Why Perpetual Motion Machines Never Work, Despite Centuries of Experiments

in History, Physics, Technology | January 16th, 2024

According to the laws of physics — at least in simplified form — an object in motion will stay in motion, at least if no other forces act on it. That’s all well and good in the realm of theory, but here in the complex reality of Earth, there always seems to be one force or another getting in the way. Not that this has ever completely shut down mankind’s desire to build a perpetual-motion machine. According to Google Arts & Culture, that quest dates at least as far back as seventh-century India, where “the mathematician Brahmagupta, who wanted to represent the cyclical and eternal motion of the heavens, designed an overbalanced wheel whose rotation was powered by the flow of mercury inside its hollow spokes.”

More widely known is the successor design by Brahmagupta’s twelfth-century countryman and colleague Bhāskara, who “altered the wheel design by giving the hollow spokes a curved shape, producing an asymmetrical course in constant imbalance.” Despite this rendition’s memorable elegance, it does not, like the earlier overbalanced wheel, actually keep on turning forever. To blame are the very same laws of physics that have dogged the subsequent 900 or so years of attempts to build perpetual-motion machines, which you can see briefly explained in the TED-Ed video above.

“Ideas for perpetual-motion machines all violate one or more fundamental laws of thermodynamics, the branch of physics that describes the relationship between different forms of energy,” says the narrator. The first law holds that “energy can’t be created or destroyed; you can’t get out more energy than you put in.” That alone would put an end to hopes for a “free” energy source of this kind. But even machines that just keep moving by themselves — much less useful, of course, but still scientifically earth-shattering — would eventually “have to create some extra energy to nudge the system past its stopping point, breaking the first law of thermodynamics.”

Whenever machines seem to overcome this problem, “in reality, they invariably turn out to be drawing energy from some external source.” (Nineteenth-century America seems to have offered endless opportunities for engineering charlatanism of this kind, whose perpetrators made a habit of skipping town whenever their trickery was revealed, some obtaining patents and profits all the while). But even if the first law of thermodynamics didn’t apply, there would remain the matter of the second, which dictates that “energy tends to spread out through processes like friction,” thus “reducing the energy available to move the system itself, until the machine inevitably stopped.” Hence the abandonment of interest in perpetual motion by such scientific minds as Galileo and Leonardo — who must also have understood that mankind would never fully relinquish the dream.

M. C. Escher’s Perpetual Motion Waterfall Brought to Life: Real or Sleight of Hand?

Leonardo da Vinci’s Inventions Come to Life as Museum-Quality, Workable Models: A Swing Bridge, Scythed Chariot, Perpetual Motion Machine & More

How the Brilliant Colors of Medieval Illuminated Manuscripts Were Made with Alchemy

A Complete Digitization of Leonardo Da Vinci’s Codex Atlanticus, the Largest Existing Collection of His Drawings & Writings

Based in Seoul, Colin Marshall writes and broadcasts on cities, language, and culture. His projects include the Substack newsletter Books on Cities, the book The Stateless City: a Walk through 21st-Century Los Angeles and the video series The City in Cinema. Follow him on Twitter at @colinmarshall or on Facebook.

by Colin Marshall | Permalink | Make a Comment ( 1 ) |

AI “Completes” Keith Haring’s Unfinished Painting and Controversy Erupts

in Art, Technology | January 12th, 2024

The story behind this painting is so sad! 😢
Now using AI we can complete what he couldn’t finish! ❤️ https://t.co/RuASoTfFdk pic.twitter.com/uAwM6SBUGW

— Donnel (@DonnelVillager) December 31, 2023

The celebrity graffiti artist Keith Haring died in 1990, at the age of 31, no doubt having completed only a fraction of the artwork he would have produced in a life a few decades longer. Upon first seeing his Unfinished Painting of 1989, one might assume that his early death is what stopped him from finishing it. In fact, painting only about a quarter of the canvas was his deliberate choice, intended to make a visual commentary on the AIDS epidemic that had claimed so many lives, and, not long thereafter, would claim his own. Presumably, it never occurred to anyone to “finish” Unfinished Painting — not before the age of artificial intelligence, anyway.

