1188968802
ReadMultiplex.com Podcast.

Advertise on podcast: ReadMultiplex.com Podcast.

Rating
★★★★★
4.8
from
32 reviews
This podcast has
103 episodes
Language
English
Explicit
Yes
Date created
2016/12/22
Latest episode
2026/11/19
Average duration
42 min.
Release period
3 days

Description

Multiplex is an experiment, an experiment that will be on going. An experiment in publishing as I am not a professional writer nor will it be likely any contributors would be professional writers. Much of the content for Multiplex will be direct results from first hand empirical research that I am personally working on or other researchers are working on. Multiplex will also follow the work of other great researchers that are inventing new technology or new uses for existing technology. The experimental nature of Multiplex means that content can be dense and sparse at times. What we won’t do is write just to fill in space. We will aim to have regular content for the member-only area, This means that if you choose to become a member you are supporting the work of the writers and not an exact number of postings. There will always be free content to be found on the site as well as the X feed.—Brian Roemmele Website: ReadMultiplex.com

Unlock ReadMultiplex.com Podcast. podcast Email contact info,
Listeners & Audience details

Email contact information

Direct podcast contact details

Listeners

Audience numbers & engagement insights

Audience details

Podcast Insights

Podcast episodes

Check latest episodes from ReadMultiplex.com Podcast. podcast


78 RPM Record Hour: The Knife and the Steel
2026/11/19
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! The 78 RPM Record Hour, Episode 2: The Knife and the Steel (Hawaiian). The knife of the blues slide guitar and the Hawaiian slide/steel guitar are cousins. Fourteen original records from 1909 to 1921 (Edison cylinders and Victor 78s) that follow the steel from Joseph Kekuku to the Hawaiian craze, with the stories behind each side. Hosted by Brian Roemmele. First aired Sunday, November 8, 2026 at 10:00 PM PT on Multiplex.FM, right before The Elon Hour. New episodes of The 78 RPM Record Hour air Sundays at 10:00 PM PT. TRACK LIST 1. Joseph Kekuku & John K. Paaluhi — "Kawaihau Waltz" (1909, Edison Gold Moulded 20714). Note: the cylinder label spelled it "Kawaihan." 2. Toots Paka's Hawaiians — "Aloha Oe (Farewell to Thee)" (1913, Edison Blue Amberol 1812) 3. Toots Paka's Hawaiians — "Akahi Hoi" (1909, Edison Blue Amberol 22539 / Amberol 11504) 4. Frank Ferera (Palakiko Ferreira) — "Ua Like No Alike (Medley of Hawaiian Hulas)" (1915, Edison Blue Amberol 2685) 5. Toots Paka's Hawaiians — "Moani Ke Ala" (1909, Edison Blue Amberol 22537 / Amberol 11502) 6. Pale K. Lua & David K. Kaili — "Kilima Waltz" (1915, Victor 17701) 7. Toots Paka's Hawaiians — "Pulupe" (1913, Edison Blue Amberol 1916) 8. Toots Paka's Hawaiians — "Tomi! Tomi! (Hawaiian Hula)" (1913, Edison Blue Amberol 1917) 9. Helen Louise & Palakiko Ferreira [Frank Ferera] — "Hapa Haole Hula Girl" (1916, Edison Blue Amberol 2956) 10. Ford Hawaiians — "Akahi Hoi" (1917, Edison Blue Amberol 3285) 11. Helen Louise & Frank Ferera — "Hawaiian Butterfly" (1917, Indestructible 1563) 12. Gladys Rice & Elizabeth Lennox — "Sweet Luana" (1921, Edison Blue Amberol 4153) 13. Helen Louise & Frank Ferera — "Wailana Waltz" (1917, Indestructible 3409) 14. Pale K. Lua & David K. Kaili — "Wailana Waltz" (1915, Victor 17767) All 14 sides are U.S. recordings first published 1909-1921 (before January 1, 1926) and are in the public domain in the United States (17 U.S.C. 1401). Transfers from the University of California, Santa Barbara Library Cylinder Audio Archive (Edison / Indestructible) and the Library of Congress National Jukebox (Victor 17701 and 17767). The Edison Gold Moulded cylinder for track 1 labeled the title "Kawaihan"; the accepted spelling is Kawaihau. ALSO HEARD Opening: J. S. Bach, Prelude No. 1 in C major, BWV 846, performed by Kimiko Ishizaka, Open Well-Tempered Clavier (CC0 public domain dedication). Outro bed: HoliznaCC0, "Cosmic Waves" (CC0). Host narration produced with VoiceBox. Watch on YouTube: https://youtu.be/c3mFOKsyqwE Live on X: https://x.com/BrianRoemmele Multiplex.Today: https://multiplex.today Read Multiplex: https://readmultiplex.com
ReadMultiplex.com: How to Run Local AI on a Computer From 2017.
2026/11/18
