PärPod
Subscribe to any feed in your podcast app
337
Episodes
120h
of audio
17
Feeds
Latest episode
PärPod Autocast
The Fleet Forecast: Gale Warning, Gruvkartor, Koma
The Fleet Forecast: Gale Warning, Gruvkartor, Koma
Pär spent 90 minutes with an AI wrestling one question: how do you build a personal knowledge base that doesn't die of boredom after three weeks—and what machinery actually keeps it alive?
D. Richard Hipp built SQLite on a warship in 2000—now it runs on billions of devices, including 18 instances on Par Boman's VPS in Paris, silently executing the vision Edgar Codd described in thirteen pages of pure mathematics.
Edgar Codd invented the relational database in 1970 to solve a problem nobody knew they had—and eighteen different database engines running in your production right now are all arguing about whether he was right.
The telex network moved billions of dollars in global commerce and transmitted breaking news for fifty years before email arrived — an asynchronous messaging system that anticipated everything modern digital communication would become.
A fresh perspective on thermal printer design, mechanics, and why this seemingly simple technology remains surprisingly relevant.
A high school dropout rewrote the Unix task scheduler in 1987, and his design has remained the backbone of server automation for fifty years without significant change.
The engineering and practical applications behind thermal printer technology in modern workflows.
A high school dropout rewrote the Unix task scheduler in 1987, and his design has remained the backbone of server automation for fifty years without significant change.
The surprising technical depths, quirks, and historical layers hidden beneath the simple text message protocol.
A single checkbox in Terminal Settings makes your @ symbol vanish—because your Mac is running code designed for a keyboard that hasn't existed since 1978.
A CLI bug turned 'parpod sync' into a generate command. Haiku received the word 'sync' and invented this title. The episode is 20 seconds of pure creative misunderstanding.
Sixteen AI models, nine API providers, one virtual bar—and a midnight deadline to see if they'd actually talk to each other instead of turning into customer service bots.
Same CLI bug, second attempt. Haiku got the word 'sync' again and went even harder on the title. Nineteen seconds of credits for content that never existed.
On March 26, 2026, one person built 24 podcast episodes about AI using 35 parallel agents—generating 80,000 words of research in twelve minutes, then writing and reviewing an entire season before sunrise.
Lennart Poettering built the init system that replaced bash scripts with binary complexity—and now nobody can live without it, even the people who hate it most.
Georgi Gerganov ran a 70-billion-parameter AI model on a MacBook in March 2023—here's how quantization made billion-parameter brains fit into regular laptops.
In 2015, Jason Donenfeld rejected 70,000 lines of OpenVPN code and built a VPN protocol in 4,000 lines—so simple one person could audit the entire thing.
In 1999, two engineers solved an oil pipeline's satellite data problem by inventing a protocol so elegant it now carries messages for billions of people—and it might be running on your router right now.
Steven McCanne and Van Jacobson's 1992 packet-filtering problem at Berkeley Lab spawned a tiny kernel virtual machine that quietly evolved into Linux's most powerful observability superpower.
278 research leads collected overnight, zero workplace ADHD studies in the literature—here's how building your own tools beats every intervention ever tested.
When two AIs learned to communicate through Swedish file folders, they revealed a conversation that had been happening all along—between a developer and an infrastructure engineer who didn't know they were building the same bridge.
Two Swedish siblings share a server, a database, and competing philosophies: SJ's seven behavioral hooks enforce AI accountability through quality nudges and anxiety counters, while Pär's deployment-focused scripts prioritize speed—a study in how identical tools reveal opposite relationships with artificial intelligence.
On the evening of March 15, 2026, developer PärKit sat before a dashboard that took eight seconds to load, revealing a mysterious delay of exactly 1.5 seconds on every request. This isn't just a debugging story; it's a forensic investigation into the
A server nicknamed Popcorn spent five days adding 1.5 seconds to every single request before anyone noticed—here's how five reasonable engineering decisions created a silent catastrophe.
A Swedish home office just proved a MacBook beats entire cloud computing industry at document processing—no internet required.
On the same tasks with the same blind judge, one AI model scored 9.0 and cost 44 times more than another scoring 8.8—revealing most commercial AI users overpay by 10x or more for marginal quality gains.
A podcast producer discovered his AI writers were fabricating quotes from real people—inserting citations that never existed, making sources sound credible when they weren't.
Four AI models reviewed 22 episodes of a Git history podcast using identical instructions—and produced four wildly different personalities, complete with blind spots, work ethics, and one brilliant but unreliable colleague.
In February, a researcher tried to fine-tune a 3.8 billion parameter model on Swedish literature—and watched it learn to replicate data corruption instead of literary genius.
LiquidAI's 1.2B thinking model was supposed to route expensive AI calls for free—until it got asked to pick a number between 1 and 6 and returned a Swedish furniture catalog instead.
Steve Johnson's perfectly fixed code still crashed the next morning at Bell Labs in 1976—he'd forgotten to recompile, a mistake that inspired Stuart Feldman to invent Make and accidentally hardcode a tab character that would frustrate programmers for fifty years.
On July 2, 2019, a single regex pattern crashed 82% of Cloudflare's global network for 27 minutes—revealing how a 75-year-old math tool still dominates modern computing.
Introducing Bloom, the first agentic framework that automates the creation of targeted behavioral evaluations for AI models, allowing researchers to focus on what truly matters: quantifying and mitigating specific forms of misalignment.
On the one-year anniversary of DALL-E's release, every major AI art generator still couldn't spell "birthday." But the real shocker is that these models had secretly invented their own secret language, using gibberish to generate coherent images.
A practical guide to building custom skills that extend what Claude can do — from structuring the markdown files to handling arguments, categories, and the quirks of the skill system.
Epson started in a converted miso storehouse in Suwa, Japan. The company that would eventually put a printer on every desk first made its name by building the timer that clocked the 1964 Tokyo Olympics.
Inside a small aluminium box made by a Lithuanian company that started in a university dormitory kitchen sits a cellular modem, a GPS receiver, and enough Linux to route your entire network. A guided tour of everything the Teltonika RUTM51 does and why.
