People assume a special interest is just a hobby turned up too loud. It isn't. It's the one environment where my actions reliably produce consequences I can see, understand, and stand behind. Mine maps onto software, and that single fact explains more about my career than any decision I ever consciously made. This piece is also published on dev.to.
What a special interest actually is
The casual version of the term is something like "an autistic person who is really into trains." Technically true, completely missing the mechanism.
What's actually happening is that a particular domain becomes disproportionately capable of activating things that are otherwise hard for me to reach: motivation, sustained attention, emotional regulation, a sense of competence, a sense of coherence. Not because the subject is objectively more important than everything else in my life, but because my nervous system responds to it in a way it doesn't respond to almost anything else.
That's a very different claim from "I enjoy this." Plenty of things are enjoyable. Very few things reliably switch me on.
There's a name for this, and it isn't mine. Monotropism, developed by the autistic researchers Dinah Murray, Mike Lesser, and Wenn Lawson, treats attention as quantitatively limited: it pools into a small number of intense channels instead of spreading thinly across many. The image in the original work is a torch beam rather than diffused light, and the state it produces is an attention tunnel. Once you model attention that way, most of what follows here stops being a personality quirk and starts being a predictable consequence of how the attention is allocated.
So when someone asks why I'll work on a problem at 2am while physically exhausted, the honest answer isn't discipline and it isn't recklessness. It's that this is one of the few places where my brain will actually start. Telling me to apply that energy to something more sensible assumes the energy is fungible. It isn't. It's tunnel-bound, and the tunnel picked me rather than the other way around.
Same wiring, different surface
I read an account once by an autistic person describing why she spent enormous amounts of time in video games. The word she reached for wasn't "addictive." It was control. Agency. Autonomy. A world where her actions had consequences, the rules were legible, competence was achievable, progress was visible, and mistakes were recoverable. Everything her actual environment, at home and in institutions, did not reliably offer.
That turns out not to be an idiosyncratic account. It's close to a consensus finding in the qualitative research on autistic gamers, to the point where one study in the British Journal of Developmental Psychology takes its title straight from a participant: "I can actually do it without any help or someone watching over me all the time and giving me constant instruction". A large survey of autistic gamers puts stress relief and a sense of achievement at the top of the list of reasons, and work on escapism in autistic adults frames it as a self-regulation strategy rather than a pathology.
I recognized my own relationship with software in all of it immediately, because structurally it's the same thing.
A game gives you a bounded world that responds to you. Software gives you almost exactly the same properties, except the thing you produce exists outside the world you were playing in. You think of something, build it, hit a wall, learn, solve it, see it work, publish it, and someone can use it.
Same wiring. Different surface. The pull was never the subject matter, it was the structure: a domain where my actions count.
What my brain wants, and what software supplies
Here's the mapping, as concretely as I can state it.
My brain wants legible rules. Software is a system where the rules, however complex, are ultimately knowable. When something breaks, there is a reason. That reason can be found. Nothing about it depends on correctly reading a room.
My brain wants fast, honest feedback. Software tells you unambiguously and quickly whether what you did worked. The test passes or it doesn't. The build succeeds or it doesn't. No ambiguity, no social interpretation layer, no waiting weeks to find out whether you misjudged something.
My brain wants depth without a ceiling. Most interests eventually exhaust what's available to learn about them. Software does not. However far down I want to go, there is always another layer, and going deeper is rewarded rather than treated as excessive. A domain that never says "that's enough detail now" is an unusually comfortable place to put a tunnel.
My brain wants self-directed scope. I choose the problem, the approach, the architecture, the stopping point. Nobody assigns me the next thing. That matters enormously, because externally imposed structure is exactly what my executive function handles worst, and self-chosen structure is what it handles best.
My brain wants recoverable mistakes. Version control means almost nothing I do is irreversible. Being able to try something badly, see what happens, and roll it back is what makes exploration possible for someone who cannot reliably plan the whole route in advance.
My brain wants something that persists. At the end, an artifact exists. A repository, a release, a service, a skill I didn't have last week.
Put those six together and you get a description of an environment that could have been designed for how I work. It wasn't, obviously. It just happens to be the shape it is, and my attention found it and stayed.
The autonomy I don't get anywhere else
This is the part that's hardest to explain to people who assume I'm describing a preference.
I actually have a lot of control over my physical environment. I live alone. I arrange my space exactly how I need it. I eat a small, consistent set of foods that work. I go outside at night, when it's quiet and dim and there are no people, because the daytime version of that same errand costs more than my nervous system can consistently pay.
That's real autonomy and it genuinely helps. But controlling my environment is not the same as being able to participate in the world.
