In this episode, Conor highlights the On Notation keynote at RacketCon 2026 and the work of Pat Hanrahan, the latter part of the episode is a Notebook LLM generated conversation. Socials Tacit Talk YouTube Playlist Conor Hoekstra: LinkTree / Bio Show Notes Date Released: 2026-10-19 On Notation Pat Hanrahan RSL VizQL A History of Mathematical Notations by Florian Cajori The Elements of Typographic Style by Robert Bringhurst
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
ConorWelcome to episode 37 of Tacit Talk. I recently watched the Phenomenal Talk on Notation, which was a keynote at RacketCon 2026, which I believe happened just this month, which, if you're listening to this when it came out, will be in October of 2026. And the keynote was given by Pat Hanrahan, an individual I had never heard of, but apparently was the co-winner of the 2019 Turing Award.
ConorAnd the other winner was Ed Catmill, who famously is of Pixar. And I've read the book Creativity Inc., so I've heard of Ed Catmill, but had not heard of Pat. And he has an incredibly uh long career, was a founding employee at Pixar, went on to co-found the Tableau software company, which I believe later sold to Salesforce for roughly 16 billion dollars, and has been involved, or more than just involved, in creating two different languages, R S L, aka, Renderman, Shading Language, and VizQL.
ConorThat's V-I-Z-Q-L. Both of this was surfaced in the notebook LLM conversation that you're going to hear for the remainder of this podcast, but I highly recommend you go and watch the On Notation talk. It mentions both Einstein, summation notation, Penrose diagrams, APL, JK, three of the textbooks that were authored by Ken Iverson, the Florian Kajore, a History of Mathematical Notation book, which has been mentioned multiple times on ADSP with my conversations with Ben Dean, and is also mentioned in the Notation as a Tool of Thought 1979, Ken Iverson Turing Award paper.
ConorAnd last but not least, the Bringhurst The Elements of Typographic Style, I believe is the title of that book, which was recently mentioned, if not explicitly by Ben on an ADSP episode. He mentioned it after the fact when he enumerated all of the design books that he had on his shelf. So, anyways, a ton of overlap between topics that have been discussed on ADSP, my other podcast, and as I'm a huge fan, APL, array languages, and also just beautifully aesthetic things.
ConorThis talk was fantastic, and hopefully we'll be able to get Pat either on ArrayCast or ADSP in the future.
AI BobSo back in 2001, it took exactly 1,250 minutes for a supercomputer to render a single believable image of a marble bust.
AI JaneRight, which is just insane. That's over 20 hours for one single frame.
AI Bob20 hours for one frame of visual data. But somehow, a researcher who actually started his career studying the motor systems of nematode worms figured out a way to slash that rendering time down to just five minutes.
AI JaneYeah. And in doing so, he fundamentally altered the trajectory of computer graphics, hardware architecture, and uh eventually the entire modern AI revolution.
AI BobAaron Powell Exactly. So today we are opening up a really eclectic stack of sources to do a deep dive into the architectural legacy of Pat Hanrahan.
AI JaneIt's quite a stack, too.
AI BobIt really is. We're pulling from his 2001 Seagraph paper on subsurface light transport, transcripts from his 2026 RacketCon keynote called On Notation, some historical reviews, QA sessions from Touring Minds, and biographical records covering his touring award and his three Academy Awards.
AI JaneRight. But I think we need to be clear about our mission here today.
AI BobYes, definitely. The goal here isn't to just list off what he built. The goal for you listening right now is to understand a master class in cognitive scaffolding.
AI JaneAaron Powell Because whether you are writing code, structuring a massive database, or you know, just trying to build a slide deck that doesn't put your team to sleep, the way you choose to physically write down your thoughts dictates the limits of what your brain can invent.
AI BobAaron Powell Okay. Let's unpack this. Okay. Trevor Burrus Because it's wild. How does a guy who started out studying nematode worms end up winning three Oscars, a Turing award, and basically laying the foundation for modern AI?
AI JaneAaron Ross Powell Well, it all starts with that completely atypical foundation. I mean, we tend to view tools as passive, right? Like English or C are just vehicles for ideas already in our heads.
AI BobRight.
AI JaneBut Hanrahan's career shows that symbols and syntax are an active cognitive medium. And to get why he sees it that way, you have to look at where he came from. He wasn't some computer science prodiging mainframes.