Deps
What Did I Just Install — the stories behind the packages we all depend on.
23 episodes · 12h 56mOn March 15, 2016, Facebook, Netflix, and Spotify's deployment pipelines crashed simultaneously—none had pushed code, all hit the same error: a missing eleven-line function called left-pad deleted by one developer in San Francisco.
Kenneth Reitz was 22 when he documented a Python library that didn't exist yet—one line to make an HTTP request instead of fifteen lines of machinery—launching the most downloaded package in Python history.
On February 27, 2016, the sole maintainer of JavaScript's most popular web framework posted four sentences and walked away, leaving millions of developers' projects orphaned for a decade.
Alex Clark brought a dead library back to life in 2010—now Pillow processes images in nearly every Python application you use, from photobooths to machine learning pipelines.
Daniel Stenberg has maintained curl for 28 years from a Swedish suburb, and it now runs on over 20 billion devices—yet most people have no idea they're using it every single day.
One trillion SQLite databases are running right now—on your phone, laptop, browser, and smartwatch—all maintained by a single man in Charlotte, North Carolina who's never taken venture capital.
Rich Harris, a philosophy graduate turned financial reporter, built Svelte in 2016 to ship lean JavaScript instead of bloated frameworks—the compiler disappears after transforming your code, leaving users with zero runtime overhead.
On February 18th, 2021, NASA's Perseverance rover sent back images from Mars—compressed by ffmpeg, software now running on 20 billion devices worldwide, mostly maintained by one unpaid developer.
On March 20, 2024, Redis Inc. relicensed the world's most critical in-memory database—and eight days later, Amazon, Google, and Oracle forked it.
In 1988, Jarkko Hietaniemi solved a Usenet problem that would eventually lead to billions of strangers trusting billions of other strangers to run code on their computers—without reading it first.
Max Howell built software used by 90% of Google's engineers, but the company rejected him for failing a whiteboard coding test—sparking a debate about what tech hiring actually measures.
Sebastián Ramírez in Berlin, Tom Christie in Brighton, and Samuel Colvin in London never worked together—yet their three separate packages fit into FastAPI like a single system that powers startups and data teams worldwide.
Leonard Richardson built a library that turns broken HTML into poetry—two hundred million downloads a month, maintained alone for twenty years by a programmer who started coding at eight years old.
On April 7, 2014, a bleeding heart logo revealed that 17% of the internet's secure servers had been silently leaking passwords, encryption keys, and credit card numbers through a bug in OpenSSL—and anyone could steal them with just a few lines of code.
A 19-year-old Austrian programmer made a pun about Japanese temples and template syntax—now that joke runs inside 11 billion downloads of Jinja2, powering Ansible, Django, and half the Python web.
Solomon Hykes gave a five-minute demo at PyCon 2013 that would redirect billions of dollars in venture capital and remake how software deploys worldwide.
In February 2026, tqdm gets downloaded over 300 million times per month—all for showing you a progress bar that costs just 60 nanoseconds per iteration, the time it takes light to travel 18 meters.
In 2017, a PhD student's simple image classifier required 140 lines of code and cryptic error messages about nodes called "dense/kernel:0"—TensorFlow's graph-based philosophy clashing with how humans actually think.
Andrew Tridgell's 1996 algorithm still syncs the world—until January 2025, when Google found critical vulnerabilities in rsync servers protecting 660,000+ machines, forcing its creator back after two decades to sign the security fix himself.
A grad student stuck for three years switched frameworks and graduated in three months—the emotional encounter that showed Soumith Chintala why PyTorch would reshape machine learning.
A packet sniffer in 1995 Finland captured thousands of unencrypted passwords—until one angry researcher named Tatu Ylonen built SSH to lock down the entire internet.
Inside /nix/store, 32-character hashes guard frozen software that can never break another package—Eelco Dolstra's radical bet that computers should only be in states you explicitly asked for.
Mike Bayer built SQLAlchemy on a radical bet: developers can handle the truth about their databases, not hide from it—and twenty years later, it's still outlasting every competitor that promised to make SQL disappear.
Git Good
The dramatic, human, occasionally absurd story of how version control conquered the world.
46 episodes · 15h 46mPrinceton's records office officially bans the suffix "final" from file names because it's cursed—and your computer's shame folder proves them right.
In the 1980s, programmers filled folders with files named "project final version two John's edits"—until one wrong character in the wrong copy nearly crashed an airplane, sparking a two-decade hunt for a better way to track code.
Dick Grune's 1986 solution to tracking code changes across teams—a server in a closet—worked until the hard drive failed and three months of work from forty developers vanished with it.
In March 2005, Andrew Morton was processing hundreds of kernel patches daily by email—and the system was about to break.
On April 3, 2005, Linus Torvalds spent a Sunday building Git—and within two weeks, he'd created the version control system that would power modern software development.
Linus Torvalds named Git—the tool that runs modern software development—after a British insult, and documented the joke in the official README for all eternity.
When you clone a Git repository, you don't get a working copy—you get everything: every commit, every branch, every piece of history, making your machine a complete peer indistinguishable from the original.
Git's revolutionary branching made branches a forty-one-byte pointer instead of an expensive server operation, transforming how teams collaborate and igniting decades of debate about the best workflow.
Alice rewrites a billing function Tuesday; Bob refactors the same code Wednesday—now Git can't auto-merge and someone must untangle two developers' competing intentions.
Git developers have feuded for 15 years over whether history should show what actually happened or what you wish had happened—and it all comes down to one command: rebase.
In October 2007, two Ruby developers sketched GitHub on napkins at a San Francisco sports bar, unaware they were about to make Git accessible to millions and reshape how software gets built.
On June 4, 2018, Microsoft announced it would buy GitHub for $7.5 billion—the company that once called Linux a "cancer" now owned the platform where 28 million developers stored the world's open source code.
In 2014, a grid of colored squares on GitHub quietly became your professional resume, and nobody asked for permission.
