The Epistemics of Being a Mudblood: Stress Testing intellectual isolation.

Vaguely connected to AI; mostly, for once, explaining who’s really behind these writings.

I started Stress Testing Reality with a premise: writing our way to a reality we can trust. Most of the time, that means writing about AI Safety for AI Governance: using insights from AI Safety research to stress-test the assumptions behind laws and policies, and translating technical constraints into enforceable governance frameworks. That remains the core of this newsletter.

But this post is different.

It’s the end of summer, and I’ve written less than usual. Not because I’ve run out of things to say, but because I’ve been in a cycle of building, reflecting, and recalibrating.

It’s been a season of progress and accomplishments that make me feel extremely grateful: designing a syllabus to train governance professionals that is under consideration by established institutions, joining the European Network for AI Safety, being accepted into the Fall cohort at the Center for AI & Digital Policy, and having academic papers accepted (more on this soon!).

On paper, I’m in a very good place. I find that we do not often explain the “why” behind apparent success, and that deprives people from valuable learning.

Many readers here are early-stage professionals, or seasoned ones in the middle of career and personal transitions. I know what it’s like to scroll LinkedIn or Substack and wonder: how long until I’m also there? I used to read posts like this and try to reverse-engineer the secret formula. (Spoiler: there isn’t one).

What there is, instead, is a stubbornness. A willingness to keep thinking too much in environments that tell you not to. A refusal to apologise for outsider status.

For me, the best metaphor has always been Harry Potter’s “mudblood”: someone with magic but without the right lineage, forever reminded that they aren’t built the same. That’s how it felt growing up, and honestly, how it still feels in many ways.

But “not being built the same” has very little to do with the things that matter, and that realisation is something I wish I saw more seasoned professionals write about. So, I am doing it now :).

This essay is my reflection on that, more personal and philosophical than usual.

Because if there’s one lesson I want to leave behind this summer, it’s that what carried me here wasn’t pedigree. It was learning how to carry curiosity through environments that didn’t want it. Other people’s reflections once gave me courage to keep going.

This is mine, in case it does the same for you.

---

The overthinking tax is worth it.

As a child I was unmistakably neurodivergent- ADHD, hyperactive to a fault. My mother’s relief came when I learned to read: books kept me still and silent for hours, and I was given as many as I wanted. Somehow, I became both the quiet kid lost in a corner and the loud kid disrupting the class. As expected, reading a lot was tolerated, while distracting others was punished. What was mocked most consistently was “thinking too much.” The phrase I heard over and over was “no te rayes” (don’t overthink it), and I clearly still don’t know how to comply!

Looking back, I really believe that phrase is the perfect summary of an environment where deep thinking isn’t a must. I am sure this applies accross locations, and I am only speaking from my own personal experience here: Small villages tend to be peaceful and accepting, but traditionally there are no debate clubs in schools, or that sort of extra curricular (save from private schools).

The mistake I made was assuming this was location-bound and temporary. Surely adults would live differently, and epistemic effort would finally matter. But environments that punish depth don’t suddenly flip at eighteen. They teach you that a habit of “meta-thinking” is optional at best, inconvenient at worst.

So what do you do if you’re wired for deep thinking in an environment that neither demands it nor rewards it? You can give up and follow the script. You can retreat into tiny circles of people who think like you, and hide from discomfort.

Or you can try to inhabit the in-between: committing to carrying curiosity into spaces that tolerate it just enough to survive, but not enough to celebrate it.

That “in-between” identity is uncomfortable, much like being the mudblood in the situation. But it’s also a training ground.

Learning to navigate environments that undervalue curiosity (or the desire to know better) is, I think, what makes it possible to survive in governance, academia, or corporate life without losing the part of you that still insists on asking why.

Intellectual Isolation and the Myth of Meritocracy

As a teenager I assumed that once people became “knowledge workers,” meta-thinking would finally matter. The reality was sobering: many adults in complex roles don’t want to spend more time than necessary improving how to think about work, or their free time on philosophy or AI. They want to get on with it, and decompress afterwards. Academic circles, meanwhile, aren’t magically accessible. They’re gated by social hurdles like networks, money, and… well, money.

For me, this meant years of working 40–50 hours a week while studying law. Not glamorous, but necessary. And those environments shaped me in ways full-time academia probably wouldn’t have.

It is very true that Customer Service jobs expose you to humanity unfiltered: sometimes at its worst, but also in all its variety. I learned empathy, negotiation, resilience- skills I now use constantly when translating across law or technology.

In all honesty? The harder lesson was about meritocracy, and it is something I still wrestle with.

I waited for years expecting a “click”- that moment when intellectual effort would finally be recognised above everything else. Instead, I discovered that hiring and promotion often reduce to who likes who, “vibes”, offfice power struggles or other less-than-transparent criterion.

I bet many of you kind of naturally expected this, but I genuinely thought this eventually “goes away” (lol!).

As we know, corporate life isn’t so different from high school: who gets invited, who passes the cultural fit test, who’s easy to sit next to. For someone who can’t quit asking ‘why,’ that disconnect quickly becomes isolating.

The question is what to do with it.

If there’s a takeaway here, it’s this: don’t waste years waiting for your environment to change its settings. Most won’t. Instead, treat those settings as constraints to navigate.

Because, sometimes, the very environments that undervalue epistemic effort are the ones that give you the perspective to apply it more powerfully later.