Every one of those adaptations still reads as "I can do this, but only under these conditions." I didn't choose 2am because it's my favorite hour to be alive. I chose it because it's when the cost drops to something payable. So even inside a life I've deliberately optimized around myself, I'm permanently aware of the set of things that remain out of reach, and no amount of rearranging my flat changes that.
Software doesn't work that way. In software the barrier between what I intend and what I can execute is thin and getting thinner, because the entire discipline is built out of tools whose whole purpose is to absorb load. I don't have to hold everything in working memory. I don't have to remember every API. I don't have to independently sustain a research thread across days, which is precisely the kind of continuous perseverance I'm worst at. The tooling carries the part I can't, and I do the part I can: the judgment, the design, the decisions, the responsibility for what ships.
That's the difference. Everywhere else I'm adapting the world to the boundaries of my nervous system. Here I'm using tools that let me cross them.
So when the interest activates, it isn't activating in a vacuum. It's activating the only reliable supply of something scarce in the rest of my life: I decided this, I did this, it worked, and it's mine.
Why the artifact matters so much
There's an extra layer on top of the loop, and I underrated it for years.
The work leaves evidence.
Each finished thing is a durable record that says: I learned this, I solved this, I built this, I can stand behind this. That matters disproportionately to someone whose memory doesn't reliably preserve their own history, and whose sense of capability fluctuates with energy levels that swing hard. On a bad day I can't feel competent. But I can look at something I built and read the receipt.
That's also why the interest didn't stay private. Publishing the work turns the internal loop into something with an external footprint: a portfolio, a service, a skill someone else can benefit from. A special interest that produces artifacts stops looking like a fixation and starts looking like a career, which is a fairly arbitrary distinction, but it's the one the world actually uses.
Why leaving is hard, and why "addictive" is the wrong word
I want to be honest about the cost, because the same mechanism that makes this good for me can consume everything I have.
Addiction imports a set of claims that don't fit here. What's actually happening is simpler: an activity becomes extraordinarily reinforcing when it supplies something that's scarce everywhere else in your life.
Monotropism gives a cleaner account of the exit cost than willpower ever did. Leaving an attention tunnel isn't like closing a tab, because you aren't next to the thing, you're inside it. The original paper puts it starkly: for someone in an attention tunnel, unanticipated change is briefly catastrophic. That's not a metaphor for mild annoyance. It's a description of why being pulled out of something feels the way it does.
There's a companion term for the other half of it. Autistic inertia covers both the difficulty of starting and the difficulty of stopping, the latter sometimes called termination resistance, and autistic people describe it as among the most disabling parts of being autistic while simultaneously describing deep absorption as one of the most restorative states available to them. Both things are true at once, which is exactly why this is hard to talk about honestly.
It's also worth separating two things that often get merged. Hyperfocus describes an unusually intense stretch of attention, hours or maybe days. It doesn't explain why someone returns to the same domain for years. A special interest is the long-horizon pattern, and it can become a genuine organizing feature of a life and an identity. The two interact, but they're not the same phenomenon, and collapsing them makes the whole thing sound more pathological and less structural than it is.
The consequence is that stopping isn't experienced as "I'm done for today." It lands closer to "I'm leaving the place where my actions count." That's a far stronger pull than enjoyment alone, and it's why the physical exhaustion always arrives long before the motivation does. There's no backpressure in the loop. Nothing in it signals that the upstream resource is depleted.
The only off-ramp I've found that actually works is making stopping cheap. Before I close everything, I leave myself enough state to resume without reconstructing the whole mental picture: the symptom, the current hypothesis, the last thing I changed, the next concrete step. That note isn't a reminder that the task exists. It's proof the thread hasn't been lost, which is the specific fear that keeps me working when I should be asleep. Once my brain accepts that the tunnel will still be there tomorrow, it will let me put it down.
The interest is not the problem
If there's a failure mode here, it isn't "I care about software too much."
It's that a single domain is carrying the entire burden of proving that my actions matter. That's a load-balancing problem, and the fix is not to take away the thing that works. Autistic accounts consistently point the other way: special interests function as protection against burnout precisely because they restore a sense of control. The fix is to widen the set of places where agency is available, even in small, unimpressive ways, so the interest gets to be one strong source instead of the only one.
What I'd resist entirely is the framing where a special interest is something to be moderated on principle, or grown out of. Mine is where my motivation lives. It's where I'm most competent, most regulated, and most recognizably myself. It produced a skill set, a body of work, and an income.
So if one idea survives this whole essay, let it be this:
I didn't build a career and then discover I enjoyed it. My attention found the one environment where it could reliably act, went all the way in, and the world happened to call the result a profession.
That's not a hobby that got out of hand. That's the only place my agency was ever fully available, and I'd be foolish to treat it as anything less.