“Last summer, artist Brooke Peachley … posted a photo of the work on X” — the social media platform formerly known as Twitter — “alongside a prompt asking others to respond with a visual art piece ‘that never fails to destroy [them] every time they see it,’ ” write Elaine Velie and Rhea Nayyar at Hyperallergic. “Over six months later, another user responded to the original post with a generative AI image that ‘completed’ Haring’s purposely half-painted work, writing, ‘now using AI we can complete what he couldn’t finish!’ ”

One might, perhaps, sense a joking tone in that post, though the many incensed commenters it continues to draw seem not to take it that way. “The post swiftly caught the ire of the X community, with users describing the action as ‘disrespectful,’ ‘disgusting,’ and a ‘desecration,’ ” says Artnet News. “Some praised the powers of A.I. for ‘showing us a world without AIDS,’ while others deemed the tweet excellent ‘bait’ on an Elon Musk-led online platform that newly rewards outrage with engagement.” As often these days — and very often when it comes to applications of artificial intelligence in popular culture — the reactions to the thing are more compelling than the thing itself.

“The A.I.-generated image doesn’t appear to be faithful to Haring’s style, which often included images of human figures,” writes Julia Binswanger at Smithsonian.com. “These kinds of figures are visible in Haring’s original piece, but the image generator wasn’t able to replicate them.” The algorithmically filled-in Unfinished Painting may be without aesthetic or intellectual interest in itself, but consider how many viewers have only learned of the original work because of it. Nevertheless, stunts like this (or like zooming out the Mona Lisa) ultimately amount to distractions from whatever artistic potential these technologies may actually hold. A.I. will come into its own not by generating images that Haring or any other artist could have created, but images that no human being has yet imagined.

A Short Biography of Keith Haring Told with Comic Book Illustrations & Music

Keith Haring’s Eclectic Journal Entries Go Online

Behold the World’s First Modern Art Amusement Park, Featuring Attractions by Salvador Dalí, Jean-Michel Basquiat, Keith Haring, Roy Lichtenstein & More (1987)

Artificial Intelligence Brings to Life Figures from 7 Famous Paintings: The Mona Lisa, Birth of Venus & More

An AI-Generated Painting Won First Prize at a State Fair & Sparked a Debate About the Essence of Art

Based in Seoul, Colin Marshall writes and broadcasts on cities, language, and culture. His projects include the Substack newsletter Books on Cities, the book The Stateless City: a Walk through 21st-Century Los Angeles and the video series The City in Cinema. Follow him on Twitter at @colinmarshall or on Facebook.

by Colin Marshall | Permalink | Make a Comment ( 10 ) |

The Oldest Voices That We Can Still Hear: Hear Audio Recordings of Ghostly Voices from the 1800s

in Computer Science, History, Life, Technology | December 8th, 2023

What history nerd doesn’t thrill to Thomas Edison speaking to us from beyond the grave in a 50th anniversary repeat of his groundbreaking 1877 spoken word recording of (those hoping for loftier stuff should dial it down now) Mary Had a Little Lamb?

The original represents the first time a recorded human voice was successfully captured and played back. We live in hope that the fragile tinfoil sheet on which it was recorded will turn up in someone’s attic someday.

Apparently Edison got it in the can on the first take. The great inventor later reminisced that he “was never so taken aback” in his life as when he first heard his own voice, issuing forth from the phonograph into which he’d so recently shouted the famous nursery rhyme:

Everybody was astonished. I was always afraid of things that worked the first time.

His achievement was a game changer, obviously, but it wasn’t the first time human speech was successfully recorded, as Kings and Things clarifies in the above video.

That honor goes to Édouard-Léon Scott de Martinville, whose phonautograph, patented in 1857, transcribed vocal sounds as wave forms etched onto lampblack-coated paper, wood, or glass.

Edison’s plans for his invention hinged on its ability to reproduce sound in ways that would be familiar and of service to the listening public. A sampling:

A music player
A device for creating audiobooks for blind people
A linguistic tool
An academic resource of archived lectures
A record of telephone conversations
A means of capturing precious family memories.