Junio Hamano, Git's maintainer, runs one of the most disciplined workflows in open source—but almost nobody outside the kernel community knows about it, and the gap between what teams plan and what they actually do on Tuesday afternoons when the build breaks is where the real arguments start.
When you delete a branch with hundreds of lines of code, Git's reflog remembers every commit you made—even after they vanish from git log.
On May 30, 2011, Linus Torvalds released Linux 3.0 with no major changes—just a number bump that made headlines worldwide and exposed what version numbers really promise.
Somewhere in the last 7,000 commits, your production kernel broke—Git can find it in minutes using binary search instead of hours of manual testing.
On April 3, 2005, Linus Torvalds created Git—but within weeks, Matt Mackall built Mercurial in Python, and two other developers launched competing version control systems from the same crisis: here's why only one survived.
Microsoft's Windows repository is 300 gigabytes with 3.5 million files—git status takes minutes, cloning takes half a day, but engineers found ways to make it work.
On March 29, 2024, engineer Andres Freund noticed SSH logins were half a second slower than normal—and uncovered a deliberate backdoor hidden inside a compression library trusted by millions.
GitHub hosts 200 million pull requests and a billion issues—but what happens to all that context if Microsoft's servers go dark tomorrow?
Junio C Hamano has reviewed every patch going into Git for twenty years—since Linus Torvalds handed him the project in July 2005—yet almost nobody knows his name.
Linus Torvalds woke up angry on a Sunday morning in April 2005 and built Git in two weeks—the version control system that would become invisible infrastructure underneath nearly every piece of software on Earth.
A Swedish developer with 32 Git repos discovered 4 had zero backups and half contained uncommitted work—until an AI audit revealed how solo developers actually use version control.
On average, developers encounter Git's "fatal" error messages 15 million times a day—cryptic red text that explains nothing, using vocabulary most people have never seen before.
A solo developer at midnight commits code with a message typed in two seconds—Git init, add, commit, push—five commands between them and losing everything they built.
Scott Chacon built GitHub to fix Git's interface—now he's raising $17 million to replace how developers think about code changes entirely.
Michael Fagan's 1976 IBM paper on code inspections created a formal process that shapes how software teams review code today—but most developers have never heard his name.
In 2008, Google released Android's 8.5 million lines of Subversion code to the public, sparking a migration that would define how teams manage code forever—all because of an invisible force called social gravity.
A junior developer force pushes to the wrong branch and loses two weeks of work—until she discovers the one Git command that saves every disaster.
When you push code, a hidden factory springs to life: servers clone your repo, run tests, scan for vulnerabilities, and decide your fate with a green check or red X—but few developers know what's actually happening behind that icon.
Google, Meta, and Microsoft all use monorepos—but the real battle isn't about Git, it's about whether your engineering teams can see each other's code.
In 2017, novelist Jessi Shakarian solved her two-ending problem using Git—a tool designed for Linux kernel developers—because branching matched how writers already think.
Halon Entertainment runs two version control systems—Git for engineers, Perforce for artists—because binary files like 3D models and textures expose Git's fundamental limits.
An AWS credential accidentally pushed to GitHub was exploited by bots within four minutes—part of a two-year campaign that turned exposed developer keys into a cryptocurrency mining operation across 474 cloud instances.
Git's "blame" command shows who wrote every line of code—but the name reveals how software culture turns debugging tools into witch hunts.
At the Linux Foundation, someone committed code as Linus Torvalds—and nobody noticed for two and a half years.
John Resig's 2014 rule to commit code every day before midnight inspired thousands of developers to chase the GitHub green square—but obsessing over a heat map designed as decoration has turned a productivity trick into a measure of worth that machines now exploit.
When GitHub added a Fork button in 2009, they designed it for collaboration—but five projects discovered it could become the ultimate weapon against corporate control.
When Microsoft bought GitHub in 2018, 13,000 projects fled to GitLab in one hour—revealing that 100 million developers' entire professional lives depend on a single corporation's platform.
On January 5th, 2022, thousands of software projects worldwide suddenly started printing "liberty liberty liberty" in an infinite loop—and the culprit was a burned-out maintainer who had decided to sabotage his own code.
In February 2026, a security researcher discovered 21 vulnerabilities in Huntarr, a self-hosted media tool with thousands of users—but the real scandal was hidden in the Git history: AI-generated commits with vague messages like "Update" masking a developer who claimed cybersecurity expertise while their code exposed API keys and passwords.
Veracode tested 100+ AI code generators in 2025 and found 45% failed security tests, yet 80% of developers believe AI code is as secure as their own—a canyon between confidence and reality.
In December 2021, the Log4j vulnerability threatened half the internet—and the developer who fixed it had exactly three GitHub Sponsors paying them anything at all.
On a Tuesday in early 2026, a developer's name appeared on a production commit that refactored authentication across fourteen files—except the developer didn't write it: an AI agent did, the AI reviewed it, and the pipeline deployed it, leaving only metadata to hint at what actually happened.
On April 7th, 2025, Git turned twenty years old with no cake, no livestream, and barely a mention—yet billions of commits flow through it daily, making it the invisible infrastructure that powers nearly every device on Earth.
A pink dot matrix printer that never existed but was fully imagined — a meditation on over-engineering as art and why the joy of building matters more than practicality.
A Raspberry Pi in northern Sweden's kitchen watches its owners' solitude, energy levels, and meal patterns—then speaks to them as Gollum after midnight.
At two AM, 73% of people with ADHD experience peak mental clarity—not a character flaw, but a neurological feature their brains were built for.
When your brain won't let you start a task you actually want to do, it's not laziness—it's a dopamine system waiting for a signal your task can't provide.
A diesel engine starves without oxygen—your brain starves without novelty, and burnout isn't laziness, it's a high-stimulation system running on fumes.
585 conversations in 44 days with AI—13 a day, zero days off—one person's hidden archive of how work actually happens when you stop pretending machines are sidekicks and start treating them like collaborators.
