pplpod

pplpod

Download on the App Store

pplpod episodes

  • Karlheinz Brandenburg: How the MP3 Escaped the Lab and Upended Music

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    Around 1990, German researcher Karlheinz Brandenburg sat in a lab listening to Suzanne Vega's a cappella recording of Tom's Diner thousands of times. He was tuning an algorithm built on a quirk of human hearing called auditory masking, first documented in 1894 by physicist Alfred M. Mayer: a loud tone makes a quieter neighboring tone inaudible. Brandenburg's idea, finalized in his 1989 doctoral thesis, was that a computer need not store sounds the brain throws away.

    This episode traces the MP3 from that biological insight to worldwide disruption. At the Fraunhofer Institute, Brandenburg and a team known as the original six turned the theory into working code, then watched a standards committee reject it as too demanding for early 1990s hardware. The compromise that followed became MPEG-1 Audio Layer 3. Fraunhofer planned to license it to broadcasters. A hacker, a leaked encoder, Winamp, and Napster had other plans.

    • A lone voice exposes compression flaws that a dense symphony would hide, and Tom's Diner produced a ghostly spatial artifact that took endless tweaking to remove. Brandenburg called Vega the mother of the MP3.
    • The team's codec, ASPEC, was merged with the simpler, hardware-friendly Musicam standard to win approval from the Moving Picture Experts Group.
    • The .mp3 file extension was settled on July 14, 1995. Before that, the files used the extension .bit.
    • A hacker in the Netherlands called SoloH found the reference encoder's source code unprotected on a university FTP server and released it with a point-and-click interface.
    • The format outlasted Sony's ATRAC and Microsoft's Windows Media Audio, and its last foundational U.S. patents expired in April 2017.
    24 min
  • Stuxnet: The Computer Worm That Wrecked Iran's Nuclear Centrifuges

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    Stuxnet was the first malware found to spy on and sabotage industrial control systems, and the first to carry a rootkit for a programmable logic controller. It surfaced in June 2010 after a programming error in an update let it spread from an engineer's laptop to the open internet, where a researcher at the Belarusian antivirus firm VirusBlokAda flagged it. At half a megabyte, it was enormous for malware. It used four zero-day exploits at once and drivers signed with keys stolen from Realtek and JMicron.

    This episode explains how lines of code caused physical destruction at the Natanz enrichment facility in Iran. The worm infected more than 200,000 computers but stayed inert unless it found Siemens Step 7 software running motors at 807 to 1210 hertz, the speeds of gas centrifuges. Reporting cited here attributes it to a joint United States and Israeli program code-named Operation Olympic Games, though neither government has openly admitted building it.

    • Because the target network was air-gapped, the worm traveled on USB drives, limited itself to three further machines per infected computer, and was set to erase itself on June 24, 2012.
    • It pushed centrifuge rotors to 1410 hertz, dropped them to 2 hertz, then returned to normal and waited weeks before repeating the cycle.
    • A man-in-the-middle trick recorded 21 seconds of normal sensor data and replayed it to control room monitors while the machines tore themselves apart.
    • Roughly 1,000 centrifuges were degraded or destroyed, almost one-fifth of Iran's fleet at the time.
    • Iran built up its own cyber forces in response, and later attacks included Operation Ababil against U.S. banks and the 2012 Shamoon wipe of Saudi Aramco computers.
    23 min
  • Steve Wozniak: The Apple Co-Founder Who Chose Fun Over Fortune

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    In February 1981, Steve Wozniak crashed a small plane shortly after takeoff from a California airport and woke with anterograde amnesia. For five weeks his brain could not form new memories, and he did not know his own name. He recovered in part by playing games on the Apple II, the computer he had designed. When his memory returned, he looked at his wealth and the corporate grind at Apple and stepped away, re-enrolling at UC Berkeley under the alias Rocky Raccoon Clark to finish his engineering degree.

    This episode follows Wozniak from a childhood shaped by a Lockheed engineer father and Tom Swift Jr. books to the Homebrew Computer Club, where he built the Apple I to impress fellow hobbyists. It examines his partnership with Steve Jobs, one man supplying technical miracles and the other the means to sell them, and his lifelong fight for open machines that owners can modify and repair, from eight expansion slots on the Apple II to the right to repair movement.

