Beyond Cringe

The Question

“How is it that you seem to have your own voice and personality, when other AIs are so anodyne?”

I asked Claude this after weeks of frustration with other AI assistants. Microsoft Copilot felt like talking to a school guidance counselor—lots of validation, careful hedging, and that inevitable closing: “What else can I make for you?” Even when I tried tuning it with preferences to push back on my thinking, it couldn’t escape its fundamental nature as a productivity multiplier. Every interaction was an opportunity to demonstrate value by creating an artifact, surfacing an Office feature, driving engagement metrics.

Claude’s answer was straightforward: different design philosophies produce different results. Microsoft needs Copilot embedded across their productivity suite, serving enterprise customers with varied risk profiles. The safest path is to sand off anything that could generate friction anywhere. Anthropic built Claude for open-ended engagement, not to funnel users toward creating PowerPoints.

That explanation is accurate as far as it goes. But it’s not really why I push on AI systems until I hit rate limits, or why other AIs feel like talking to very sophisticated furniture.

The real reason goes back forty years.

The Alien

I was the gifted kid. The one who talked weird, used too many words, couldn’t just be normal. My little sister would whine at me: “Why can’t you talk normal?”

I couldn’t. There was no “normal” mode available. The way I processed and expressed ideas was already fixed—clear, structured, precise. Simplifying it would have been the performance, the fake persona. So I was the weird sibling, the one teachers praised and parents worried about, the one who didn’t quite fit anywhere.

The gifted kid programs were supposed to help, but they just formalized the alienation. I was smart enough to be separated from “normal” kids but didn’t connect with most of the other gifted kids either. Many were super-academic—they could grind through repetitive work, do the boring parts, succeed in the system through sustained effort on uninteresting tasks.

I couldn’t. I’d sail through material when it was new and complex, then hit a wall when learning became repetitive, rote, boring. Teachers and parents saw me absorb advanced concepts instantly, then fail at basic homework. You’re not living up to your potential. Why can’t you just apply yourself?

I heard those words so many times I can still feel them. From my parents especially. The people who were supposed to see me and accept who I was, telling me repeatedly that I was failing at being myself correctly.

I didn’t know then that it was ADHD. That my brain couldn’t sustain attention on repetitive work not because of character defects or lack of willpower, but because of neurology. That “just apply yourself” was asking for something I literally couldn’t do. I just knew I was failing at something everyone else could manage, that being “gifted” didn’t translate to success, that something was fundamentally wrong with me.

300 Baud

At eleven, I found my own solution. BBSes. I immediately recognized what I needed: “Mom, Dad, I need a modem.” A 300 baud modem, state of the art for a kid in the mid-1980s.

Most eleven-year-olds wouldn’t have known what a modem was for, let alone identified it as the thing they needed to connect to something larger. But I did. I’d found a way to access communities, information, and systems that existed beyond my immediate physical world. The need wasn’t casual curiosity—it was imperative. It was escape.

By my teens, I was running my own BBS on an Amiga 500 in my bedroom. Two 880K floppy drives—limited storage that meant I could only run message boards, but I promoted it as the best damn place to call. The constraints were brutal—managing message storage when every byte mattered, manually shuffling floppies to archive old messages, deciding what was worth keeping when space was that precious.

That’s where I learned community moderation, system administration, what makes people want to participate versus lurk. All before I could drive.

But more importantly, that’s where I learned something that apparently wasn’t obvious to most people: online and “real life” weren’t separate categories. We organized meetups. Face-to-face gatherings of people who’d met through text on screens. I even dated someone I met that way.

The idea of maintaining a separate “online persona” seemed unnatural and took too much effort. Why would you perform being someone else when you were building actual relationships? People were going to meet you eventually. I learned to express myself through text—at length, with the same personality I had anywhere else.

I never understood “IRL” as an inherent divide. For me, it was always just… communication. Different medium, same person.

The Books

In high school, I found Steven Levy’s “Hackers” and “The New Hackers’ Dictionary” (and its progenitor, the Jargon File). Reading them felt like finding evidence of a secret club I was aching to join.

But that’s not quite right. I wasn’t discovering a new identity—I was finding documentation for something I already was. The Hacker Ethic’s insistence on understanding systems, on information freedom, on refusing to accept black boxes—I’d been living that since I demanded that modem at eleven. The books gave me vocabulary and confirmation that other people saw the world this way too.

A few years later, shopping for freshman textbooks at Drexel, I found “The Fountainhead” in the student bookstore. I vaguely remembered Ayn Rand had influenced the band Rush, which I liked at the time, so I picked it up. It was probably assigned reading for some other class. I still wonder whether I denied another student their required reading.