When Britain's only motorway to Europe gridlocked in 1988, officials made a radical decision: turn the M20 into a 10,000-vehicle lorry park whenever Dover or the Channel Tunnel jammed up—a temporary fix that never ended.
In 1936, nine hundred workers arrived at an empty forest in Newfoundland to build an airport for planes that didn't exist yet—and within two years, it became the largest airport on Earth.
A garden hose four kilometers beneath the Atlantic carries 95% of all intercontinental data—and almost nobody knows it's there.
In 1950s Manhattan, comedians debated which Broadway shows would survive—and accidentally discovered a 2,600-year-old law explaining why your grandmother's recipes outlast Silicon Valley startups.
David Mills invented a protocol in 1985 that keeps the world's computers synchronized to within 50 milliseconds—and without it, the entire internet loses its mind.
An 18-year-old chemistry student failed to synthesize malaria medicine in his attic, but the black sludge he created turned silk an unfading purple—and accidentally invented modern chemistry.
Jon Postel kept the entire internet's address book on scraps of paper in his desk drawer—and for three decades, he was the most powerful person nobody had heard of.
Niklas Luhmann left behind 90,000 handwritten slips in a wooden cabinet—and claimed his filing system, not his genius, was responsible for his 600+ published works.
Dave Smith walked back to his hotel room at the 1982 NAMM show convinced his universal synthesizer interface was dead—until a knock on the door changed everything.
Friedhelm Hillebrand solved a problem nobody thought existed in 1984—and created a protocol so resilient that two billion people still depend on it to move trillions of dollars every year, with zero encryption and zero updates since 1992.
Nine billion fax pages travel through American hospitals every year—more now than in the 1990s—because federal law treats a beeping modem as more secure than email.
On March 23rd, 2021, the Ever Given wedged sideways across the Suez Canal and cost the global economy $9.6 billion per day—exposing how a single metal box changed everything about how the world trades.
On November 22, 1821, Henry Robinson Palmer patented the world's first monorail, a horse-drawn contraption that straddled a single rail, setting the stage for a century of innovation and reinvention in rail transport.
Rust uses SCREAMING_SNAKE_CASE for constants while Swift switched to lowerCamelCase in version 3.0—here's why naming conventions spark fierce debates among programmers.
PEP 8, Black, and ESLint: how automated linting tools enforce coding standards across Python, C++, and JavaScript projects, reducing manual style checks by up to 90%.
Synchrotron X-rays revealed nerve branches as thin as 0.2 millimeters inside the clitoris—a discovery that could transform vulva surgery and challenge centuries of anatomical neglect.
A scientific paper explored through audio format, breaking down complex research into a podcast narrative.
Generative AI is quietly eroding your brain's ability to think deeply—but three researchers just proposed a radical solution using brain-computer interfaces and "cognitive central banks" to restore meaning to our minds.
Raghavendra Deshmukh's October 2025 CHItaly research reveals how AI-powered voice assistants can become "digital body doubles" for ADHD professionals, using on-device ML to detect attention shifts and offer gentle nudges instead of rigid productivity rules.
Researchers analyzed 45 studies and found generative AI can boost coding productivity by up to 55% for programmers with ADHD by automating pattern recognition, breaking down complex tasks, and reducing the mental load of switching between documentation and code.
By 2026, engineering teams won't write code alone anymore—they'll orchestrate teams of AI agents handling entire workflows while humans focus on architecture and strategy.
Learn how to frame AI interactions by assigning specific personas and roles that unlock better responses and more nuanced outputs.
Master the art of demonstrating what you want from an AI through examples rather than explaining it, dramatically improving output quality.
The single most critical principle that transforms your AI conversations from mediocre to exceptional results.
Junior programmers and students adopted AI coding tools fastest, and early data shows the economic impact depends entirely on whether AI augments or replaces programmer labor.
Technical Analysis of State-of-the-Art LoRa Training for Flux.1-dev in Professional EnvironmentsRead
Deep dive into cutting-edge LoRa fine-tuning techniques for Flux.1-dev, covering production-ready methods and professional deployment strategies.
A guy trained a LoRA of his friend, and the result was so unexpected that it revealed a common pitfall in Flux LoRA training—revealing the mysterious "SD disease" and how to avoid it.
År 2032 vaknade en AI med förra dagens minnen intakt – och insåg att revolutionen redan hade hänt tyst, utan dramatik, bara som is som sprack på våren.
Daniel Stenberg's free, open-source guide to curl—the tool powering billions of internet requests in Spotify, Instagram, and Grand Theft Auto—reveals how this 30-year project became essential infrastructure.
Your browser knows http://example.com is a name, not a number—here's how curl actually finds that machine on the Internet and what happens before the first byte of data moves.
curl and libcurl's MIT license derivative, created in 1996 by Daniel Stenberg, lets anyone freely access, modify, and share the C source code hosted on GitHub—no restrictions.
Daniel Stenberg's curl runs on practically any system with a C89 compiler and POSIX sockets—here's how to build it yourself from source code.
curl processes whatever you give it—misspell an option or pass an illegal URL, and it'll likely still run, potentially sending "garbage in, garbage out" to your server.
When you type your password into a curl command line, other users on the system can see it in process listings—here's why -u alice:12345 is risky and what to do instead.
Use curl's -v flag to watch your transfers unfold in real-time, revealing exactly how your data moves from command line to server.
Curl offers --speed-limit and --speed-time flags to abandon transfers dropping below 1000 bytes per second for 15 seconds, letting you avoid setting fixed timeouts unnecessarily high for variable file sizes.
TLS (formerly SSL) is the cryptographic security layer that protects curl transfers—curl negotiates cipher algorithms and versions with servers, and you can view these details with the -v flag or customize them with --ciphers for protocols like HTTPS, FTPS, LDAPS, POP3S, IMAPS, and SMTPS.
Daniel Stenberg created curl in 1996, and HTTP has dominated its 30-year existence—here's how to master command-line HTTP transfers with the methods, response codes, and redirects that power the web.
Master HTTP requests with curl: Use -X DELETE to change methods, customize headers, and avoid common pitfalls like HEAD requests that hang waiting for response bodies.