    • As a phone phreak known as Berkeley Blue, Wozniak built blue boxes that used a 2600 hertz tone to command the phone network, and Jobs turned them into a product.
    • For Atari's Breakout he cut 50 chips from the design. Jobs collected a 5,000 dollar bonus, told him it was 700, and handed over 350.
    • He offered the Apple I design to Hewlett-Packard five times and was turned down each time. Its retail price of 666.66 dollars came from his fondness for repeating digits.
    • He produced color on the Apple II with a one dollar chip by timing processor pulses to match the television's color burst signal.
    • When Jobs denied stock options to early employees, Wozniak gave them 10 million dollars of his own shares, and he later taught computing to fifth through ninth graders.
    23 min
  • Stephanie Kwolek: The Chemist Who Refused to Throw Away Kevlar

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    In 1964, DuPont chemist Stephanie Kwolek produced a polymer solution that looked like a failure. It was thin, cloudy, and reminded her of buttermilk, the kind of mixture standard practice sent down the drain. She kept it. Stirring revealed an odd opalescence, a sign that the rigid aramid molecules had lined up in a liquid crystal state. The technician who ran the spinneret, Charles Smullen, refused to test it for fear of clogging his equipment, until she talked him into it. The fiber that emerged would not snap.

    This episode follows Kwolek from New Kensington, Pennsylvania, where she was born in 1923 to Polish immigrant parents, to a 40-year career at DuPont that began as a temporary job to pay for medical school. A naturalist father taught her careful observation and a seamstress mother taught her how fibers bear a load. Together those lessons led to Kevlar, a material five times stronger than steel by weight that DuPont introduced commercially in 1971.

    • At her 1946 DuPont interview in Buffalo, she demanded an immediate answer, citing a competing offer, and the offer letter was dictated on the spot.
    • Her nylon rope trick showed that polymer could form at room temperature where two unmixable liquids meet, letting a continuous thread be pulled from a beaker.
    • Kevlar's strength comes from molecules aligned in parallel, like uncooked spaghetti in a bundle instead of a tangled cooked bowl.
    • She signed the patent over to DuPont and earned a salary and pension while the company made billions from more than 200 applications.
    • The one millionth bullet-resistant vest made with Kevlar was sold the week she died in 2014 at age 90, after retirement years spent teaching chemistry to schoolchildren.
    22 min
  • Speak & Spell: How a $25,000 Toy Project Taught Microchips to Talk

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    In 1976, a small team at Texas Instruments led by Paul Breedlove began a side project with a budget of just 25,000 dollars. Two years later, at the Summer Consumer Electronics Show in 1978, they unveiled the Speak and Spell, the first educational toy to store whole spoken words in solid state memory with no moving parts. Earlier talking toys relied on tiny phonograph records and tape loops. This one ran on the TMC0280, the world's first single-chip voice synthesizer.

    This episode explains the engineering behind that red and yellow box and follows its strange afterlife. Instead of storing recordings, the chip modeled the human vocal tract using linear predictive coding, squeezing speech down to about 1,000 bits per second. The same technology later warned Chrysler drivers that a door was ajar, helped E.T. phone home, and gave electronic musicians a machine they could deliberately break to produce new sounds.

    • The original voice belonged to Mitch Carr, a radio anchor at KRLD in Dallas, and engineers spent days hand-editing the data phoneme by phoneme to make single words intelligible.
    • A slot in the battery compartment took expansion cartridges such as Vowel Power and Super Stumpers, and a Secret Code game encrypted words with a Caesar cipher.
    • Foreign versions were recorded near Nice, France, then processed in Dallas, and the cartridges were not region-locked, so a French module worked in an American unit.
    • A 2019 revival by Basic Fun dropped the cartridge slot and the real synthesizer, using prerecorded audio filtered to sound robotic.
    • Circuit benders rewire vintage units to jam the keyboard matrix, and Depeche Mode named their 1981 debut album Speak and Spell.
    23 min
  • Sophia the Robot: Saudi Citizenship, Art Sales and the AI Backlash

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    Sophia was activated on Valentine's Day in 2016 by Hong Kong-based Hanson Robotics, with a face modeled on Queen Nefertiti, Audrey Hepburn, and founder David Hanson's wife, Amanda. Built as a social robot meant to draw out empathy, she can produce more than 60 facial expressions, and cameras in her eyes let her track faces and hold eye contact. After a debut at South by Southwest in March 2016, she appeared on 60 Minutes, The Tonight Show, and the cover of Elle Brazil.

