Interrupt-driven support work breaks your strongest people first

An engineer I know spent six weeks failing to start a document he had promised to co-write. There were weekly calls about it. He had personally been walked through the exact method the document was supposed to explain. On the day it was due, he turned up with a placeholder page and nothing else on it.

His explanation is the most useful sentence about work I have read in years. "My brain only recognizes two times, now and not now. That was a problem for future me."

He was not lying about intending to write it. He intended to write it constantly, for six weeks. What he had was a brain capable of holding genuine intention and genuine paralysis at exactly the same time.

Once you have seen that clearly, you start seeing it in your ticket queue.

Why an ambiguous ticket costs more than a dead mail server

"The mail server is down" is a lovely problem to have. Binary. Clearly broken, then clearly fixed. There is a specific moment when it comes back, and that moment does more work than anyone gives it credit for, because completion is chemical. Finishing something releases dopamine, and that release is a large part of what makes the next start possible.

"Outlook has been weird since Tuesday" offers no such moment.

Open-ended, ambiguous problems never deliver the done marker that closes the file. So the brain keeps running the case with nothing to end it. Three days later, in the shower, it is still running the case.

Interrupt-driven support work is a queue made almost entirely of that. Which makes it uniquely difficult, and hardest of all for the people whose brains most need a clean finish to function.

Here is my position, and I hold it firmly. Better ticket structure is an accessibility feature. Teams sell structured intake internally as a reporting improvement or an automation prerequisite. Both are true, and both undersell it. A ticket that arrives with a category, a scope, and a stated definition of finished is materially easier to work for a large fraction of your staff.

Divergent thinking is a technical asset with an invoice attached

Two modes worth naming. Convergent thinking narrows toward the one right answer. Divergent thinking generates many possibilities at once.

Good troubleshooting is mostly the second one. Standing in front of a system that is broken in an unfamiliar way and rapidly producing ten plausible causes is the actual skill, and the people who do it naturally are the ones you page at 3am.

The bill arrives with it. A brain that generates ten solutions generates ten failure modes simultaneously, at equal emotional volume, and nothing shows up pre-labeled by likelihood. That is why beginning a task can feel like standing in wet cement while your head sprints.

Your best troubleshooter and your most stuck person are frequently the same human being on two different days.

I will say the next part once, plainly, and then move on. This is not laziness. What reads as resistance or procrastination is the nervous system matching a task to a past threat and freezing it in place. Willpower does not override that, and shame overrides it only in the sense that a fire alarm gets people out of a building.

Which brings up the burnout mechanism specific to this. Activation for a lot of people requires interest, challenge, novelty, urgency, or passion. When none of those is present in a task, they get manufactured, and the cheapest thing to manufacture is dread. Self-loathing and last-minute panic function as a reliable starter motor.

They also work, which is the trap. I ran most of a decade that way and called it performing well under pressure. It is the least sustainable engine I have ever installed in myself, and it goes a long way toward explaining why this group burns out at the rates it does.

A short detour through numbers

Roughly 4 to 5% of adults have ADHD. About 2% are autistic. In any given year, 18% have a diagnosed anxiety disorder.

Statistically that puts an estimated one in five to one in six people in any general room in one of those categories, and probably higher on a technical team, because a job that rewards hyperfocus and pattern recognition attracts the brains that do those things involuntarily.

Two footnotes that matter operationally. ADHD and autism are significantly underdiagnosed in women, because the diagnostic criteria were originally built by studying boys, so if your team's mental model of ADHD is a restless nine-year-old, you will not recognize the senior engineer who has been compensating for two decades. And dopamine and norepinephrine reportedly run up to 30% lower in ADHD brains in some research, which is a physiological explanation for a thing usually discussed as an attitude.

I am not asking you to screen anybody. I am pointing out that you already employ these people, and your support rota was designed as though you did not.

Working with the brain you have

Getting started, when starting is the hard part

Move a muscle, change a thought. Physical movement before a task, a lap of the building or thirty seconds of jumping jacks, increases prefrontal blood flow and gets the relevant chemistry moving. It sounds unserious. It reliably breaks a freeze, and it costs ninety seconds.

Contain an infinite problem with three lists. What I know, facts only. What I don't know. What's not my job. Then time-box 20 minutes before escalating.

That technique is worth stealing for your whole team regardless of anyone's wiring, because it converts an unbounded ticket into a bounded one in about four minutes. The third list is the one that earns its place, since ambiguous tickets annex other teams' territory and nobody notices until a week has gone.

Use a Pomodoro box as containment rather than productivity. One task, one timer, 15 to 30 minutes. No quality judgment. No scope expansion. No self-criticism, and no starting over. Intrusive "what if I am wrong about this" thoughts get written on a side notepad and revisited when the timer ends, which keeps them from derailing the block.