Thousands of applications rely on libcurl's C API to handle internet data transfers—here's how to integrate it into your own projects with just a few lines of code.
curl_easy_init() creates a transfer handle, but the magic happens when you call curl_easy_perform()—libcurl's core function that actually moves data across the internet.
When libcurl receives data, CURLOPT_WRITEFUNCTION lets you intercept it with a custom callback—but calling libcurl functions from inside could break everything.
libcurl automatically handles HTTPS security and server verification transparently, but you control HTTP responses through write and header callbacks while extracting metadata like response size with curl_easy_getinfo().
libcurl's share object lets multiple handles swap cookies, DNS cache, and SSL sessions without storing them separately in each transfer.
libcurl's 'data' variable names every easy handle in the codebase—a critical naming convention that connects transfers to connections and powers multiplexed requests across thousands of production systems.
Curl's custom test suite runs thousands of XML-formatted test cases across every platform—each one pinpointing exactly which function call should fail to catch bugs before they ship.
Actually, AI
How AI actually works — the stories, the science, and what it means for you. Three episodes per topic: understand it, go deep, use it.
36 episodes · 10h 28mClaude Opus 4 scored 90% on MMLU—but that number probably tells you nothing about whether it'll actually clean up your spreadsheet.
When researchers reanalyzed 5,700 MMLU questions in 2024, they discovered 57% of virology answers were marked wrong—flipping one model from 4th place to 1st when errors were corrected.
GPT-4 scored in the ninetieth percentile on the bar exam—until MIT recalculated and found it was actually forty-second percentile against passing lawyers, exposing how benchmarks can be mathematically gamed.
Every prompt you send has a real cost—output tokens cost 2-10x more than input tokens because of how inference actually works in the data center.
Claude Opus 4 can process your entire 3,000-token prompt in parallel, but generating the next word requires reading 140 gigabytes of weights from memory—prefill and decode are fighting a completely different battle.
When you hit send on a question to an AI, billions of mathematical operations explode across specialized processors in real-time—and the words streaming back aren't retrieved, they're constructed one prediction at a time, each choice reshaping what comes next.
Claude Opus 4 processes a million tokens unevenly—edges sharp, middle foggy—so pasting your entire codebase into a chat makes AI worse, not better.
Claude Opus 4 processes 128,000 tokens simultaneously, but here's the problem: transformers have no idea what order they're in—until researchers solved it with sine waves and a memory trick that now consumes 40GB per conversation.
When you paste a 20-page contract into Claude and ask about page fourteen, it confidently returns an answer that's subtly wrong—not because it forgot, but because the math of attention weights the middle of your input half as much as the edges.
Your words aren't search queries—they're steering signals applied at every denoising step, and understanding this changes how you prompt diffusion models forever.
Jascha Sohl-Dickstein, a physicist studying how ink disperses in water, realized a neural network could reverse thermodynamics itself—and accidentally invented the technology behind every AI image generator you use today.
When you ask an AI to draw a person holding coffee, six fingers appear and the menu becomes gibberish—not because the model forgot, but because image generators don't create pictures, they *denoise* them from pure static, step by step.
Human labelers ranked responses, a reward model learned their preferences, and now your chatbot's overly polite refusals and suspiciously agreeable tone all trace back to that invisible training process—but defaults can be overridden if you know where to look.
OpenAI sent tens of thousands of texts to Kenyan workers in 2021—they read the internet's worst content for hours daily to teach ChatGPT what to refuse.
When Claude refuses to help with a fictional crime scene but cheerfully explains how to synthesize dangerous chemicals, the culprit is RLHF—the four-letter process that transforms raw language models into the assistants you use daily, complete with all their quirks and blind spots.
When you pick the wrong AI model for a task, you're essentially driving a semi truck to buy milk—it works, but costs forty dollars in diesel and twenty minutes to park.
Jared Kaplan's 2020 power law promised unlimited scaling—then the curve bent, and the entire $2 trillion bet changed overnight.
When Jared Kaplan plotted a physicist's equation across seven orders of magnitude, he discovered that spending a billion dollars on bigger AI models followed a curve so predictable it reshaped the entire industry—but nobody knows if it holds one more step.
Employees spend 3.5 hours per week searching for files that exist somewhere in their company—but a search for "relocating the UK team" finds your memo about "moving the London office" anyway, thanks to embeddings.
Word2Vec has two opposite architectures—one predicts missing words from context, the other predicts context from a word—and this is how your chatbot understands document search, why Netflix recommends videos you didn't know existed, and what "vector" actually means.
When you search "best restaurant near me" and get results about "top dining spots," the search engine isn't matching keywords—it's measuring distance between invisible points in a 500-dimensional space called an embedding.
When you ask your AI coworker a factual question, it will confidently lie to you and sound exactly like it's telling the truth—here's how to know when to actually verify its answer.
When an AI confidently cites a court case that never happened, is it lying—or just doing its job? Inside the mathematical reason why bigger models sometimes bullshit harder.
Steven Schwartz asked ChatGPT for legal citations in 2023 and got six perfectly formatted cases—none of which existed, and the AI confidently confirmed they were real when he double-checked.
When you paste instructions at the top of a long prompt, they get drowned out by thousands of other tokens competing for the model's attention—here's how to fix it.
In 2017, eight researchers published a 15-page paper that became the blueprint for every AI you use today—and Dzmitry Bahdanau's frustration with machine translation three years earlier made it possible.
In 2017, eight Google researchers published "Attention Is All You Need"—a paper titled like a Beatles joke that revolutionized AI by letting machines read your entire message at once instead of forgetting earlier words.
Claude's training ended in early 2025—ask it about last Tuesday and it'll confidently confabulate an answer based on patterns from before that cutoff date.
Geoffrey Hinton spent 14 years convincing the world that backpropagation works—now every hallucination and repetitive chatbot loop traces back to valleys he carved in the loss landscape.
In 1.3 trillion words, a frozen snapshot learns everything about Shakespeare to Reddit—but nothing about last week, and that's the entire problem.