Léon Scott’s vision for his phonautograph reflects his preoccupation with the science of sound.

A professional typesetter, with an interest in shorthand, he conceived of the phonautograph as an artificial ear capable of reproducing every hiccup and quirk of pronunciation far more faithfully than a stenographer ever could. It was, in the words of audio historian Patrick Feaster, the “ultimate speech-to-text machine.”

As he told NPR’s Talk of the Nation, Léon Scott was driven to “get sounds down on paper where he could look at them and study them:”

…in terms of what we’re talking about here visually, anybody who’s ever used audio editing software should have a pretty good idea of what we’re talking about here, that kind of wavy line that you see on your screen that somehow corresponds to a sound file that you’re working with…He was hoping people would learn to read those squiggles and not just get the words out of them.

Although Léon Scott managed to sell a few phonautographs to scientific laboratories, the general public took little note of his invention. He was pained by the global acclaim that greeted Edison’s phonograph 21 years later, fearing that his own name would be lost to history.

His fear was not unfounded, though as Conan O’Brien, of all people, mused, “eventually, all our graves go unattended.”

But Léon Scott got a second act, as did several unidentified long-dead humans whose voices he had recorded, when Dr. Feaster and his First Sounds colleague David Giovannoni converted some phonautograms to playable digital audio files using non-contact optical-scanning technology from the Lawrence Berkeley National Laboratory.

Dr. Feaster describes the eerie experience of listening to the cleaned-up spoken word tracks after a long night of tweaking file speeds, using Léon Scott’s phonautograms of tuning forks as his guide:

I’m a sound recording historian, so hearing a voice from 100 years ago is no real surprise for me. But sitting there, I was just kind of stunned to be thinking, now I’m suddenly at last listening to a performance of vocal music made in France before the American Civil War. That was just a stunning thing, feeling like a ghost is trying to sing to me through that static.

Scanning technology also allowed historians to create playable digital files of fragile foil recordings made on Edison devices, like the St. Louis Tinfoil , made by writer and early adopter Thomas Mason in the summer of 1878, as a way of showing off his new-fangled phonograph, purchased for the whopping sum of $95.

The British Library’s Tinfoil Recording is thought to be the earliest in existence. It features an as-yet unidentified woman, who may or may not be quoting from social theorist Harriet Martineau… this garbled ghost is exceptionally difficult to pin down.

Far easier to decipher are the 1889 recordings of Prussian Field Marshall Helmuth Von Multke, who was born in 1800, the last year of the 18th century, making his the earliest-born recorded voice in audio history.

The nonagenarian recites from Hamlet and Faust, and congratulates Edison on his astonishing invention:

This phonograph makes it possible for a man who has already long rested in the grave once again to raise his voice and greet the present.

Suzanne Vega, “The Mother of the MP3,” Records “Tom’s Diner” with the Edison Cylinder

A Beer Bottle Gets Turned Into a 19th Century Edison Cylinder and Plays Fine Music

400,000+ Sound Recordings Made Before 1923 Have Entered the Public Domain

The Web Site “Centuries of Sound” is Making a Mixtape for Every Year of Recorded Sound from 1860 to Present

Stream 385,000 Vintage 78 RPM Records at the Internet Archive: Louis Armstrong, Glenn Miller, Billie Holiday & More

– Ayun Halliday is the Chief Primatologist of the East Village Inky zine and author, most recently, of Creative, Not Famous: The Small Potato Manifesto and Creative, Not Famous Activity Book. Follow her @AyunHalliday.

by Ayun Halliday | Permalink | Make a Comment ( 1 ) |

A Busy Person’s Introduction to Large Language Models (LLMs)

in Online Courses, Technology | December 7th, 2023

You’re busy. You don’t have much time to figure out the deal with Large Language Models (aka LLMs). But you have some curiosity. Enter Andrej Karpathy and his presentation, “A Busy Person’s Introduction to Large Language Models.” It’s a one-hour tutorial that explains “the core technical component behind systems like ChatGPT, Claude, and Bard.” Designed for a general audience, the video explains what Large Language Models (LLMs) are, and where Karpathy sees them going. Andrej knows what he’s talking about. He currently works for OpenAI (the maker of ChatGPT), and, before that, he served as the director of artificial intelligence at Tesla.