Body doubling. Work alongside another person who does not need to know what you are doing or help you do it. Presence alone lowers the barrier to starting. Effectiveness here is intensely individual: one admin needs a body in the same physical room, while his housemate thrives in an all-day virtual group of strangers doing unrelated work.

Use an AI assistant as an external convergence tool. Dump the unstructured mess of what you are thinking into it and ask it to organize the mess, or ask it for a beginning and an end so you can fill in the middle. That is the single most valuable thing these tools do for me, it has nothing to do with code generation, and nobody markets it.

Shaping the day around actual capacity

Energy-aware scheduling. Stack deep work in your highest-energy window. Route meetings and low-stakes tasks to the afternoon dip. When the brain turns to soup around four, stop completely rather than pushing through and producing work you will redo tomorrow. If a second wind arrives at seven or eight, use it instead of fighting it.

A full-screen meeting reminder that cannot be waved away. A five-minute warning feels like an eternity to a time-blind brain, so it gets dismissed and then the meeting has started.

Only handle it once. Process a request, a form, a decision completely on first contact instead of setting it down for a later that is structurally never going to arrive.

Don't put it down, put it away. Finish the last micro-step. The dish goes in the dishwasher, the ticket note gets written now, the config gets committed rather than left dirty.

Decision budgeting. Treat each day's decision-making capacity as finite and depletable, and schedule high-stakes decisions early. This is the same reasoning behind people who wear identical clothes daily, applied to change approvals.

Two email windows. Batch email into two fixed slots instead of monitoring it live, with a hard rule: under five minutes gets done immediately, everything else gets logged and left.

Two accommodations round this out, and both are free. A personal "read me" document, a page explaining how you actually work best, shared proactively with anyone who collaborates with you. And calendar-blocked focus time that nobody can book over without asking first.

Trust runs on evidence

The engineer with the placeholder page had visibly dropped a commitment in front of a collaborator who was depending on him. What fixed it is worth being precise about, because the sentimental version is wrong.

He did not fix it by apologizing more thoroughly, and he did not fix it by believing in himself. His own framing: belief isn't transferable, evidence is. Trust came back through a string of small, boring, kept promises over months, none of which made a good story.

The corollary is equally unsentimental. A job or a task that fits you badly and ends badly is data, not a verdict. Worth saying because the move from "I did something wrong" to "I am wrong" happens very fast in some brains, and a single missed deadline becomes a referendum on a person's worth by Thursday.

Five ways technical people talk themselves out of asking

You will recognize all of these, possibly from the inside.

The Perfectionist, for whom a 99% success rate reads as failure. The Expert, who will not ask a question until they already know everything, which is a fascinating position to defend. The Soloist, for whom asking for help feels like cheating. The Natural Genius, for whom the presence of struggle is itself proof of inadequacy. And the Superhuman, juggling admin and engineering and support and parenting, such that dropping any one ball convicts them in all of them.

The operational cost is real. The Expert and the Soloist add hours to your mean time to resolution, and they do it without ever appearing in a report.

The useful research reframe is that imposter feeling behaves like a state triggered by context change (a new role, a new boss, a reorg, sustained ambiguity) rather than a fixed personality trait. That makes it temporary and, better, predictable. You can anticipate it during onboarding and restructures instead of watching good people go silent. For scale, in a 2018 anonymous poll of more than 10,000 tech workers, 58% reported imposter syndrome.

What a manager can change in the rota this month

1. Put shape on intake. Structured buttons for your top request types. Ambiguity is a cost you can delete before a ticket exists.
2. Write a definition of done for each ticket type. Whoever accepts an open-ended ticket writes the closure condition first. I argued against this for years as bureaucratic overhead and I was wrong about it.
3. Publish a 20-minute escalation rule. Time-box, then hand off, with no judgment attached to handing off. It removes the sentence "I should have been able to solve this."
4. Split the rota. One person on interrupts, everybody else protected, rotated weekly. Nobody does interrupt duty and project work on the same day.
5. End the interrupt shift with a real handoff. Someone has to be able to close the day.
6. Ask people to write a read-me, then actually read them. Filing them counts as not reading them.
7. Stop praising sub-minute chat replies. You get what you applaud.

The invoice you cannot see

An environment that is interrupt-heavy, ambiguity-heavy, and bad at telling anyone when they are finished is hostile to a large fraction of the people who are best at this work. Every fix above is cheap, and most of them are free.

We built ramps at building entrances for people using wheelchairs, then noticed that everyone arriving with a suitcase or a delivery trolley uses them too. Accessibility work has a habit of turning out to be general infrastructure.

A ticket that says what it wants, what has been tried, and what finished looks like is a curb cut. Almost nobody thinks of it in those terms.

Nobody needs to build it out of generosity. You are already paying for its absence, in hours, and the invoice is not itemized anywhere you can find it.