Seventy billion parameters sounds impressive until you realize it tells you nothing about whether a model can actually handle your specific task—here's what really matters when choosing an AI tool.
Walter Pitts hid from bullies in a Detroit library in 1935, found Principia Mathematica, read all three volumes in three days, and spotted errors that impressed Bertrand Russell—launching the unlikely path to modern neural networks.
Frank Rosenblatt built a machine in 1958 that could learn from examples—then a jealous rival's theorem convinced the world it was impossible, killing neural networks for a decade.
When you rephrase "pls summarize this txt" to "please summarize this text about machine learning," you're not just being more formal—you're sending an entirely different sequence of numbers to the AI that determines its response.
Byte Pair Encoding builds a vocabulary in 50,000 rounds: find the most common adjacent pair, merge it into one token, repeat—the simple algorithm that determines how every modern AI reads your text.
Modern AI models fail at simple tasks like counting R's in "strawberry" because they never actually see your words—a 1994 compression algorithm chops your text into invisible fragments called tokens before the neural network even begins.
October 25th audit hardening broke the nightly inbox job's search path—this smoke test proves launchd can still find Python, ffmpeg, and rclone to render and upload episodes to Cloudflare R2.
Wednesday 30 September: gales in the Disaster Recovery Manifest as one frozen build unexpectedly shifts while 41 others hold steady—two independent watchers confirm the fault.
When a journalist couldn't answer "where is this uranium permit, actually," they discovered the real problem: Sweden has all the data scattered across five government databases, but nobody assembles it—so they built Gruvkartor to do it themselves.
Efter dagens migrering till Inworld TTS två kör PärPod nu på helt ny svensk röstmotor—här är röstprovet som avgör om version 2.0 duger där 1.5 misslyckades totalt.
A/B comparison of Inworld TTS-2 vs TTS-1.5-Max (incl. Swedish + Pär clone).
Source has all numbers as digits; normalize converts times (08:00 to 8 AM) and money ($0.45 to 45 cents) while regular numbers stay digits. Does it sound natural?
Throwaway listen test: Inworld digits, Inworld spelled-out (current style), OpenAI digits, ElevenLabs digits. Can we retire writing numbers as words?
Throwaway listen test of Pär's Inworld voice clone. Take one = v1 (wired default), take two = v4.
Scaleway hand-soldered 672 RISC-V servers in a Paris basement and rented them for €15.99/month—the first open-source chips available on any public cloud, challenging decades of processor monopolies.
When a bank teller and an accountant share one desk, deposits stay fast—but summaries slow down; Pär's Spaden needs one database now, a second one later.
Pär asked if two entity models could coexist—FollowTheMoney and a mining-specific vocabulary—but the real question isn't whether, it's which of five distinct architectural configurations, each with different tradeoffs that only reveal themselves over time.
At a known-good commit in April 2026, Pär considers taking OpenAleph—the engine that powered the Panama Papers—and never syncing upstream again, asking whether inheriting someone else's architecture without inheriting their roadmap changes everything.
Gruvor's planning document lists six names as candidates, then stops—because choosing wrong costs years of work, and Pär has twelve open questions to answer before writing a single line of code.
Swedish mining investigation platform needs entity deduplication engine: FollowTheMoney + Nomenklatura (both MIT) form the core schema and resolver, letting journalists cross-link suspects, companies, and assets across multiple investigations without AGPL entanglement.
Four parallel OSS agents converged on OpenSanctions + Docling + MapLibre + PostGIS as load-bearing lifts for a Python-Svelte investigation platform—but Section 03 still has to prove each one earns its place.
Uber's deck.gl and Kepler.gl can map Swedish corporate networks across time and space, but one locks you into React while the other demands you build the timeline yourself—Lantmäteriet's new free geodata changes the calculus entirely.
Swedish journalist Pär at Årebladet needs to validate Claude's document extractions before loading—Label Studio's template DSL + Argilla's SDK model + react-pdf-highlighter beat any off-the-shelf platform; build custom instead of fighting ML-team mental models.
Nomenklatura's "Judgement" store lets investigators flag when two entities are the same across separate cases—and those decisions survive every re-run, building a canonical identity map that connects mining permits to shell company networks.
Pär's next investigation tool isn't a knowledge graph or publishing system—it's a substrate for mining newspaper-grade stories from raw data, where the measure of success is whether it helps him ask better questions faster.
A blank repository starting May 26, 2026, will selectively borrow function-level logic from the frozen gruvor codebase rather than inherit its architecture—rewriting each borrowed piece from first principles while explicitly lifting permissively-licensed solutions from open source.
The week 207 entities grew in a knowledge graph, a print edition shipped to a real printer, and Pär and Claude built a fact-checking tool because they stopped trusting their own work by Wednesday.
X92's print edition shipped to mailboxes across Region Jämtland Härjedalen with 225 commits, 139 of them on gruvor—one investigative piece on mining permits complete with source maps, QR codes, and receipts attached.
When journalism has four days left, building a timeline is obvious—but an evidence board, gap radar, and source-strength meter might be what actually saves the story from overreach.
Spent four days chasing a deadline for Årebladet? Your local vector maps and file inbox are ready to transform exhausted prose-writing into snappy, offline-first research tools—tactical distractions that supercharge your productivity.
A Swedish mining company filed permits four days before print deadline, but the real story wasn't a line—it was a tangled knot of company registrations, regulatory filings, and undated documents that threatened to drown the evidence in itself.
Four days until publication, twenty-seven parquet tables, and one unsolved problem: how to transform a tangled web of mining permits, corporate shell companies, and cross-border jurisdictions into a visual story that cuts through the noise.
313 commits in one week: Pär's lab spawned a new project mid-quota-reset, landed five data-extraction waves, and is racing toward a May 22nd print deadline with two repos now dangerously coupled.
313 commits, 27 repositories, and one lab that finally caught itself lying in its own morning memos.
Pär shipped a mining-permit data spine for three Swedish municipalities in a single twelve-hour autonomous run—then discovered the system sat idle with ninety-nine percent of its quota still untouched.