As one YouTube commenter put it, “Andrej is hands-down one of the best ML [Machine Learning] educators out there.” At Stanford, he was the primary instructor for the first deep learning class, which has become one of the largest courses at the university. Enjoy.

Neural Networks for Machine Learning: A Free Online Course Taught by Geoffrey Hinton

Google Launches a Free Course on Artificial Intelligence: Sign Up for Its New “Machine Learning Crash Course”

Stephen Fry Reads Nick Cave’s Stirring Letter About ChatGPT and Human Creativity: “We Are Fighting for the Very Soul of the World”

Noam Chomsky on ChatGPT: It’s “Basically High-Tech Plagiarism” and “a Way of Avoiding Learning”

by OC | Permalink | Make a Comment ( 1 ) |

Rome Reborn: A New 3D Virtual Model Lets You Fly Over the Great Monuments of Ancient Rome

in History, Technology | November 30th, 2023

Thirteen years ago here on Open Culture, we first featured Rome Reborn 2.2, a digital 3D model of the ancient metropolis at the height of its glory in the fourth century. And that rebirth has continued apace ever since, and just last week bore the fruit of Rome Reborn 4.0, through which you can get a flying tour in the video above. Intercut with the computer-generated reconstructions is footage of the ruins of the very same parts of the city as they exist in Rome today. The opportunity for comparison thus provided allows us to appreciate not just the upgrades in the latest Rome Reborn’s level of detail, but also its degree of realism.

With each revision, the fourth-century Eternal City recreated in Rome Reborn looks more like reality and less like a video game. But that doesn’t mean you can’t get the same thrill of exploring it that you would from a video game, which is part of the appeal of loading up the latest version of the model on the virtual-reality app Yorescape, a product of the “virtual tourism” company Flyover Zone Productions founded by Rome Reborn’s project leader Bernard Frischer.

And it is Frischer himself who leads the in-app tour of “sites exemplifying the city’s geography, markets, temples, and much, much more,” enriched by “Time Warps spread around the city that allow you to toggle between the view today and the view from the same vantage point in antiquity.”

This is heady stuff indeed for enthusiasts of ancient Rome, who will no doubt be eager to see for themselves the new and improved digital models of ancient Roman structures like the Circus Maximus, the Arch of Titus, the Porticus Liviae, and the Temple of Minerva. These and many others besides appear in the Rome Reborn 4.0 demo reel just above, which shows off the culmination of 27 years of work so far by Frischer and his team. A digital archaeologist at Indiana University, Professor Frischer has pointed out still-absent features to come, such as “avatars infused with AI” with whom the twenty-first-century tourist can interact. We’ll have to wait for future iterations to do so, but surely we can summon the patience by remembering that Rome isn’t reborn in a day.

A Huge Scale Model Showing Ancient Rome at Its Architectural Peak (Built Between 1933 and 1937)

The Changing Landscape of Ancient Rome: A Free Online Course from Sapienza University of Rome

An 8‑Minute Animated Flight Over Ancient Rome

The Oldest Known Photographs of Rome (1841–1871)

High-Resolution Walking Tours of Italy’s Most Historic Places: The Colosseum, Pompeii, St. Peter’s Basilica & More

Based in Seoul, Colin Marshall writes and broadcasts on cities, language, and culture. His projects include the Substack newsletter Books on Cities, the book The Stateless City: a Walk through 21st-Century Los Angeles and the video series The City in Cinema. Follow him on Twitter at @colinmarshall or on Facebook.

by Colin Marshall | Permalink | Make a Comment ( 1 ) |

The Surprisingly Long History of Auto-Tune, the Vocal-Processing Technology Music Critics Love to Hate

in Music, Technology | November 29th, 2023

In the fall of 1998, pop music changed forever — or at least it seems that way today, a quarter-century later. The epochal event in question was the release of Cher’s comeback hit “Believe,” of whose jaggedly fractured vocal glissando no listener had heard the likes of before. “The glow-and-flutter of Cher’s voice at key points in the song announced its own technological artifice,” writes critic Simon Reynolds at Pitchfork, “a blend of posthuman perfection and angelic transcendence ideal for the vague religiosity of the chorus.” As for how that effect had been achieved, only the tech-savviest studio professionals would have suspected a creative misuse of Auto-Tune, a popular digital audio processing tool brought to market the year before.