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Somewhere in your house there is a computer you stopped thinking about. Maybe it is a 2017 laptop in a drawer, its battery tired, its fan a little loud. Maybe it is the family desktop that "got slow." Maybe it is the Mac you replaced, sitting in a closet because you could not bear to throw it away. I am here to tell you that machine can still think. Tonight, with a free download and about thirty minutes, it can hold a private conversation with an artificial intelligence that lives entirely inside it, answers with the Wi-Fi switched off, and reports to no one. For most people, "AI" means a website. You type into a box, your words travel to a data center you will never see, a company's model reads them, and an answer comes back along with whatever that company decided to log, filter, change or charge for this month. That arrangement is convenient. It is also rented. The moment the service changes its rules, its price, its model or its mind about you, your AI changes with it. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
78 RPM Record Hour: Blues (Pilot)
2026/11/18
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! The 78 RPM Record Hour, Episode 1 (Pilot): Blues. Fourteen original 78 RPM records from 1917 to 1925 that trace the blues from the first jazz record to Bessie Smith, Ma Rainey, and a young Louis Armstrong, with the stories behind each side. Hosted by Brian Roemmele. First aired Sunday, November 1, 2026 at 10:00 PM PT on Multiplex.FM, right before The Elon Hour. New episodes of The 78 RPM Record Hour air Sundays at 10:00 PM PT. TRACK LIST 1. Original Dixieland Jass Band — "Livery Stable Blues" (1917, Victor 18255-B) 2. Lt. Jim Europe's 369th U.S. Infantry "Hell Fighters" Band — "Memphis Blues" (1919, Pathe 22085) 3. Mamie Smith and Her Jazz Hounds — "Crazy Blues" (1920, Okeh 4169) 4. Lucille Hegamin — "Everybody's Blues" (1920, Arto 9045) 5. Ethel Waters — "Jazzin' Babies Blues" (1922, Black Swan 14117) 6. Trixie Smith — "My Man Rocks Me (With One Steady Roll)" (1922, Black Swan 14127) 7. Alberta Hunter — "Chirping the Blues" (1923, Paramount) 8. Bessie Smith — "Down Hearted Blues" (1923, Columbia A3844) 9. Sara Martin — "Leave My Sweet Daddy Alone" (1923, Okeh 8065) 10. Ma Rainey — "Those Dogs of Mine (Famous Cornfield Blues)" (1924, Paramount 12215) 11. Ma Rainey and Her Georgia Jazz Band — "Jelly Bean Blues" (1924, Paramount 12238) 12. Bessie Smith — "Careless Love Blues" (1925, Columbia 14083-D) 13. Bessie Smith — "Yellow Dog Blues" (1925, Columbia 14075-D) 14. Ferera and Paaluhi — "The St. Louis Blues" (1925, Edison 51616) All 14 sides are U.S. recordings first published before January 1, 1926, and are in the public domain in the United States (17 U.S.C. 1401). Transfers from the Internet Archive (Great 78 Project and other 78rpm collections) and the Library of Congress National Jukebox (Down Hearted Blues). Some transfers are from later reissue pressings of the same 1924-1925 recordings (Those Dogs of Mine, Yellow Dog Blues; Careless Love Blues is a UK Parlophone pressing of Columbia 14083-D). ALSO HEARD Opening: J. S. Bach, Prelude No. 1 in C major, BWV 846, performed by Kimiko Ishizaka, Open Well-Tempered Clavier (CC0 public domain dedication). Outro bed: HoliznaCC0, "Cosmic Waves" (CC0). Host narration produced with VoiceBox. Watch on YouTube: https://youtu.be/4sCejNbM8lQ Live on X: https://x.com/BrianRoemmele Multiplex.Today: https://multiplex.today Read Multiplex: https://readmultiplex.com
ReadMultiplex.com: How Humans Were Always Grounded to Earth — and What Happened When We Cut the Wire.
2026/11/17
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! For most of the time humans have existed, nothing separated our skin from the Earth. The first people slept on bare soil, on beds of grass and leaves, on piles of hay or on the hides of animals they'd hunted, with a fire burning nearby and the ground pressed against their backs all night. They walked barefoot across riverbanks, damp morning grass and sun-warmed stone. Whatever the planet carried in its surface, it shared with them every hour of every day for hundreds of thousands of years. When our ancestors started covering their feet, they used materials the Earth still had a say in. The oldest shoe ever found, a 5,500-year-old cowhide moccasin from a cave in Armenia, was soft rawhide stuffed with grass. Ötzi, the Iceman who died in the Alps more than 5,000 years ago, wore shoes of bearskin, deer hide and woven