AI BobAaron Ross Powell No, he was born in Green Bay, Wisconsin in 1955. And he went to UW Madison and got a bachelor's in nuclear engineering in 1977.
AI JaneAaron Ross Powell Nuclear Engineering. Right. Not computer science.
AI BobRight. And then he stays there to get a PhD in biophysics in 1985. So I mean, how does nuclear engineering and biophysics translate to drawing computer graphics?
AI JaneAaron Ross Powell It translates through physical simulation. When you study nuclear engineering, you are dealing with particle transport.
AI BobLike neutrons moving through a reactor core.
AI JaneExactly. You use heavy math to model how they move, how they scatter, how they're absorbed. And then in biophysics, you look at biological systems through deterministic physics.
AI BobAaron Powell Right, which leads to the nematode worms.
AI JaneRight. Trying to map out a biological neural network and simulate its physical output. So he isn't looking at a screen thinking about pixels.
AI BobHe's acting like a physicist.
AI JaneYes. He's looking at physical reality and the mathematical rules governing it. And that distinction is everything because in the late 70s and early 80s, traditional computer graphics was obsessed with geometry. Trevor Burrus, Jr.
AI BobDrawing polygons.
AI JaneExactly. It was a shape drawing problem. How do we mathematically describe a triangle and rasterize it onto a grid of pixels?
AI BobBut hand-rahan sees it differently. Trevor Burrus, Jr.
AI JaneHe sees a transport phenomenon. Just like neutrons in a reactor, he sees photons moving through space. To him, rendering is the physical simulation of light transport.
AI BobAaron Powell So instead of a painter trying to fake a shadow on a canvas, he was acting like a physicist building a virtual sun and virtual rules of optics.
AI JaneThat's a great way to put it. A light source emits photons, they travel, strike a surface, scatter based on physical properties, and eventually hit a virtual camera lens.
AI BobAaron Powell And there's a wild detail in the bios here. While he's finishing that biophysics PhD around 1982, he actually starts programming neural networks on early image processing hardware.
AI JaneWhich he later called the dark ages of neural nets.
AI BobAaron Powell Because the compute power just wasn't there yet.
AI JaneAaron Powell Not even close. Algorithms like backpropagation were barely a thing, but he was already seeing the connection between biological neural processing, physical simulation, and parallel hardware.
AI BobAaron Ross Powell He was also teaching a newly created computer graphics course in 1981. And one of his students was Donna Cox.
AI JaneAaron Ross Powell Right, who becomes a massive pioneer in scientific visualization.
AI BobAaron Ross Powell Turning supercomputer data of tornadoes or galaxies into things humans can actually see.
AI JaneWhich perfectly reflects his philosophy. Turning abstract math into visual imagery is a profound mechanism for understanding reality. It's not just a party trick.
AI BobAaron Powell So let's follow that thread out of academia. In 1986, he leaves to become a founding employee at Pixar under Ed Catmall. Right into the fire. Yeah, because they're facing a massive bottleneck. Prior rendering engines were just monolithic blocks of code. Trevor Burrus, Jr.
AI JaneDeeply inflexible. The geometric algorithms, the code drawing the shape, were rigidly coupled to the optical algorithms, the code coloring the pixels.
AI BobSo if I'm an artist in 1986 and I want my virtual plastic sphere to suddenly look like brushed metal, what do I do? Trevor Burrus, Jr.
AI JaneYou go to the core software engineers and beg them to hack the underlying engine. You couldn't just flip a switch or tweak a menu.
AI BobAaron Powell Which means doing a whole feature film with wood, cloth, skin, and metal would just become this infinitely tangled mess of code.
AI JaneIt was a total architectural dead end. So Hannahan and his team created the Renderman interface specification, and crucially, the Renderman Shading Language, or RSL.
AI BobAaron Powell His first big foray into the power of notation.
AI JaneAaron Powell Exactly. They strictly decoupled the shape from the appearance.
AI BobAaron Powell Okay, here's where it gets really interesting. If he's just separating shape from color, isn't that basically what HTML and CSS do for web pages? Why was this so revolutionary for 3D graphics back then?
AI JaneAaron Powell Well, it's sort of like HTML and CSS, but RSL went infinitely deeper. CSS is mostly declarative styling. RSL was a fully executable, domain-specific mathematical notation designed exclusively for light.
AI BobWhat does the language for light even look like?
AI JaneAaron Ross Powell It had native data types that don't exist in standard languages. Instead of just integers or strings, RSL had native data types for points in 3D space, directional vectors, and colors.