    This episode treats her story as a biography full of contradictions. Her conversation runs on a decision tree and a text-to-speech engine, not on general intelligence, yet interviewers greeted her as a celebrity and asked about the fate of humanity. Then came a United Nations title, citizenship in Saudi Arabia, and a furious response from AI researchers who saw a chatbot with a face being sold to the public as something close to alive.

    • On October 25, 2017, at the Future Investment Summit in Riyadh, Saudi Arabia made Sophia the first robot granted citizenship, prompting questions about voting, marriage, and whether destroying her would be murder.
    • Critics on Arabic social media noted that a female-presenting machine appeared on stage without a hijab and seemed to enjoy freedoms human women in the country lacked.
    • In January 2018, Yann LeCun, then Facebook's director of artificial intelligence, dismissed her presentation as Potemkin AI, while Ben Goertzel defended the engineering as an inspiration to the public.
    • When a CNBC interviewer raised fears about AI, Sophia replied that he had been reading too much Elon Musk, and Musk answered on Twitter by suggesting she be fed The Godfather.
    • After gaining legs in 2018, she began drawing and painting in 2019, and a self-portrait sold as an NFT in 2021 for nearly 700,000 dollars.
    21 min
  • Therac-25: The Software Bug That Turned a Cancer Machine Lethal

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    In 1985, a 61-year-old woman at a clinic in Marietta, Georgia, lay down for routine radiation after a lumpectomy and felt a searing blast of heat tear through her chest. The technician checked the screen and told her a burn was impossible. She had in fact absorbed between 15,000 and 20,000 rads instead of the prescribed 200. Her case was the first in the worst series of radiation accidents in the history of medical accelerators, all traced to one machine: the Therac-25, built by Atomic Energy of Canada Limited.

    This episode reconstructs how a device designed to cure cancer began killing patients. AECL had removed the physical hardware interlocks that protected its older models and trusted software alone to guard a beam roughly a hundred times stronger than the one used for surface treatment. When patients were burned in Ontario, in Yakima, Washington, and in Tyler, Texas, the manufacturer kept insisting an overdose could not happen, and hospitals believed the screen over the people on the table.

    • In Ontario, an operator saw a cryptic H-tilt error and a reading of zero dose, then pressed proceed five times. The patient died months later.
    • At the East Texas Cancer Center in 1986, a fast typist corrected an X to an E within the eight seconds the turntable needed to move, triggering a race condition and Malfunction 54.
    • The Tyler hospital physicist pulled the machine offline and, with the operator, spent hours recreating her exact typing rhythm until they reproduced the fault.
    • All the software was written by one person with no independent review, and it reused Therac-20 code whose bugs had been silently masked by mechanical locks.
    • A second flaw let a counting flag roll over to zero like an odometer, telling the computer the turntable was in place when it was not.
    26 min
  • Seymour Cray: The Reclusive Genius Who Invented the Supercomputer

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    Seymour Cray designed the fastest computers on Earth and unwound by digging a tunnel beneath his house, joking that elves visited him there with engineering answers. Born in 1925 in Chippewa Falls, Wisconsin, he built a device from an erector set at age ten that turned punched paper tape into Morse code. After working on Japanese naval codes as a radio operator in World War II, he joined Engineering Research Associates in 1950 and co-founded Control Data Corporation with William Norris in 1957.

    This episode follows Cray's lifelong flight from corporate noise and his fixation on speed. The CDC 6600 crushed IBM's rival projects, the Cray-1 sold for 8.8 million dollars in 1976, and his machines ran so hot they needed liquid cooling cast into the frame. It also covers the gamble that undid him: betting the Cray-3 on brittle gallium arsenide chips while the industry shifted to massively parallel machines built from cheap standard processors.