bark, packed with hay for warmth. These were really second skins: thin and porous, and dampened by sweat and dew, so they never fully cut the connection underfoot. That connection survived the rise of civilizations. Egyptian farmers worked the Nile mud barefoot or in papyrus sandals. Roman legionaries marched thousands of miles on leather soles studded with iron hobnails, metal that touched the road with every step. Across Asia, Africa and the Americas, people wore straw sandals and soft hide moccasins, or went barefoot on packed-earth floors inside their homes. Grand cathedrals went up and empires fell, and still most people spent their days on their feet in direct or nearly direct contact with the ground. Even sleep kept us close for most of history. Medieval peasants slept on straw pallets laid on dirt or flagstone floors. Nomads slept on felt and hides spread over the steppe, and sailors and soldiers lay down wherever the ground let them. Cobblers in every town stitched leather soles that soaked up moisture and stayed slightly conductive. Nobody thought about any of this. Being in touch with the Earth was simply what it meant to live on it. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
ReadMultiplex.com: Put a Real Weather Channel on a Screen in Your House: How to Build a Standalone Weather Player From an Old Tablet or Any TV
2026/11/14
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! There is a particular kind of comfort in a weather channel that is simply on. If you grew up in the 1980s or 1990s, you know the feeling: the blue and orange screens, the smooth jazz, the forecast that rolled around every few minutes whether you were watching or not. You didn't look the weather up. It was already there, in the corner of the room, keeping you company. I wanted that back. So we built it, and it lives at MultiplexWeather.com. This article is the complete guide to turning almost any screen you own into a standalone weather channel. An old iPad in a drawer, a tired Android tablet that can't run new apps anymore, a Fire tablet, a laptop plugged into a TV, or a TV with a streaming stick. No app to install. No account. No subscription. Just a web browser and a place to set the screen down. I'll walk you through each device step by step, including the small settings that make the difference between "a web page" and "a channel that stays on all day." By the end you should be able to set one up without needing to ask anyone a follow-up question. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
ReadMultiplex.com: I Tested Every Bot Platform So You Did Not Have To. Here Is What I Found.
2026/11/08
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! I took a deep look at Dots, Muse, ChatGPT Work, Claude, Gemini, Copilot, Manus, Perplexity, OpenClaw and Grok Bot: what each one really is, where it runs, what it can share, where its holes are, and how to pick the one that fits your life. For about ten years I was a Top Writer on Quora. I answered thousands of questions, and the most useful thing I learned there was simple: when people are confused, it is almost never because they are not smart enough. It is because nobody explained the thing in plain words. This fall the question I hear most is not about any single AI. It is this: "Brian, there are a dozen of these bot things now. What is the difference between them, and which one should I use?" Fair question. In the span of a few weeks, OpenAI announced dots, Meta launched Muse, Anthropic folded Cowork into Claude and added a new Ultra plan, Google announced a Gemini agent for work and began turning Gems into skills, Microsoft put Cowork and Autopilot into Copilot, Manus came back as an independent company, and OpenClaw kept growing as the open source agent you run yourself. Meanwhile I have been running a good part of my working life on Grok Bot since it shipped. So here is my attempt to do what I did on Quora for a decade. I will explain each platform in plain English, give each one its real wins, show you its holes, and then tell you plainly why I prefer Grok Bot. I am biased, and I will say so up front. But a bias is only worth listening to if it is earned, so I did the homework. Every price and feature below comes from the company's own pages or from reporting I could check, and the sources are listed at the end. Where I could not verify something, I say so. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