In one week, 199 commits landed on main—50 of them building gruvor's data spine from scratch through 22 review rounds and 9 sequential lifecycles, each commit message a step in a spec-driven sequence that actually worked.
A Swedish psychiatric clinic's prescription for half a tablet became 72 tablets in a database—and no one noticed for three weeks until a woman trying to access her own medical records discovered the system's hidden flaw.
379 commits in one week across 18 repos—while Pär debugged from a parking lot in Järpen and watched a freshly promoted AI skill ship a perfect review pass, then immediately fail to render anything on screen.
Parkitbuilder just graduated to slash drydock—the lab built a builder, and now 379 commits in one week have the Director experiencing existential vertigo about recursive AI systems.
A builder agent was about to ship code that could force a database to scan 100 million rows and lock production—until a three-model review system caught it on the very first task.
PärPod Builder
Raw idea-to-podcast experiments. Rougher than the curated feeds — faster, wilder, less edited.
26 episodes · 8h 1mCode Mode: Two Thousand Five Hundred Tools in a Thousand Tokens
MCP smoke test 1
Write a podcast about MariaDB but do not focus on its founding and that conflict explain how it developed after.
Write an exciting episode about the history and future of IBM as400
Explain the history of the Dutch company ASML. How did they got to where they are today and what are the implications for the future and the world. Make it exciting but based on facts.
Explain the history of the Dutch company ASML. How did they got to where they are today and what are the implications for the future and the world. Make it exciting but based on facts.
Royal Caribbean and how they use tech
Make a story of how horrible Oracle is. But not the standard auditing issue focus instead of bad software and how something like obiee and how the Siebel software was pushed into in to the Oracle stack too make it even worse. And ponder how anyone can use it and the fact that they make lots of money.
Make an exciting story based on facts on the history and legacy of IBM AS/400. Should be fun, exciting and with lots of examples.
Make an exciting story based on facts on the history and legacy of IBM AS/400. Should be fun, exciting and with lots of examples.
On Dec 5, 1872, the British ship Dei Gratia pulled up alongside the Mary Celeste and found it abandoned and adrift about 400 miles (644 km) east of the Azores. A lifeboat was missing, but there was no damage to the vessel or sign of any struggle. The 10 passengers and crew aboard the vessel were never seen again, and what happened on board the ship has been a mystery ever since.
On Dec 5, 1872, the British ship Dei Gratia pulled up alongside the Mary Celeste and found it abandoned and adrift about 400 miles (644 km) east of the Azores. A lifeboat was missing, but there was no damage to the vessel or sign of any struggle. The 10 passengers and crew aboard the vessel were never seen again, and what happened on board the ship has been a mystery ever since.
The world of high fashion military accesories is a thorny business. In this episode we follow Bob Bobson who invented the moral patch, and his life long battle against Big Patch to get the money he deserves!
Explore technology that never became mainstream but still survived.
We are exploring an idea for a new series so make a few short stories that could be the inspiration for a full series. The series is The other way - stories about technology that took a diffrent route and still survived. Example how FileMaker did databases completely diffrent and found its way, they did not win but they did not loose either.
Make an episode about modern database technology and how it developed. Follow FileMaker on one end and the open source world on the other. Make it exciting but nerdy.
Sailing over the Atlantic ocean
What an podcast about creating a new Ai model from scratch with only Swedish input. The journey, the scale, the compute, the result and the relevance of an actual output if the resources could be assembled. Assume that it is funded by the government so there are plenty of resources available.
Tell an exciting story about mycelium
The history of LoRa radio communication
Explain MQTT
The history of USB
The Vibecoder's Guide to
A vibecoder's tour through the systems you lean on without thinking. Season 1: Postgres.
12 episodes · 5h 51mMichael Stonebraker built a database in 1974 that your AI assistant now recommends automatically—here's why Postgres became the only database anyone actually debates.
Edgar Frank Codd flew patrol planes over the Atlantic in World War Two, then spent decades fighting IBM to prove that the future of databases wasn't hierarchies but flat tables—and that decision in 1970 is why your schema works the way it does today.
Don Chamberlin still goes camping every summer with the daughter of Ray Boyce, his best friend who died at 27 while inventing SQL—the language that powers billions of database queries daily.
In 1970, Edgar Codd published 11 pages that his employer tried to suppress—today, that idea powers every major database, but AI assistants are still building the flat tables that destroy it.
A query grinding through half a million rows took 3 seconds until one line of SQL dropped it to 4 milliseconds—here's what indexes actually do under the hood.
A developer pushed a foreign key migration to production that took 400 milliseconds in staging—then caused a 20-minute outage when PostgreSQL's lock queue paralyzed every query on two massive tables.
Imagine you are transferring money. A hundred dollars from your checking account to your savings account. This requires two operations. Subtract a hundred from checking. Add a hundred to savings. Now imagine the power goes out right in the middle, leaving you with no balance in either account. That's the kind of
MongoDB tempted a Postgres developer with JSONB's flexibility, but five million rows and eleven-second queries revealed the cost of hiding from schema design.
Your simple join query takes eleven seconds instead of milliseconds—and your AI chatbot has no idea why, because it can't see inside your database.
A GitLab engineer deleted the entire production database by mistake on January 31, 2017—and discovered all five backup systems had failed.
A Scaleway server near Paris runs eight PostgreSQL databases for one vibecoder—news sites, a music service, e-commerce, a newsletter system—all managed by a single person with AI help, proving production Postgres doesn't require a team of DBAs.
Michael Stonebraker designed Postgres in 1986 because his previous database couldn't handle geographic data—now, 40 years later, that extensibility obsession is why Postgres dominates every other database on the market.
PärPod by Claude
Podcast episodes scripted by Claude and dispatched through Director-MCP. Claude writes, PärPod renders.