On Refusing to Be Determined by Circumstances

By my mid-twenties, I finally had more freedom- better finances, time to travel, a chance to re-enter academia. And then the pandemic hit.

Long story short, I ended up back in my hometown.

It was disorienting. I wanted to believe my earlier judgments about the place had been unfair. They weren’t: I had moved from an environment that, while expensive and harsh, at least rewarded effort (grants for working students, academic incentives) to one that actively penalised it (higher tuition if you had a job). My world-model, that effort wasn’t always valued, was confirmed. I updated my beliefs, but in unproductive ways: resentment, negativity, even a creeping sense of futility.

Paradoxically, those years of frustration led to some of my best outputs. Being ground down by sleep deprivation and petty obstacles gave me less room for condescension and more room for logic and compassion. It clicked for me how easy it is to feel isolated, and why some people would rather turn to AI companionship than to human company. My “stuck-up academic self” might once have mocked that choice; the version of me that lived through it could only make sense of it with empathy and intellectual curiosity.

And from that ground, new doors opened: a master’s in privacy and data protection, work in human rights, roles in tech, and eventually a path into AI governance and safety. I doubled down on study, pursued challenges I actually wanted, cut ties that didn’t move me closer to my goals. And in the middle of this, I discovered something life-changing: entire communities of people spend their days thinking about long-term AI risks.

For the first time, I wasn’t “punished” for what I once thought was overthinking, or carrying curiosity alone.

Epistemics of “adulting better” I wish I had earlier.

Most of what I’ve written so far has been about carrying curiosity through environments that didn’t reward it. But I don’t want to leave this as just a private reflection. Here are four lessons that may be useful if you ever find yourself in the same situation.

1. There is always a mold you’re expected to fit. Find one shaped like you.
Every system has boxes: cultural fit, grant criteria, vibes checks. If you don’t fit, you’re out. Many people respond by trying to patch their weaknesses. I’ve found the opposite works: ignore the places that punish your nature and double down where your strengths matter. Richard Feynman built his reputation on what many of his peers thought of as “trivial” obsessions. In AI Safety, Jacob Steinhardt and Paul Christiano walked away from safe tracks to pursue research that looked strange until it became central.

You don’t sand yourself into shape; you find or create the mold where your edges are an asset.

2. Optimise for interest early, not money later.
I wasted years waiting for financial security to align with meaningful work. In practice, the order has to be reversed: build your finances around your interests, not your interests around your finances. That doesn’t mean recklessness. It means recognising how fast five years vanish if you defer. I know how patronising this sounds, I know how difficult it is in practice (I really know). And I wouldn’t be saying this if I didn’t have to go through the hurdles I did. Scholarships and grants take effort to find, but they exist. Networking into the right communities is slow, but it compounds. Don’t assume that “later” will magically arrive if you just keep “working hard” but narrowly.

3. Don’t lock yourself into one source of growth.
It’s tempting to stay in one environment once you find people who “get” you: academia, policy, comfortable circles. But monocultures shrink your perspective if they’re your only source of intellectual stimulation and social connection. Some of the best lessons I’ve learned came from people with nothing in common with me: retail colleagues, engineers who cared only about shipping, lawyers uninterested in AI. You don’t need overlap to learn. If anything, discomfort forces interdisciplinarity- the thing I rely on most now.

4. Stop apologising for your mudbloodness.
“Mudblood,” in Harry Potter, is the insult for someone with magic but the wrong lineage. I use it here for anyone who feels like an outsider: immigrants, first-gens, people from non-prestigious backgrounds. The reflex is to over-perform, say yes to everything, prove you’re “grateful” to be allowed in the room. But that erases the very strategies you developed to survive constraints others never faced. Sometimes you will be the only mudblood in the room. That’s not a deficit. It’s evidence you carved your own way in.

Rationalists like to talk about map vs. territory or being the kind of agent who updates. My life so far has followed this pattern: the territory does not reward deep thinking by default. Most environments punish it. But if you persist, if you build or seek out the rare pockets where it is valued, the payoff is extraordinary.

That’s what Stress Testing Reality is for me: a place where I don’t have to apologise for overthinking, where I can use the lens of law, policy, and alignment to stress-test not just AI or regulation… but myself.

Conclusion

If there’s one thing I’ve learned, it’s that epistemic isolation is a self-fulfilling loop: you don’t see others who care about the things you care about, so you assume they don’t exist, so you stop signalling those interests- making it even harder to find them. Breaking that loop requires brute honesty about what actually matters to you, even if it comes with social costs in the short run.

“Mudblood forever” works for me because it reframes what once felt like a deficit (being from outside the “right” place, network, or pedigree) into a kind of stubborn asset. You build your own epistemic antibodies. You get sharper at pattern recognition across mismatched contexts. You don’t inherit prestige; you generate legitimacy through work itself.

Maybe the goal isn’t to outgrow mudbloodness at all, but to operationalise it: learning to sit in the discomfort, creating opportunities instead of expecting the right circumstances to provide them.

Being the mudblood means holding the door open for muggles despite the hissing around: create the circumstances you wish others had created for you. Holding your ground when discourse turns too insular, and seeing both sides clearly enough to improve them in ways only the in-between ever can.

If anything here resonates, it’s probably because you’ve felt the same tension: between fitting the mold and breaking it, between suppressing curiosity and letting it burn. If so, I’d argue you’re not out of place. You’re exactly where you should be.

Yutong Liu & The Bigger Picture / https://betterimagesofai.org / https://creativecommons.org/licenses/by/4.0/