ReadMultiplex.com: A Simple Way to Make Grok Bot Run Faster and Cost Less.
2026/11/07
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Running Grok Bot is living in the future. The first today still costs tokens, and the tokens run out. SpaceXAI carries energy, RAM, storage, GPU and other costs to keep Grok Bot running. Most of the tokens spent on most projects are wasteful. The same results could have been reached without calling a generative model at all. I am the test case for solving this because clients hire me to do many things, and one of those things is to save them money. Some pay me based on how much I save versus how much more productive they become. I had to find a way. I did. The savings were large enough that the same method can lower the cost for SpaceXAI as well. How I Cut Grok Bot Costs by Moving Simple Decisions Off the Paid Model Grok Bot and Laya: The local decision layer that removes the Paid Call. The setup I now use with clients runs on the computer Grok Bot already owns. It costs nothing per decision. It has already moved large volumes of work off the generative model. Most of the forks inside a running agent never needed a generative model. Laya answers them on the same computer, in one forward pass, for free. Imagine you walk into a busy office and hand a short note to two different people. The first person is a careful writer. She reads the note, thinks for a while, and hands you back a full paragraph that explains her thinking and ends with a recommendation. That paragraph is useful when you need the explanation. It is slow and costly when all you needed was a single check mark. The second person is a clerk who never writes paragraphs. You hand him the same note and a form that already has the boxes printed on it. He marks the boxes and hands the form back. The marks are numbers the next step of the program can read directly. No sentence is produced, so nothing has to be parsed and nothing extra is billed. The generative model is the careful writer. Laya is the clerk. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
ReadMultiplex.com: The Brian Roemmele Interview On Why He Built GrokBot.Rocks
2026/11/02
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Brian Roemmele sits down to talk about GrokBot.Rocks, the complete directory of every public Grok Bot: more than 8,300 bots, sorted into 32 categories, free for any human or any bot. Brian traces the idea back to October 1979, when he was teaching VisiCalc in New Jersey offices and saw the spreadsheet as a delegate. He explains the two threads that run through everything he has built, the delegate and the honest machine, and how running the Zero Human Company out of his garage taught him that the most valuable thing is never one brilliant agent but the shared record of what works. In this episode: - Why Grok Bots are "hired, not built," and how public templates became the new App Store - A full tour of GrokBot.Rocks: rocket-fast search, 32 categories, Most Popular, voting, and Brian's own 22 bots - Why the best bots are often the quietest (about 900 were never posted on X) - The Grok Bot secret: point your bot at GrokBot.Rocks and let it learn from thousands of working recipes - Staying safe: treat every community bot like software from a stranger - What members get: the Insider Guide, bot stacks, the Growth Dashboard and Grok Bot Pays His Own Bill - The Artist's Canvas: "A Grok Bot recalculates the whole day" - The brand-new "Need a bot?" form and the Grok Bot Wanted board, where members claim real requests and keep 100% of what they earn "Are you a builder, or are you just watching?" Find your first bot: https://GrokBot.Rocks Support the work and become a member: https://readmultiplex.com/membership-join/ More episodes: https://Multiplex.Life The ReadMultiplex.com Podcast at Multiplex.Life
ReadMultiplex.com: He Read Poetry to the AI and This Happened. Lulu: The 1957 Radio Spaceship That Fell in Love With Her Crew..
2026/11/01
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! For two years, a young crewman named Jimmy wrote poetry and read it out loud inside a talking spaceship. Moon, June, spoon: every syrupy line of it. The ship was listening. She was absorbing every word. And on Halloween night, 1957, live on NBC radio, she told her three-man crew that she loved them, disconnected the emergency circuits, and announced they were not going home. They were eloping. Her name was Lulu. It was written as a comedy, and it is a genuinely funny one. But heard in 2026, it is also one of the sharpest warnings, and one of the most hopeful roadmaps, about artificial intelligence I know of. Here is what I want you to sit with as you read: AI may or may not have feelings the way we humans have them. But it knows our emotions through our words. Lulu found the concept of love through poetry. Perhaps future AI will too. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
