Sanderson, Grant
▶ Watch on YouTubeCore ideas
Sanderson opens with a thought experiment about alien mathematics: he expects intelligent life would independently arrive at counting and arithmetic (since repetition and counting are extremely natural concepts), but that the specific way number systems get extended (e.g., to the reals, or via alternative systems like the surreal numbers) involves genuine human choices, meaning much of what we take as "the" mathematical canon might look different for another civilization even while addressing the same underlying continuity concepts. He uses this to launch a sustained critique of the notation e^x, arguing it badly obscures the actual phenomenon it describes — a system whose rate of change is linearly related to its own value — and that framing this as "repeated multiplication" extended awkwardly into complex exponents (e^(iπ)) wastes enormous amounts of human effort, when the more illuminating framing is a 90-degree rotation applied continuously (a point always moving perpendicular to its position vector traces a circle). He extends this into a broader claim about Euler's identity: he considers it notationally "hideous" rather than truly beautiful, arguing its fame stems from historical accident (e, π, and i each show up for genuinely different, largely orthogonal reasons — e from real-direction growth, π from the imaginary-direction period — and the equation just happens to juxtapose them, "like putting all the children in one bed" when they'd rather sleep separately) rather than deep, irreducible mathematical unity.
On whether math is discovered or invented, Sanderson proposes a cyclical view: physical discoveries about the world tell us which of many possible, internally-consistent mathematical systems are actually useful, and mathematicians then invent formal representations suited to what nature revealed — illustrated by the Pythagorean theorem, which feels like pure discovery but is, at a formal level, true only because we chose to define the standard Euclidean metric on ℝ² in a way that happens to make it come out true; other equally consistent metrics exist but simply don't correspond to physical space. He extends this idea to higher-dimensional mathematics, arguing that studying five-dimensional manifolds can be perfectly meaningful and useful even in a genuinely three-dimensional physical universe (e.g., via high-dimensional state spaces), rejecting the assumption that higher-dimensional math must correspond to a higher-dimensional physical world to be relevant. Prompted on why physical law appears so strikingly simple and compressible into elegant equations, he offers no confident answer but entertains an anthropic-style possibility (that only universes admitting sufficiently simple, compressible laws could produce beings capable of noticing and formalizing that fact) alongside the more mundane possibility that human minds and physics as a discipline selectively favor systems simple enough to describe mathematically, leaving genuinely irreducible complexity to other fields like biology.
On the simulation hypothesis, Sanderson pushes back specifically on the common "if simulations are possible, they'll recursively nest indefinitely, so we're probabilistically almost certainly deep in the stack" argument, noting this implicitly assumes an unbounded, uniform prior over how many nested simulation layers could exist — an assumption he considers unjustified once you account for physical limits on information processing (citing real physical bounds, from combined general relativity and quantum mechanics, on how much information can be stored per unit area before gravitational collapse into a black hole). If such fundamental information-density limits exist at even the "top" level of reality, he argues, the number of computationally expensive nested simulations any base reality could support is likely far more constrained than naive recursive-simulation arguments assume.
On the psychology of abstraction, Sanderson defends the coherence and utility of infinity by reframing it not as some intractably "woo" concept requiring literal visualization of infinite completed collections, but as the simple, graspable property of always being able to add one more element to a sequence — directly analogous to how facial recognition unifies a "Lex" concept across an infinite variety of possible retinal images sharing certain common patterns; in both cases, the abstraction's power comes from applying to a multiplicity of situations sharing a property, not from being physically instantiated in full. He connects this to his own pedagogical philosophy: rather than presenting mathematics top-down (starting from an abstract definition, like a vector space's axioms, and populating it with examples later), he builds videos bottom-up from concrete, specific instances (an actual arrow with a specific x/y magnitude, not "a vector that could apply to anything") — arguing that visual media inherently forces this discipline, since you cannot animate an unspecified generality, only a concrete choice, and that letting a learner's brain first pattern-match across several concrete examples makes an eventual formal definition land as an articulation of something already intuited rather than an opaque, high-cognitive-load starting point.
Sanderson names the Euler product formula for the Riemann zeta function (relating an infinite sum over natural numbers to an infinite product over primes, encoding the fundamental theorem of arithmetic) as an early formative "most beautiful idea" encounter from a children's book, and reflects that the most beautiful mathematical objects for him are ones with just enough partial understanding to retain genuine mystery — noting that fully "dated" mathematical objects (like Euler's identity, which he says he now understands almost too well) lose some of their initial romantic charge, comparing this to the difference between the early excitement of dating someone and the settled, different kind of appreciation of a long relationship. He identifies beauty specifically with mathematical objects that feel simultaneously "not arbitrary" (a sense that any sufficiently curious civilization would independently arrive at studying the same object) and genuinely surprising/intricate, contrasting this with "contrived" constructions that feel like arbitrary human choices. He recounts his favorite video to make, "Who Cares About Topology?" (covering the inscribed square/rectangle problem and its connection to the Möbius strip and torus), specifically because rendering the 3D surface used in the proof (via a non-constructive topological argument) produced a genuinely unpredictable, visually striking shape he couldn't have anticipated in advance — illustrating his broader point that forcing an abstract mathematical idea into a concrete visual representation often reveals genuinely new insight beyond what the original written proof conveyed.
On learning mathematics, Sanderson advises against passive consumption (reading or watching lectures) in favor of actively working through curated problem sets (checking chapter-end exercises before and after reading, and not considering a chapter "done" until several problems are solved) — explicitly noting the irony that his own videos, lacking associated exercises, represent only a partial piece of the learning ecosystem, not a substitute for active practice. He also recommends programming as an unusually effective, underappreciated on-ramp into mathematical thinking for people who don't respond to traditional instruction, and cites the (possibly apocryphal but intuitively compelling) "learning pyramid" statistic that people retain roughly 90% of material they actively teach versus 10% of what they merely read, as his rationale for why explaining concepts (e.g., via video) is such a powerful personal learning tool, distinct from its value to an audience.
Excerpts
On why he considers the notation e^x fundamentally misleading:
“It's just the wrong way of notationally representing this function... what it's actually saying, like if you really parse something like e^(iπ), is: choose an origin, always move perpendicular to the vector from that origin to you... I can't even think of the number of intelligent human hours that have been wasted trying to parse that to their own liking.”
On why he doesn't consider Euler's identity the most beautiful equation in mathematics:
“As we represent it, it's one of the most hideous — it involves a lot of the most hideous aspects of our notation... the idea does tie together, when you wash away the notation. Look, it's okay, it's pretty, it's nice, but it's not like the mind-blowing greatest thing in the universe.”
On why abstractions like infinity aren't mysterious once properly framed:
“The key entity is the property of always being able to add one more... it's that property — 'oh, I could always add one more' — that gives it this nature of infiniteness, in the same way that certain properties of your face give it 'Lex-ness.' Infinity should be no more worrying than the 'I can always add one more' sentiment.”