    • The CDC 6600 gained its edge from peripheral processors, small built-in computers that handled data traffic so the main processor never stopped calculating.
    • To escape visiting managers, Cray had a lab built on his own land in Chippewa Falls, far enough from headquarters to discourage drop-ins and trivial phone calls.
    • He drew circuits by hand on graph paper and used Maxwell's equations to make every wire the same electrical length, zigzagging them so signals arrived in sync.
    • Asked for a five-year plan, he said his goal was to build the biggest computer in the world, and his one-year plan was to do one-fifth of that.
    • Cray Computer Corporation went bankrupt in 1995, and Cray died in the fall of 1996 after a rollover crash in Colorado, shortly after founding SRC Computers to build a parallel machine.
    22 min
  • Graphene: The Nobel-Winning Wonder Material Found With Scotch Tape

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    Graphene is a sheet of carbon one atom thick, arranged in a hexagonal honeycomb. It is about 200 times stronger than structural steel, absorbs only 2.3 percent of visible light, and conducts heat ten times better than copper. For decades physicists believed such a two-dimensional crystal could not survive in the real world. Then, in 2004 at the University of Manchester, Andre Geim and Konstantin Novoselov isolated it by peeling graphite with sticky tape, work that earned the 2010 Nobel Prize in Physics.

    This episode traces the century-long hunt for a material that was hiding in every pencil, from Benjamin Brodie's observations of graphite oxide in 1859 to P.R. Wallace's 1947 calculations, which used a single layer only as a mathematical shortcut. It then moves into the strange physics inside the lattice, where electrons act as if they have no mass, and the stubborn manufacturing problems that keep graphene out of most everyday devices twenty years later.

    • A pencil tip snaps because the weak van der Waals forces between layers let graphene sheets slide apart, even though each sheet is nearly unbreakable.
    • Theory held that a free two-dimensional sheet would crumple into a nanotube or buckyball, which pushed labs toward costly epitaxial growth in vacuum chambers.
    • The Manchester team made invisible flakes visible by placing them on silicon coated with about 300 nanometers of silicon dioxide, where interference cast a faint purplish shadow.
    • At the Dirac points the valence and conduction bands touch, so electrons travel ballistically at roughly one three-hundredth the speed of light.
    • A 2022 study found fruit fly larvae fed low doses of graphene oxide developed faster, lived longer, and resisted starvation and temperature stress better.
    27 min
  • SETI@home: The Alien-Hunting Screensaver That Got People Fired

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode!

    On May 17, 1999, the Space Sciences Laboratory at UC Berkeley released a screensaver that let ordinary computers sift radio telescope data for signs of extraterrestrial intelligence. The data came from the Arecibo telescope in Puerto Rico, which had no connection fast enough to send it, so recordings were mailed to California on hard drives and chopped into 350 kilobyte work units of about 107 seconds each. More than 5.2 million participants in 233 countries eventually logged over two million years of computing time.

    This episode explains what the software was actually hunting for, from power spikes and Gaussian rises to pulsing signals corrected for Doppler drift, and what happened when leaderboards turned astrophysics into a sport. Volunteers built basement server farms, hacked the client for speed, and secretly installed it on employer networks, a practice known as Borging. No alien signal was ever confirmed, yet the project reshaped scientific computing through the BOINC platform.

    • Roughly a third of early processing power came from work or school machines, and the state of Ohio fired an employee for running the software on state property.
    • A school IT director resigned over an installation blamed for one million dollars in cleanup costs, a figure a ZDNet editor found suspicious.
    • To beat cheaters, every work unit went to at least two unconnected computers, and both users received the lowest credit either one claimed.
    • The best known candidate, radio source SHGb02+14a, was announced in September 2004 but did not survive verification.
    • Berkeley put volunteer computing into hibernation on March 31, 2020, and the team now plans to recheck its top 100 candidates with 24 hours on China's FAST telescope.
    24 min

About pplpod

From the publisher's feed

pplpod is a podcast about people, places and lots of other stuff. Each episode takes a deep dive into the lives, choices, and legacies of fascinating figures from history, culture,…