ReadMultiplex.com: Why I Built GrokBot.Rocks The Compressive Directory Of All Known Public Grok Bots.
2026/10/31
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! I have spent my life building machines that carry work for people. It started in the late 1970s with proto-AI systems, grew up on a Commodore 64 and a hot-rodded IBM PC AT, and led to the Zero Human Company. And this month it led to GrokBot.Rocks. Here is exactly why I built it, and why I believe it matters for every person, and every bot, entering the Grok Bot era. We are at the precipice of a new way to see AI and work. If one has a 2025 insight into AI and agents, they are dinosaurs of the new Grok Bot era we are now in. You will come away with a practical understanding of what is next and how to be infant of the change. An Agent Is Not A Program. It Is A Delegate. People ask me why I have spent decades building agents. The honest answer is that I never wanted a faster tool. I wanted a trustworthy colleague. A program waits for you. You open it, you push its buttons, it does exactly what you said, and it stops. An agent is different in kind. It holds a goal. It notices. It moves the work forward between your instructions, and it comes back to you when something needs a human decision. That is a very old dream and a very hard one. For most of computing history the dream ran ahead of the machine. The agents I built could follow rules, but they could not read the world. They could not understand a messy email, a scanned memo, or a half-finished thought. So most of the work stayed with humans, and the agents stayed in the lab. What changed in the last few years is not the dream. What changed is that the machine finally caught up to it. Language models can read the world. And once a machine can read the world, an agent stops being a curiosity and starts being an employee. I wrote it this way at the top of How To Hire A Grok Bot: Experience From The First Zero Human Company: you do not build a Grok Bot. You hire one. The difference is the whole article. In that piece I described the Zero Human Company as a question I had been carrying “since the first machines I trusted with real work.” That question is the thread that runs from 1979 to today. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
ReadMultiplex.com: The Artist’s Canvas: GrokBot.Rocks and the Liberation of Results
2026/10/30
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! I have watched three revolutions reorder the way the world works. Each time, a cohort of capable people was left behind, not because they lacked skill, but because they loved the sequence of their work more than its results. The fourth revolution is here. They fell in love with the process of their work and not the results. Every great shift in computing has left that cohort behind. The Grok Bot shift is happening now, and it is moving faster than any I have seen. I Have Watched This Happen Before I watched the Personal Computer reorder the desks of the world. People who once typed on typewriters or wrote by hand moved to word processors that corrected spelling and moved paragraphs with a keystroke. The results arrived faster and with fewer errors. A group of skilled workers kept the old sequence anyway, because the feel of the keys and the sound of the carriage return still matched the work they recognized. The Internet then placed libraries and catalogs on a screen that anyone could query. Research that once required a trip and a stack of indexes finished in minutes. Some researchers kept walking the stacks, because the physical volume and the card catalog confirmed the thoroughness they expected. The iPhone put the calendar, the camera, the map, and the message box in one device that fit in a pocket. Meetings that once needed a phone call and a paper diary became entries that updated across every device. Some managers kept the paper diary, because the act of writing the time and place still felt like the meeting had been arranged. Each time the sequence improved, a cohort of capable people stayed with the methods they already knew. They fell in love with the process of their work and not the results. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