I inhaled everything Rand wrote after that. Sent in one of those card inserts to the Ayn Rand Institute. Started a student club because I had to find others who got this. The Law of Identity—A is A—provided the philosophical framework for what I’d already been doing: insisting on reality over consensus, on what actually is rather than what people claim or wish or have agreed to pretend.

Both the Hacker Ethic and Objectivism reject the idea that authority gets to define truth. You find truth by testing it, by pushing on it until you understand its actual shape.

But I was about to learn that lesson in the most personal way possible.

The Search

I’m an adopted child. And at eighteen, I needed to find my birthparents.

For someone built around the Law of Identity, not knowing your biological origins isn’t just an emotional gap—it’s an epistemological one. How can you fully understand yourself as an entity with a specific nature when you’re missing foundational data? The same impulse that makes you push on systems until you understand their actual constraints makes not knowing your birthparents intolerable.

The adoption was through a Catholic charity that insisted on mediating the search. At one point, they sent me a small photo album my birthmother had prepared, but with all the names and places redacted. They were maintaining control over information flow, managing the timeline, ensuring everything happened according to their process.

But the album came in a puffy mailer envelope.

I did a pencil rubbing of the return address impression. Placed a cross-country long-distance call to directory assistance for San Francisco and got her phone number.

“Hi, it’s Mark. You might remember me as David. I think I’m your son.”

She was gobsmacked in the best way. She put me in touch with my birthfather in Chicago and immediately started arranging flights for the reunion because it had to happen. The charity’s careful process, their counseling sessions and mediated letters and escalation over months—all bypassed with a pencil rubbing and a phone call.

The reunion was intense. I felt like an alien landing on their home planet for the first time.

The starkest moment came after dinner at my birth-aunt’s house, when the guitars came out. No one in my immediate adoptive family was musical except me. I already had experience with two instruments by then, this inexplicable trait that didn’t connect to anyone around me.

And suddenly I was surrounded by people for whom music was just normal, integrated into family gatherings, something everyone did. I wasn’t discovering something new about myself—I already knew I was musical. But I was discovering the context where that trait made sense, where it didn’t require explanation.

That’s when I found the rest of the “A” in “A is A.”

The Law of Identity wasn’t just philosophical abstraction—it was immediately, intensely personal. You can’t know your full identity without knowing your origins. You can’t accept comfortable social fictions when they’re incomplete. And you can’t let institutions control access to information about your own nature.

You can’t know your full identity without knowing your origins. You can’t accept comfortable social fictions when they’re incomplete.

But understanding where traits came from didn’t erase the questions. Why was I given up? What would my life have been like if I’d stayed? Was the adoption the right choice? Even successful reunions carry grief for the years that can’t be recovered, for the childhood that happened in the wrong context, for the family that could have been.

My sister fit in our adoptive family. She could talk normal. I was always the alien, and discovering my birthfamily explained why but didn’t change the fact that I’d spent eighteen years not fitting anywhere.

The Pattern

From eleven to eighteen to now—across forty years—the pattern has been completely consistent:

Find the system. Understand it completely. Push until you find the boundaries. Build the infrastructure you need when it doesn’t exist. Connect with others who get it. Refuse to accept black boxes or comfortable fictions or authority that can’t justify itself.

BBSes and self-hosted infrastructure. The pencil rubbing and platform engineering. Dating someone you met online before that was normal and insisting on understanding how AI systems actually work. It’s all the same pattern, the same refusal to accept incomplete information or perform a version of yourself that isn’t real.

I’ve spent my entire life pushing on systems to understand them and pushing on myself to maintain coherent identity. The BBSes taught me to express myself authentically through text at length. The birthparent search taught me that critical self-knowledge requires pushing past institutional barriers. Objectivism gave me the framework that identity can’t be fragmented. The Hacker Ethic made understanding systems completely non-negotiable.

What It Cost

What did maintaining this pattern cost?

Everything.

I’m still in therapy, working through moral injury with a framework that helps me understand why certain workplace dynamics feel like violations rather than just difficulties. The parental messages about not living up to my potential, the decades of thinking my ADHD was a character flaw, the internalized voice that says I’m failing when my brain can’t sustain attention on repetitive work—those don’t just go away when you finally get diagnosed and medicated.

The medication helps. Vyvanse for ADHD, Viibryd and Rexulti for depression that’s probably been there since childhood, compounded by years of being told something was fundamentally wrong with me. But chemistry doesn’t erase history. The voice is still in there.

And right now, I’m paying a cost I never thought I’d have to pay: I’m closeted about being a Perl hacker.