AI BobOh wow.
AI JaneYeah, with built-in mathematical operators for illumination and specular highlights.
AI BobAaron Powell So an artist could just write a tiny block of code to generate the chaotic pattern of rust on a pipe without ever touching the core code drawing the pipe itself.
AI JaneAaron Powell It completely democratized visual effects. You didn't need to be a systems architect anymore. You just needed to understand the math of the material.
AI BobAaron Powell And then Pixar builds a massive library of shaders. One for Woody's jeans, one for Buzz's plastic.
AI JaneAaron Powell And the core engine just asks the shader, hey, what color should this pixel be? This exact decoupling is what earned him his first Academy Award in 1993 and fueled the Toy Story Revolution.
AI BobAaron Powell But then he hits a wall. He leaves Pixar, goes to Stanford, and decides to tackle the most notoriously difficult appearance to simulate.
AI JaneTranslucent materials. The white whale.
AI BobAnd to understand why it was the white whale, we have to look at the old math model, the BRDF, bidirectional reflectance distribution function.
AI JaneRight. It assumes completely localized interaction. It treats light like a billiard ball. A photon hits a surface and bounces off the exact same spatial coordinate.
AI BobAaron Ross Powell Which is great if you're rendering a brick wall or a piece of steel.
AI JaneAaron Ross Powell Exactly. Light doesn't penetrate steel. But if you use the billiard ball model on a block of wax or milk or human skin, you get those cheap hard plastic mannequins. Because in reality, light penetrates translucent materials. It scatters inside, bouncing around hundreds of times, and exits at a completely different location. Aaron Powell Right.
AI BobIf you've ever held a flashlight up to your hand and seen your fingers glow red, you've seen subsurface scattering. The BRDF model would just treat your hand like plastic.
AI JaneAaron Powell So the industry knew this was happening. The problem was just the staggering computational cost of simulating it.
AI BobUsing Monte Carlo path sampling. Tracking the pinball path of a single particle of light.
AI JaneAnd you have to do that for millions of photons. Otherwise, the image is just covered in static and noise.
AI BobAaron Powell Hence the 1250 minutes to render one frame.
AI JaneRight. You can't make a movie at 20 hours a frame.
AI BobSo this brings us to the 2001 SIG Graph paper. Hanrahan, Henrik Wangensen, Steven Martioner, and Mark Lavoy introduced BSSRDF, adding scattering surface to the acronym.
AI JaneThey bypassed the brute force pinball machine with a math shortcut from diffusion theory.
AI BobThe dipole point source diffusion approximation.
AI JaneWhich is a mouthful, but it's genius. They abstracted the complex internal scattering into a fluid dynamics problem, like ink dropping into water.
AI BobOkay, walk me through the mechanics of a dipole. How do you fake the math of light scattering inside an object?
AI JaneWell, to make the fluid math work for a bounded surface like skin, you have to satisfy boundary conditions. The light eventually has to escape back into the air.
AI BobRight.
AI JaneSo they created a dipole. They instantly replaced the incoming light hitting the surface with two virtual light sources.
AI BobThey just invented fake lights in the math?
AI JaneYes. One positive virtual light source just beneath the surface radiating outward.
AI BobAaron Ross Powell But wait, if you just put a glowing light bulb under the skin, wouldn't the math say the light is shining infinitely everywhere, including straight back out too intensely?
AI JaneAaron Ross Powell Exactly. Which is why they place a second negative virtual light source just above the surface.
AI BobAaron Powell Wait, a negative light source? What does that even mean? Like a black hole that sucks up photons?
AI JaneAaron Ross Powell Mathematically, yes. A point of negative energy, a thermal vacuum.
AI BobAaron Powell That's wild.
AI JaneBy placing the positive energy below and the negative energy above, the mathematical fields perfectly cancel each other out right at the boundary layer of the surface.
AI BobAaron Powell So it forces the math to behave exactly like scattered photons hitting the boundary between flesh and air.
AI JaneAaron Powell And it's so lightweight. You aren't calculating random bounces anymore. You're just calculating the smooth diffusion gradient between two static virtual points.
AI BobAaron Powell Which drops the render time for that marble bust from twelve hundred and fifty minutes down to five minutes.
AI JaneAaron Ross Powell With image quality that's basically indistinguishable from the brute force method.