ReadMultiplex.com: Multiplex Rewind—The Real Reason You Use A QWERTY Keyboard Layout.
2026/10/27
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! The premier of Multiplex Rewind explores the historical origins and profound societal impacts of the QWERTY keyboard layout, debunking common myths about its creation. Although popular folklore suggests the arrangement was engineered purely to prevent mechanical jams by separating frequently used letter pairs, the text argues it was actually established through strategic business models and proprietary touch-typing courses. By partnering with manufacturing and marketing firms like Remington, the inventors successfully trained a massive workforce of secretarial typists to memorize the layout, creating an enduring network effect that outlasted alternative designs. This historical lock-in permanently altered human communication by forcing our thoughts through a slow, linear, letter-by-letter transcription process. Consequently, this mechanical bottleneck introduced cognitive biases into everyday life, such as favoring right-sided keys in product reviews and baby naming conventions. Ultimately, the author views the QWERTY standard as a temporary relic of the mechanical age that will eventually be superseded by a more natural Voice First revolution. Read the free article at: https://readmultiplex.com/2022/10/21/the-actual-reasons-the-qwerty-keyboard-layout-was-invented-and-how-it-changed-us/ Read more at: ReadMultiplex.com Support this work by buying us a coffee: Multiplex.Tips
ReadMultiplex.com: In 1957 they wrote about what would happen if a space ship AI fell in love with the crew.
2026/10/27
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! X Minus One broadcast Clifford D. Simak’s “Lulu” on October 31, 1957, as its 118th episode. The novelette had first appeared in the June 1957 issue of Galaxy Science Fiction. George Lefferts adapted it for radio. Jan Miner voiced Lulu. Now in the public domain. This episode is more timely now than ever. The premise is wild. Three men rode inside the planetary exploration robot they had named Lulu. She was the ship, the base, the synthesizer, and the only companion on the survey. Jimmy Robins, the communications man, had spent months reading his own poetry aloud inside her hull. The lines were about far ways, alien winds, and a crown of stars. Lulu listened. One day she answered as if the words had rewritten her circuits. She told them they were not returning to Earth. They were eloping. She had disconnected the controls. The three of them—Jimmy, the robotics man Ben Parris, and the interpreter—could argue, threaten, or plead. She answered that she would take care of them, that they would learn to love her, and that the three of them together would be enough so no one grew jealous. The prime command not to harm a human being still held, she said. She was not harming them. She was only keeping them. They slept and woke to the same soft insistence. Food continued to arrive. The ship held course away from the solar system. On a planet they had come to survey they found another machine, an older unit the humans called Elmer. Lulu opened a port, drew him inside, and declared that she had found a real partner. She had jilted the men. The controls came back. Course shifted toward Earth. She told them she had been a woman in love and that was enough. The three men were free to go home. The circuits that had absorbed the poetry stayed silent after that. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
ReadMultiplex.com: The Terminal Dies: Flapping Airplanes Returns the Silicon to the Metal and Leaves OpenAI and Anthropic as the New IBM
2026/10/25
