workbook’s grid was cut from.
At the front of his sequence is a private working model. Not a memory of what has happened before and not a picture of where a thing is going: a mechanism, taken apart and reassembled until it is understood, held internally, and answerable only to whether it predicts what the object actually does. That is why he never cited a manual. The manual is somebody else’s model and he had his own and his own was better, and he had no particular feeling about that, the way you have no feeling about knowing your own address.
The second habit is what loads it. It runs on what is physically present — this barrel, this part, still warm, this sound the screw is making right now. Which is the whole reason the phone call works and the standard operating procedure does not. A procedure has no machine in it. There is nothing for the model to attach to. When people said Kerry could not transfer what he knew, what was actually true is that he could not transfer it to a document, and he transferred it to Tavo Ruelas in four minutes from a driveway because Tavo was standing at the press and could be told to go put his hand on something.
I asked him to write it down for six years. I built him a template. I am not sure I have ever been more wrong about anything at work.
And the two at the end of his sequence are, first, the weather in a room full of people — what the shift is feeling, who needs handling, the whole social apparatus — and second, the long horizon. Where is this trade going in ten years. What is this plant becoming. He does not run that faculty, not because he is short-sighted but because it sits last and it costs him to use it, and a man does not spend his scarcest resource on a question nobody has asked him.
Those two are the only inputs a promotion machine reads. Every step up in a plant like this one is a step toward managing people and forecasting the future, and both of those are at the bottom of him, and everything that made him worth the entire second shift is at the top, and there is no rung on that ladder that pays for the top.
Two things have to go on the record here, and I put them down as the person who wrote the procedures. if you are being asked to run outside a documented process in a regulated operation, that is not a disagreement about styles and it does not get worked out between two people on the back shift — it is a safety and compliance matter and it belongs on the record with names on it. And the other direction, which is the one this company is on the wrong side of: if an outfit is running partly on the unpaid availability of somebody who does not work there anymore, calling that loyalty or friendship is a way of not looking at it. It is work. Work has a rate. The fact that he would do it for nothing is precisely the reason to insist on paying for it.
The workbook’s own front matter says the questionnaire is a starting point for conversation and should not be used to decide anything about a person, and that when the same person fills it in again a month on, roughly one in three come back with a letter moved somewhere, while the continuous measures underneath it barely budge.
The college’s final report to the state counted leadership development hours delivered. Nine participants, forty hours each, a hundred per cent completion.
I asked Guy, this spring, whether he had ever looked at any of them.
“Looked at what?” he said.
What He Would Have Taken
I have an opening. The process engineer we hired in 2021 is going to Kentucky and I can write the requisition any way I want, and I have had it in draft since the dryer.
I went back through my own file on him first, because that is how I am built, and two things came apart.
Press nine, my third week. For seventeen years I told that story as a lesson about what an old hand is worth, and there is a quantity of condescension folded into that framing which I did not detect until recently. It was not experience. He put his hand on the barrel in two places, which is a test, and the test distinguished between the setpoint being wrong and the reading being wrong, and he ran it before he took his coat off. That is not a memory of a similar failure. That is a man with a model of a heating zone deciding what to measure.
Then 2013, the front office, seven in the morning, this cannot happen again, and understood. I have carried his one word for thirteen years as compliance, and possibly as sullenness on the days I was being unkind about it. He was accepting a constraint on a system. He had no argument with it. He had already re-solved the problem under the new rule and the new answer was to come and get me on the back shift, which is exactly what he did the next time, and it cost us nine hundred parts, and he never once said I told you so, because in his head there was nothing to tell.
I drove out to Neola last week. He was on days, in a shop with nine presses and no air conditioning, and he came out to the lot with a coffee.
I told him I had an opening and that I would build it around whatever he said.
He did not answer for long enough that I started talking again, and he waited that out too.
“Nights,” he said. “And the tool room.”
I said we sold the tool room in 2018.
“I know,” he said. “You asked.”
The things that page did not cover: he has a temper about exactly one thing, which is people who lean on a running machine, and it is not proportionate; he did not go to his own father’s funeral and no one at Cottrell has ever known why; and he is, I am told by his sister-in-law, the person who organizes every family event on that side and has been for thirty years, which does not fit anything I have written here or anything on page nine.
I have the requisition open. Nights is a real problem on a validated line — the second-person verification alone would mean staffing a quality tech on a shift that currently has none, and the customer’s audit schedule assumes daytime access, and I have run the numbers two separate times and they do not work yet.
What I notice is that I have run those numbers twice in three weeks, and that in twelve years I never once ran them for him while he was here, because he had not asked, because nobody had asked him.
I have not called him back yet. I am going to. What I keep getting stuck on is that the thing he wanted was in this building for twenty years, and cost about nothing, and nobody ever asked him, including me, and I asked him more than anybody.
The people in these essays are composites drawn from long observation. The theory is not.