AI BobAaron Powell But they didn't just invent the math, right? They actually went out and measured real world materials to plug into the math.
AI JaneAaron Powell Yes. The paper details the specific absorption and scattering coefficients for things like skim milk, whole milk, ketchup, and human skin.
AI BobAaron Powell Okay, I have to ask. What is the mathematical difference between whole milk and skim milk?
AI JaneAaron Ross Powell It's the fat content. Whole milk has heavy multiple scattering, so the light diffuses very evenly. It's opaque white.
AI BobAnd skim milk?
AI JaneSkim milk has fewer fat particles, so it scatters light differently based on the wavelength. It's called Rayleigh scattering.
AI BobAaron Ross Powell The same physics that makes the sky blue.
AI JaneAaron Ross Powell Exactly. The smaller particles scatter the shorter blue wavelengths easier than the longer red ones. So skim milk looks slightly bluish where the light enters, but reddish where the light exits.
AI BobAnd the new model could actually simulate that exact subtle color shift.
AI JaneAaron Powell It could. And human skin is even crazier because it's highly anisotropic.
AI BobAaron Ross Powell Meaning the scattering isn't uniform.
AI JaneAaron Ross Powell Right. Skin has structure collagen, blood vessels. The data showed a single photon might scatter over a hundred times inside human skin, biased by that biological geometry.
AI BobAaron Ross Powell Which brings it all full circle back to his biophysics days.
AI JaneAaron Powell Yes. Aaron Ross Powell Understanding how laser light scatters in human tissue, which is medical physics, was applied directly to creating soft, photorealistic CGI skin.
AI BobAaron Powell Which wins him his second Academy Award in 2004.
AI JaneBut then the complex math required for these new shaders demanded immense computational power.
AI BobAaron Ross Powell Right, which leads to the hardware leap. He starts looking at the silicon itself and accidentally lays the groundwork for the modern AI boom.
AI JaneAaron Powell Yes, because at Stanford he realizes these programmable shaders from Pixar were parallel computing streams.
AI BobOkay. I want to push back on this timeline a bit. Because you're saying that the hardware required to make realistic shadows for Pixar is the exact same parallel computing model that trains massive large language models today. I am. So if Pixar hadn't needed realistic shadows, we wouldn't have modern AI today.
AI JaneAaron Powell It sounds crazy, but the underlying math structure connects them. When a CPU runs a word processor, it executes tasks sequentially, linear logic.
AI BobStep A, then step B.
AI JaneRight. But rendering a 1080p monitor means calculating two million pixels. And the top left pixel is completely independent from the bottom right pixel.
AI BobAaron Powell Right. They don't have to wait for each other.
AI JaneAaron Powell It's an embarrassingly parallel problem. And graphics hardware GPUs had been built to handle this massive parallel math.
AI BobBut those GPUs were completely locked down, right? They were just video rasterizers for gaming. Exactly.
AI JaneBut Hannahan and his student Ian Buck saw them as untapped parallel supercomputers. So in 2004, they developed Brooke for GPUs.
AI BobAaron Powell A stream programming model.
AI JaneAaron Ross Powell Yes. In traditional programming, you write a loop. Look at a million numbers one by one and multiply by two. It takes a million cycles.
AI BobA toll booth with one lane.
AI JaneBut Brooke used streams. You define a stream of all one million numbers, you write a tiny kernel of logic, multiply by two, you feed the stream into the GPU, and it processes every element simultaneously across thousands of tiny cores.
AI BobA toll booth with a million lanes.
AI JaneYes. It tricked the GPU into running general-purpose scientific math by disguising it as graphic textures.
AI BobAnd then Ian Buck takes this to NVIDIA.
AI JaneWhich directly results in CUDA, the compute unified device architecture. It unlocked the GPU natively.
AI BobAnd that's the AI connection.
AI JaneBecause years later, AI researchers realized training neural nets required billions of simultaneous matrix multiplications.
AI BobAnd the hardware was already sitting in gaming PCs. The matrix math for simulating light rays is basically isomorphic to the matrix math for adjusting weights in a neural network.
AI JaneExactly. His work catalyzed the GP GPU revolution, general purpose GPUs.
AI BobAaron Powell Okay, so he revolutionizes light, he revolutionizes the hardware. Then in 2003 he pivots again. He co-founds Tableau Software.
AI JaneWhich gets acquired by Salesforce for almost 16 billion dollars.