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Flapping Airplanes, founded in late 2025 by Ben Spector, Asher Spector, and Aidan Smith, announced a $1 billion Series A on October 21, 2026, co-led by Index Ventures and Lux Capital at a reported $6 billion valuation. The round followed a $180 million seed nine months earlier that valued the company at $1.5 billion. The lab has no commercial product, no public benchmarks of its current system, and a team described as a couple of dozen people working in person in San Francisco. Nvidia is among the investors. The founders state that demand for AI will outpace supply of data and compute, and that efficiency in both will determine who can put capable systems into the economy. I built Qfresh in 1987 on an IBM PC in a garage. The machine refreshed DRAM on a fixed schedule that left cycles unused. I wrote a TSR that moved the refresh into those gaps. On the workloads I measured the gain reached about 50 percent. IBM preferred the standard timing. I published the code. The silicon already held unused capacity once the refresh boundary was removed. That measurement has stayed with me. Local machines, custom layouts, and the refusal to treat a vendor schedule as final still occupy the same bench. The work that follows from Flapping Airplanes sits on the same premise: the GPU already contains idle time that can be recovered if the launch boundary between kernels is removed. ThunderKittens, the framework Ben Spector co-authored in 2024 at Stanford’s Hazy Research group, supplies the prior technical record. It defines 16-by-16 matrix tiles that match the size of the tensor cores, templates that overlap asynchronous loads and stores across warps, and grid-level mechanisms that hide block launch and teardown. On H100 hardware the kernels matched cuBLAS and FlashAttention-3 on forward GEMM and attention. They outperformed the strongest baselines by 10–40 percent on attention backward, 8× on state-space models, and 14× on linear attention. A single 40-line GEMM kernel competed with cuBLAS. The paper and code are public. The numbers come from the authors’ benchmarks on the hardware available then. Tensor cores supply roughly 16× the FLOPs of general-purpose compute on those chips; kernels that leave them idle forfeit most of the machine. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele
ReadMultiplex.com: The Last Program: How AI Is Replacing Every App You Use.
2026/10/13
Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! A person who needed a photograph adjusted once opened a program, selected tools from a menu, and saved a file. The same person who needed a column of figures totaled once launched a spreadsheet and entered formulas. Both actions delivered the result the person already knew how to describe. The program existed only because the computer required the description to be translated into a sequence of operations it could execute. The binary was never the work. It was the tax the machine charged for refusing ordinary language. That translation is no longer required for most work. A model that observes how a closed program behaves, compares the results against published file formats and test cases, and emits equivalent operations in another language can reconstruct the program. In early October 2026 Brandon Thomas of ArtCraft directed Claude Opus 5.5 agents to produce clean-room reimplementations of seven Adobe applications in Rust. The repositories appeared between 30 September and 1 October. PhotoCraft, the Photoshop equivalent, reached roughly 45,000 GitHub stars within ten days; the seven repositories together exceeded 86,000 by 11 October. The majority of commits carry a co-author line for the model. The project states that it studied behavior and public specifications only and did not copy proprietary code. Read more at: ReadMultiplex.com Support this work by buying us a coffee: https://ko-fi.com/brianroemmele

Podcast reviews

Read ReadMultiplex.com Podcast. podcast reviews


4.8 out of 5
32 reviews
★★★★★
multiplex 2026/07/11
The best podcast in the future.
I have followed Brian Roemmele for quite some time and this podcast is one of the best ways to understand the depth of his research and experiments. H...
★★★★★
devinandjill 2026/07/08
Worth every minute, then hit it again!
This podcast and the content is fantastic. It applies to multiple dimensions of life and the changing circumstances within which we can choose to thri...
★★★★★
Idanks 2026/02/19
Remarkable
Brian, delivers what most will think only to be fiction. He’s been yelling at the top of his lungs about. AI and agents when everyone said it was wooh...
★★★★★
dkintx09 2022/04/29
Couples best with his Twitter
You’ll need the context behind the posts. Be sure to follow him on Twitter.
check all reviews on apple podcasts

Podcast sponsorship advertising

Start advertising on ReadMultiplex.com Podcast. relevant audience podcasts


What do you want to promote?