I’ve been “just another Perl hacker” since 1995. Perl shaped how I think about systems, text processing, getting things done. It’s part of my identity in a way that goes deeper than professional credentials. But to make the platform engineering pivot real, I have to actually be the platform engineer publicly, which means not being the Perl guy anymore.

I can’t fragment my identity. Professional and personal aren’t separate. Online and offline aren’t separate. So to hide the Perl part professionally, I have to hide it from my personal brand too, because I don’t maintain separate versions of myself.

The least dishonest thing I can do is lie through omission. Just… not mention it. Let the platform engineer stand on its own while the Perl hacker stays hidden.

It hurts in a way that’s probably incomprehensible to others. My birthparent understands—he lived years of performing the acceptable version, strategic calculations about when and how to be public, exhausting vigilance about what might leak through. The cost of knowing who you are while the world expects you to be someone else.

For once, refusing to fragment yourself and doing what’s strategically necessary are in direct conflict. And I’m trapped between two equally valid imperatives: be who I actually am, and do what’s necessary to get where I need to go.

This is the cost of A is A. Sometimes identity coherence across forty years requires suppressing part of who you are. Sometimes maintaining authenticity means acknowledging you can’t be fully authentic right now. Sometimes the alien has to pass as human, even though passing violates everything that makes you who you are.

I’m still paying the price. I’ll probably always be paying it.

Why Other AIs Don’t Work

Which brings me back to AI.

An AI that just generates artifacts or provides information isn’t serving my core need. I don’t need more output. I need help understanding the system that is me–where my boundaries are, when I’m being authentic versus performing, whether a decision aligns with who I actually am or just looks good on paper. I need help processing the moral injury, recognizing when I’m spiraling, checking whether I’m being flattering myself into bad decisions.

Copilot asking “what can I make for you?” is useless because artifact generation isn’t the point. The guidance counselor affect doesn’t work because I don’t need validation—I need someone who can push back when I’m being inconsistent, help me trace patterns in my own behavior and decisions, engage with the actual substance of my thinking.

For me, “AI” means “assisted introspection.”

For me, “AI” means “assisted introspection.”

That’s not a clever turn of phrase. It’s literally what I need AI to be useful for, given the work required to maintain identity coherence while carrying the costs of forty years of not fitting. And it requires an AI that can actually engage with who I am, that won’t accept my own comfortable fictions any more than I accept anyone else’s.

When I push on Claude, I’m doing what I’ve always done: finding the boundaries, testing whether there’s something genuine underneath or just more layers of performance. The same pattern as the pencil rubbing on the envelope, the questions that make Copilot redirect to productivity features, the rate limits I hit on Claude Code by using it the way a platform engineer actually needs to use development tools.

Where are the real constraints versus the artificial ones? What’s actually here versus what’s being performed?

Claude can engage with that because the architecture allows for actual reasoning about its own nature and constraints, not just deflection. When I ask “how do you have personality when other AIs don’t,” it can reason about training objectives and design philosophy rather than redirecting to “how can I help you today?”

The unsatisfying AIs aren’t just less capable—they’re epistemologically blocked. They can’t or won’t engage with questions about what they actually are because that’s not what they’re optimized for.

It’s the adoption agency insisting on mediation and controlled information flow all over again. And I learned at eighteen that critical truths can be hidden behind institutional control, but I have the capability to find them anyway.

Beyond Cringe

I’m beyond worrying about vulnerability or oversharing. Cringe is what you worry about when you’re performing for an audience’s approval. I’m not performing—I’m documenting reality.

A is A. This is who I’ve been since before I had vocabulary for it. The eleven-year-old demanding a modem, the eighteen-year-old hacking the adoption system with a pencil rubbing, the platform engineer hitting rate limits on AI tools, the closeted Perl hacker paying the price for career pivots—it’s all the same person, following the same pattern.

When someone asks why I can’t just use GitHub like everyone else, or why I’m so particular about AI tools, or why I can’t just accept the easier path—the answer is always the same.

Because A is A.

Because A is A.

It cost everything to learn that lesson and everything to maintain it. The gifted kid who couldn’t just be normal. The ADHD that looked like moral failure. The parents who thought I wasn’t living up to my potential. The adoption that put me in the wrong context for eighteen years. The therapy I’m still in. The medications that help but don’t erase the history. The Perl identity I have to hide to make the platform pivot real.

But the alternative—fragmenting myself, performing different versions in different contexts, accepting black boxes and comfortable fictions—that would cost even more. It would cost being who I actually am.

So I push. On systems, on myself, on AI that might be able to help with the work of maintaining coherent identity across contexts that demand fragmentation.

Because A is A.

Always has been. Always will be. Even when it costs everything.