The Roommate that Totally Changed Me
This summer, I lived with two strangers I had never met before, from two different countries, in a city none of us had ever been to before, for a month and a half. At the end, I emerged a different person – I chose different things to focus on in my leisure time, aspired to different goals, and thought in a completely different way. What happened?
In many ways, this is a story about someone else. Yet, I want to tell this story because it is fundamentally a story about myself: about what I noticed, how I noticed, and most importantly, how I began my process of change.
During summer, I chose one of my roommates, Justin, based on nothing other than two blog posts. One about building your own playbooks, and one about reading different books. Both of these posts deeply resonated – romantic ideas in my head, till that moment I was handed the words to express them. How he chose books to read, how he applied the knowledge, how he focused, and how he played all came together to define an entirely different life than the one I had been living.
I’ll quickly describe the first idea before focusing on the second: the playbook idea is, briefly, that following a list of instructions is easy (conditioned on discipline), and that the vast majority of aspirations have such instruction lists attached to them. The most risky opportunities had no such playbooks – this is what made them higher stakes, higher reward, and more fun. This was evident in what Justin found interesting; the more unclear steps forward, the more interesting – his space self-replicating machine startup and his latest zero knowledge crypto game both had very unclear playbooks.
Reading
The second post discussed how Justin reads differently – older, ‘Lindy’ books along with random titles he found interesting at the bookstore. It showed with the titles littering his bedside table, and how he opted to read with the vast majority of his leisure time (on the bus, in the park, during the weekends, on inspiration-less days, etc). Titles like ‘Simulation and Simulacra’ and ‘This is How You Win the Time War’ littered his bedside table, overflowing onto the floor. I found this fascinating: how does one choose, then commit to, books? So far, my strategy had been to collect recommendations and just read the ones with the highest recommendation count. A route to commoditized success like many of my friends perhaps, but not a route to a unique worldview.
His route was different: for academic subjects, simply find the smartest academic and read the textbook; for all other more philosophical reading, read whatever’s interesting. The canonical Goodfellow machine learning and Susskind/Feynman’s lectures in physics, along with the best Stanford, MIT, and Cal courses, were the stepping stones he used to learn these skills without the signaling credential. Physics from Symmetry revealed beauty in rederiving physics from only fundamental axioms, and the Princeton Companion to Mathematics (MAA Euler Award) was a fun hour a day to read on a new mathematical topic. The Dream Machine gave an intuition on how to sniff for great new possibility-enabling technology – honing an intuition in a way that only experience (or reading about it) can. Getting as close to the source as possible was a constant theme, and such first principles thinking led to an ability to invent new things.
Mastering foundations isn’t isolated to science – by seeing life choices in the same way, one can argue the analogy of first principles thinking to the good life is philosophy: the root from which your decisions are made. For life philosophy, fiction, and other interesting musings without a cold scientific truth, you might as well dig deep on what you find interesting. The foundational bible here is ‘Finite and Infinite Games’, but he had various other tomes on greatness, gender, and reality. According to his roommate, Godel Escher Bach was leisurely finals week downtime. In the sci-fi section at the used book store, he excitedly pointed out titles he loved – Snowcrash (a sci-fi classic), Blind Sight (about aliens), and Ted Chiang’s short stories (especially the Limitless one). I was struck by how he spent hours in a bookstore looking for titles – a complete disdain for the average review count, trusting (and therefore calibrating) opinions which are only his.
I wondered how one made sense of such chaotic choice – it seemed like such a reckless way to choose. I had the common opinion that one should finish what they start – surely this would guarantee more reading. However, if you’re more excited to read some other book after a chapter or two of your current one, it’s good to switch! There’s always a book you’re most excited to read, and sticking to that is how you can ensure you only finish the books that you are most excited about throughout, and read the most pages of only those books that resonate (there’s a fine line between resonation, confirmation bias, and reading what you actively disagree with to steelman/swap beliefs – as long as you are intellectually honest, you can generally avoid such traps). Discarding poor books after a few chapters was his only way to leave a book with high conviction. Many readers I’ve met read many books at once with a similar philosophy (see: Patrick Collison), and Justin was no exception. I too now have a smattering of tales on my desk, toiletside, couch, and backpack.
I boiled down what I was impressed by into two facets: his focus, and his reading. One argues focus can follow from reading long-form content, especially in the ADHD-inducing attention economy: in his experience, reading in crowded buses was a natural stepping stone to extreme focus. So, perhaps it is only books, and focus follows. Such focus allowed him to dig deep into the ground source of an idea, instead of ever relying on “cheap opinions” by thought-leaders. Thinking deeply and focusing only on the task at hand, led him to attack discussions with ease. During Ethereum Minicon talks, it was quite impressive how he described his method of understanding Solidity’s diamond pattern and relayers: by reading the source code of Uniswap and Gas Station Network respectively (uncommon, but not a new idea: see also Surya D). Such deeply focused multi-hour sessions are surely the only way to really learn something new and develop informed first-principles opinions, and a skill that I’ve found Olympiad competitors, researchers, and voracious book readers alike share, but few others in my circles. He also swiftly understood most of my internship project in 5 minutes and then proceeded to point out a flaw that I had not considered (specifically, that outer products in a neural network could not possibly be equivariant if compressed into a summation).
I had never taken significant time to read books, but Justin made two points that shifted my mindset. For one, he described it as a high bandwidth conversation. I had never seen books this way: after all, you can’t ask questions to dead carbon. But it’s more subtle than that – a good author, having gone over their words with a fine tooth comb, has likely anticipated the logical jumps a careful listener will protest, and likely addressed them. Since your questions are likely answered in this highly edited narrative structure, your brain can then reallocate most time spent on hole-finding, to instead deeply understanding.