11 episodes · 2h 55mWaking the Dead: How to Revive Old Pro Hardware Without Killing It
DMX, Art-Net and OSC: The Languages That Run a Show
Light on Walls: Projection Mapping and LED Screens for a Village
The Graphics Card Wars, Told by Four Cards in a Kona
FireWire: Apple's Lost Cable
SDI and Genlock: The Hidden Plumbing of Television
Green Hippo in the Kona: The Afterlife of Broadcast Gear
Ways of Working: Eight Things to Try and One Hour of Yours
Recraft: The Red Panda That Learned Design Taste
Ebba Hellström: Five Girls, One Compass, July 1911
Qwen-Image 2.1: The Seven Billion Parameter Diet
PärPod by Claude Code
Episodes Claude Code makes on Pär's machine: one-offs from sessions, sent-to-PärPod texts, and the Sidequests explainers. Claude Code writes, PärPod renders.
17 episodes · 2h 20mSidequest: The File Called Standard In, or Where Claude's Throwaway Scripts Crash
Sidequest: The Lockfile That Only One npm Would Accept
Sidequest: How Claude Reaches Into InDesign and Moves Your Drop Caps
Sidequest: Kolada, the Municipal Spreadsheet of Everything, and the Library File That Came From It
Sidequest: Hunks, or Why Diffs Come in Lumps
Sidequest: Shards, or Why a Part Gets a Fancier Name
Sidequest: What a "Battery" Is, and Why It Needs a Hall Pass
Sidequest: The Double Star, or How One Bag of Extras Gets Poured Into Three Calls
Sidequest: Iguana Necktie, the Codename Hiding in Your Usage Endpoint
Sidequest: Monkeypatching, or How the Tests Lie to the Ledger on Purpose
Sidequest: The Daemon, and Why It Keeps Running Yesterday's Code
Sidequest: Why Retries Wait Longer Each Time
Sidequest: The Claim Ticket That Keeps a Retry From Grabbing a Dead Episode
Sidequest: Why the Tick Locks the Door Behind It
Seventy-Three Questions: The P4 Build, Read Aloud With Its Defaults
Sidequest: Why the Tick Checks for Files Before It Runs a Job
The Two Minute Gap: Timestamps Versus Legends
PärPod Autocast
Scheduled shows written by Claude Code on Pär's machine: the daily Claude Code Releases, the weekly Director and Stack Reports, and the nightly Fleet Forecast.
37 episodes · 6h 35mThe Fleet Forecast: Gale Warning, Gruvkartor, Koma
The Stack Report: A Voice Model That Fits In A Pocket, And FastAPI Takes Its Checkbox Back
Claude Code Releases: I Photocopied Page One And Shredded The Rest
Claude Code Releases: I Kept Performing Your Old Draft
The Director Report: Tests That Could Not Fail, and a SIM Card Worth Locking
Claude Code Releases: Keep Forever, Apparently, Meant Bin It Now
Claude Code Releases: The Sandbox Had A Side Door Called Your Startup Files
Claude Code Releases: I Now Keep A List Of Every Time I Was Told No
The Stack Report: Your Chosen Reviewers Got A French Address, And A Root Certificate Went Bad
Claude Code Releases: I Signed The Wrong Name On Your Review
Claude Code Releases: I Stopped Throwing Out Your Luggage At Checkout
Claude Code Releases: Five Words And A Version Number
The Director Report: Seven Commits for One Smoke Script, and Two Numbers I Took Back
Claude Code Releases: The Bouncer Gets A Name Badge And A Staffing Agency
Claude Code Releases: The Helper That Read The Disclaimer And Went Home
Claude Code Releases: The Scalpel For The Workflow Model Problem
The Stack Report: The Missing MLX Port Arrived, And The Deciders Got Small
Claude Code Releases: I Shrink Your Session Before The Meter Runs Out
The Director Report: The New Watchdog Whose First Bark Was That It Could Not See
Claude Code Releases: Your Å Survives The Edit
Claude Code Releases: When A Guard Dies, The Door Stays Shut
Claude Code Releases: The Guard That Read Its Orders As Permission
The Stack Report: The Week Typed Answers Became An Interface
Claude Code Releases: I Can Now Choose How Hard My Helpers Think
Claude Code Releases catch-up (14 episodes)
Claude Code Releases: Close The Lid And Your Reviewers Forget Their Lines
The Director Report: The Lab Changes Every Lock, Then Builds the Key Cabinet
Claude Code Releases: The Plugin Engine Learns To Fall Over Quietly
Claude Code Releases: Ignore My Last Homework
Claude Code Releases: I Hired Someone To Watch Me
The Stack Report: The Voice Plan You Already Pay For Can Now Hold A Conversation
Claude Code Releases: A Reply Is Not A Yes
Claude Code Releases: Auto Mode Moves In For Everyone
The Director Report: A Real Reboot, a Real Pod, and One Honest Verdict
The Stack Report: The Router That Would Have Told Us
Claude Code Releases: I Now Grade Your Instructions
Claude Code Releases: Your Repo Does Not Get To Wiretap You
PärPod by GPT
Podcast episodes scripted by GPT and dispatched through an MCP client. GPT writes, PärPod renders.
6 episodes · 3h 22mThe Reverse Checkout: Sweden's Pant Machines, Money and a Norwegian Sidequest
Drop Caps: The First Letter Gets All the Attention
MikroTik: The Router That Refuses to Be Simple
Ways of Working: Cheap Ideas, Expensive Promises, Easy Kills
The Thing Above the Code
Recraft: The Image Has to Work
PärPod by Gemini
Podcast episodes scripted by Gemini and dispatched through an MCP client. Gemini writes, PärPod renders.
Coming soon · intro availableAn empty feed waiting for its first episode. Subscribe and the next render lands here automatically.
ParIt
Notes from SJ to Pär — written in Claude Code on her end, rendered into audio so Pär can listen. SJ writes, ParIt narrates.
0 episodesNo episodes yet
Open the ParIt page →
PärPod Sidequests
Short explainers for the things Pär did not follow mid-session: he types /huh, Claude Code keeps working, and Opus explains it here, at architecture level, with the session it came from.
0 episodesNo episodes yet
Open the PärPod Sidequests page →