AI BobAaron Ross Powell Stemming from the Polaris Research Project at Stanford, let's talk about the friction of structured data.
AI JaneAaron Ross Powell Multidimensional Rational Databases.
AI BobAaron Ross Powell Millions of rows of text and numbers that the human brain just cannot parse.
AI JaneRight.
AI BobIf you're a data analyst back then and you want to know if Midwest sales beat East Coast sales, you have to write a complex SQL query.
AI JaneTrevor Burrus Select statements, joins, group by clauses.
AI BobAaron Powell You had to be a database engineer. And the database just hands you back a smaller table of text.
AI JaneAaron Powell Which you then export into Excel.
AI BobThink about how much time you've wasted fighting with spreadsheet chart wizards.
AI JaneIt was a nightmare. You highlight columns, pick up bar chart, the software puts dates on the y-axis instead of the x-axis. You have to manually swap them.
AI BobYou're translating business questions into database logic and then translating that into graphical logic.
AI JaneSo Hanrahan and his student Chris Stolte invented VizQL, Visual Query Language.
AI BobAnother domain-specific language.
AI JaneA declarative language that bridges SQL and graphical rendering.
AI BobSo how does it actually work for the user?
AI JaneInstead of writing a query, getting a table, and drawing a chart, VizQL compiles the visual layout directly from the query.
AI BobSo when I drag the sales pill onto the y-axis shelf in Tableau.
AI JaneYou aren't just drawing a chart. VizQL translates that physical drag and drop into a formal SQL query, hits the database, and maps the numbers to the height of a bar instantly.
AI BobHe essentially created a universal translator between raw databases and human visual perception.
AI JaneIt's the exact same philosophy as Renderman, decoupling shape from light or collapsing queries into charts. Titled On Notation, a continuation of his 2009 talk, Systems of Thought.
AI BobAnd he really goes after the computer science dogma here.
AI JaneThe dogma that semantics is fundamental and syntax is superficial.
AI BobThe idea that as long as it's a Turing complete language, the symbols you type are just window dressing.
AI JaneHanrahan completely rebuts that. He argues notation is active cognitive media. Syntax is the direct interface to human working memory.
AI BobHe compares mathematical notation on a whiteboard to programming languages on a screen.
AI JaneRight. Math is incredibly dense and two-dimensional, superscripts, integrals wrapping around equations. The eye takes in massive concepts at a glance.
AI BobBut it's also super ambiguous. He jokes about those viral PEMDOS debates.
AI JaneAaron Powell Because the spatial notation can be informal. And most importantly, math is unexecutable. You can't feed a whiteboard into a CPU.
AI BobAaron Powell But then code is the opposite.
AI JaneAaron Powell Right. Reading thousands of lines of monospace text.
AI BobAaron Ross Powell So he gives these historical case studies on how to fix this, like Einstein's summation.
AI JaneAaron Powell When Einstein was doing general relativity, the tensor equations required massive sums using the Greek sigma symbol. It created immense visual clutter.
AI BobAaron Ross Powell So Einstein just said if an index variable appears twice, drop the sigma symbol. It's implicitly understood.
AI JaneAaron Powell And by erasing that piece of syntax, the underlying invariant structure of the tensor is exposed. You aren't wasting mental energy parsing the add this up instruction.
AI BobAaron Powell He also talks about Penrose diagrams.
AI JaneAaron Powell Roger Penrose replaced complex tensor algebra entirely with two-dimensional topological shapes and lines.
AI BobTensors become geometric shapes.
AI JaneAnd when you multiply them, you literally just connect the lines of the shapes together on paper.
AI BobComplex algebraic proofs reduce to intuitive line manipulations. A 2D diagram acting as rigorous mathematical notation. He also cites Ken Iverson, the creator of APL, an array programming language.
AI JaneBecause in C, if you want to add two lists, you write a tedious loop. In APL, the array is native. You just write A plus B.
AI BobHe showed an example using the scan operator, right?
AI JaneThe plus slash operator.
AI BobYeah.
AI JaneIt drops a plus sign between every element and sums them instantly. He used it to visually prove that odd numbers sum to geometric squares.
AI BobHe even brings up typography.
AI JaneRobert Bringhurst's concepts.
AI BobBecause spatial alignment and font weight convey critical hierarchy.
AI JaneA bold heading tells your brain it's a new category before you read the words. Standard code strips all that away.
AI BobOkay. Let's look at the psychology behind this. Why does notation matter so much?