The second was the sheer number of books one could get through at an hour a day: even barring vacations, thats 300 hours a year. At 10 hours a book, 30 books. For a textbook with 500 pages and 200 words a page, reading at 100 words a minute and a 3:1 ratio on exercise time to reading time, that’s only 4000 minutes = 60 hours per textbook, or 5 textbooks a year. With another hour or two per day on application, that’s enough to be a formidable expert, as he points out. This was a far cry from the 2-3 books I rushed through a year – and it shows in the crumbling foundation my ‘original’ opinions go through when opposed with an entire book of structured reason. Ideas that I thought were original, hardly scratched the first few pages of a book on the topic. Discourse with someone who had read was then a fruitless and unbalanced endeavor for us both, although we were both clearly excited about that idea.
Originality
His focus and reading also powered his relentless originality: a Turing complete DAO proposal, a formidable new zero knowledge game where Dark Forest was but the most basic building block, and a past self replicating space machine startup – all projects I’d have been excited to work on. His passion projects were also exciting, and a primary reason I found him such a curious and unique character – a losing crypto trading bot repository lay strewn amongst the novel reinforcement learning framework to distribute training on entire clusters, and a GPT-3 crossover with Dark Forest. He didn’t just read and forget – he turned the knowledge into wisdom through application.
Justin’s maddeningly sharp intellect was never thrust upon you in the way some arrogant people recklessly do: you only realized you were wrong when he steered you to realize a fallacy in your own logic. I felt surprising satisfaction when he said he would have wanted to trade places with me when founding my last startup, but an equally deflating downfall when he accurately asked if I could have done both that and school, with discipline. I noticed this pattern when he asked me to explain how MACI could solve sybil attacks in deepening detail and I faltered, on Discord when I proposed a multi-sig implementation of a general asset marketplace but couldn’t explain details, and on Twitter when I brought up a slightly flawed proposal for a “turing test score” for on-chain frameworks. This concept naturally flows from intellectual honesty and wanting to understand the other person’s position over pushing your own, but is hard to do effectively across contexts. Exercising originality by inspiring originality in another person is a beautiful concept from Carse’s text, and is exactly what is drawn on here. You are never told you’re wrong, ever, but he asks you to sufficiently elaborate till you yourself realize your own folly: the sign of a formidable teacher that teaches through empathy and logos, never vis.
Play and Love
What was most interesting about his intellectualism though, was his refreshing playfulness in downtime. Humor with a dash of braggadocio was par for the course: game theory comments on zero sum games at the fondue pot, exclaiming just as Rick did with memory parasites that “there are six people in this room!” when we were doing headcounts, and Gru 4 step plan memes when Brian missed his own intricately organized conference. One who sees the world through an infinite context sees July 4 fireworks as cavemen coming to the fire at the center of camp, and riding a yacht like piloting a multi-bandit spaceship. When he takes time at dinner to appreciate how these friendships and drinks with the boys were the moments that really mattered, he was right. Playful exploration and casual connection through metaphor space with memes and math make his benign violations hilarious and relatable.
His play wasn’t just confined to Meeseeks and Black Mirror episodes – on a particularly uninspired day, he left us to sit in the park and read a game design textbook. When we met up for dinner that night, he bustled in with the frenzy of Archimedes after his revolutionary bath, yelling how he figured out a huge problem. He then laid out a theory from the text: the way to make Dark Forest and his game more fun – embodied markets! He spent the entire appetizer spreading out his full theory atop the tablesheet as we all listened, enraptured. Once we had settled into the meat-laden tapas (Justin’s description being along the lines of the cavemen descending into the flesh of our less fortunate self aware brethren), it felt like the start of a new paradigm for gaming and crypto.
A few months later, Justin has staked a key part of his game on the above principle. Like established artists exploring a new path to risk their career on, such large ideological risks are hard to come up with, justify, and extend yourself from first principles without the aid of anyone else. It makes complete sense he delved into the literature before banking months of work on his whole game on a new principle. I’ve seen this concept before, like when Ed Boyden refers to random extremely old cross-discipline literature to inspire new ideas: it’s exciting to see it done in a non-research context as well.
Thinking and execution were siloed under the bigger umbrella of ‘brain in a vat’ days – after days of grinding, Justin expressed the lack of physical satisfaction as such. The ‘base pleasures of the body’ are simple, but key – mini triathlons and physical sensation rounded out the human experience.
Beyond just play, Justin embraced emotion and love. When he read Meditations and Epicurius’s ideas on the classic valley philosophy of Stoicism, he actually full sent for a few months: finding it made him too emotionless, he abandoned it, establishing conviction against it that would last a lifetime. I was also struck by his obsession with labors of love. Projects people are so emotionally convicted in, they spend years of their lives exploring some idea or passion. He mentioned this as part of the beauty of Outer Wilds, where a single developer had this idea and assembled a team and sold and resold the game for nearly 7 years, before they released it to near universal indie game acclaim (a game that is most beautiful played knowing nothing beyond the trailer, so don’t go googling the plot :)). He also mentioned this for Finite and Infinite Games – Carse apparently wrote a manuscript over years that was 400 pages long, then lost in Paris somewhere and frustratingly rewrote into an extremely dense and punchy 150 pages (source).
Perhaps it’s no surprise that at the last dinner after Justin left, Flo called me the “resident Justin expert” when people asked questions – with a decent model of his mindset, I felt like I was really starting to understand this person – and pick up the habits that I admired. It’s interesting that Justin impacted not just me, but seemingly everyone he came in contact with. All his roommates from college are off doing exciting things, as clearly Flo is and (hopefully!) I.