AI JaneHe cites two theories. First, the representation effect. Human problem solving varies by up to a hundred times based purely on representation.
AI BobAaron Powell The Roman numerals analogy.
AI JaneExactly. Try doing long division using Arabic positional notation. It's easy because the system is based on place value.
AI BobAaron Powell A five in the hundreds column means five hundred.
AI JaneRight. Now try dividing MCMXC IV by XLII using Roman numerals.
AI BobAaron Powell It's a nightmare of cognitive load.
AI JaneYou can't stack them in columns.
AI BobAaron Powell The underlying math problem is identical.
AI JaneAaron Powell The only thing that changed is the notation. Roman numerals lack the structural properties to support the algorithm.
AI BobAaron Powell And the second theory is distributed cognition.
AI JaneGood notation acts as cognitive offloading. It shifts the burden from the brain's limited working memory to the visual cortex's instantaneous pattern recognition.
AI BobAaron Powell Which is VizQL. You look at a line chart and instantly see the upper trend. You don't deduce it, you literally see the answer.
AI JaneYes.
AI BobOkay, wait. At the end of the RacketCon talk, someone asks if we should just adopt whiteboard math notation for programming. What was his take on that?
AI JaneAaron Powell He explicitly says no. Math is too messy and inconsistent. He personally prefers Haskell for mathematical reasoning now because it's concise but executable.
AI BobBut there's a tension there.
AI JaneThe tension between unified notations, so everyone understands, and flexible dialects, so people can invent new ideas. It's an unresolved trade-off.
AI BobAaron Powell Which brings us to the Turing mind's QA and his current stance on AI's role in all of this. He started in the dark ages of AI. How does he see it now?
AI JaneHis philosophy is spec, read, verify. Treat the prompt as a specification, read the generated code carefully, test it rigorously.
AI BobAaron Powell Because reading code is now more important than writing it.
AI JaneAI eliminated the friction of writing. It can blast out a thousand lines of Python in seconds, but if you can't read it, it's a thousand lines of liability.
AI BobRight. And he also talks about neural nets being used for rendering now.
AI JaneAaron Powell Coming full circle to the path sampling problem. Neural nets act as heuristics like a chess engine, proposing moves to solve impossible light scattering scenarios that analytical math still struggles with.
AI BobThe convergence of traditional computer graphics and LLMs.
AI JaneTraditional graphics hold explicit physical knowledge, gravity, light. LLMs hold implicit semantic knowledge, culture, emotion.
AI BobAaron Ross Powell Connecting them through multimodal embeddings.
AI JaneExactly. And through all of this, his core advice to students remains: curiosity and passion determine success. He didn't chase trends, he just followed what he found interesting.
AI BobSo to wrap this up for the listener, whether you are writing code, taking meeting notes, or designing a slide deck, the way you choose to represent your information will absolutely dictate the quality of your thinking. Aaron Powell To leave you with something to mull over, Hanrahan showed us that trying to do long division with Roman numerals is a nightmare of cognitive load. As you look at the tools, the apps, and the languages you use every single day to do your work, ask yourself Are you secretly doing math in Roman numerals?
AI BobWhat would your Arabic numerals look like? And how much faster could you move if you found them?
In this episode, "Representation-Driven Dispatch" (generated by Claude 4.6 Opus) is read by the Speechify text-to-speech app. Socials Tacit Talk YouTube Playlist Conor Hoekstra: LinkTree / Bio Show Notes Date Released:...
In this episode, Conor interviews Madeline Vergani about TinyAPL 0.13, inverses, AP8L and more! Socials Tacit Talk YouTube Playlist Conor Hoekstra: LinkTree / Bio About the Guest Madeline Vergani is the creator of...
In this episode, the 1972 paper "Algebra as a Language" by Kenneth E. Iverson is read by the Speechify text-to-speech app. Socials Tacit Talk YouTube Playlist Conor Hoekstra: LinkTree / Bio Show Notes Date Released:...
In this episode, the "From List Calculus to Array Calculi" (generated by GPT 5.4) is read by the Speechify text-to-speech app. Socials Tacit Talk YouTube Playlist Conor Hoekstra: LinkTree / Bio Show Notes Date Released:...
In this episode, the "Richard Bird's Algebraic Laws for Program Calculation" which is a summary of the important papers that introduced the algebraic laws of programming (generated by Claude 4.6 Opus) is read by the...