As its name suggests, Auto-Tune was designed to keep a musical performance in tune automatically. This capability owes to the efforts of one Andy Hildebrand, a classical flute virtuoso turned oil-extraction engineer turned music-technology entrepreneur. Employing the same mathematical acumen he’d used to assist the likes of Exxon in determining the location of prime drilling sites from processed sonar data, he figured out a vast simplification of the calculations theoretically required for an algorithm to put a real vocal recording into a particular key.

Rapidly adopted throughout the music industry, Hildebrand’s invention soon became a generic trademark, like Kleenex, Jell‑O, or Google. Even if a studio wasn’t using Auto-Tune, it was almost certainly auto-tuning, and with such subtlety that listeners never noticed.

The producers of “Believe,” for their part, turned the subtlety (or, technically, the “smoothness”) down to zero. In an attempt to keep that discovery a secret, they claimed at first to have used a vocoder, a synthesizer that converts the human voice into manipulable analog or digital signals. Some would also have suspected the even more venerable talkbox, which had been made well-known in the seventies and eighties by Earth, Wind & Fire, Stevie Wonder, and Roger Troutman of Zapp. Though the “Cher effect,” as it was known for a time, could plausibly be regarded as an aesthetic descendant of those devices, it had an entirely different technological basis. A few years after that basis became widely understood, conspicuous Auto-Tune became ubiquitous, not just in dance music but also in hip-hop, whose artists (not least Rappa Ternt Sanga T‑Pain) used Auto-Tune to steer their genre straight into the currents of mainstream pop, if not always to high critical acclaim.

Used as intended, Auto-Tune constituted a godsend for music producers working with any singer less freakishly skilled than, say, Freddie Mercury. Producer-Youtuber Rick Beato admits as much in the video just above, though given his classic rock- and jazz-oriented tastes, it doesn’t come as a surprise also to hear him lament the technology’s overuse. But for those willing to take it to ever-further extremes, Auto-Tune has given rise to previously unimagined subgenres, bringing (as emphasized in a recent Arte documentary) the universal language of melody into the linguistically fragmented arena of global hip-hop. As a means of generating “digital soul, for digital beings, leading digital lives,” in Reynolds’ words, Auto-Tune does reflect our time, for better or for worse. Its detractors can at least take some consolation in the fact that recent releases have come with something called a “humanize knob.”

How the Yamaha DX7 Digital Synthesizer Defined the Sound of 1980s Music

What Makes This Song Great?: Producer Rick Beato Breaks Down the Greatness of Classic Rock Songs in His New Video Series

The Distortion of Sound: A Short Film on How We’ve Created “a McDonald’s Generation of Music Consumers”

How Computers Ruined Rock Music

Brian Eno on the Loss of Humanity in Modern Music

Based in Seoul, Colin Marshall writes and broadcasts on cities, language, and culture. His projects include the Substack newsletter Books on Cities, the book The Stateless City: a Walk through 21st-Century Los Angeles and the video series The City in Cinema. Follow him on Twitter at @colinmarshall or on Facebook.

by Colin Marshall | Permalink | Make a Comment ( 4 ) |

Generative AI for Everyone: A Free Course from AI Pioneer Andrew Ng

in Artificial Intelligence, Online Courses, Stanford, Technology | November 29th, 2023

Andrew Ng–an AI pioneer and Stanford computer science professor–has released a new course called Generative AI for Everyone. Designed for a non-technical audience, the course will “guide you through how generative AI works and what it can (and can’t) do. It includes hands-on exercises where you’ll learn to use generative AI to help in day-to-day work.” The course also explains “how to think through the lifecycle of a generative AI project, from conception to launch, including how to build effective prompts,” and it discusses “the potential opportunities and risks that generative AI technologies present to individuals, businesses, and society.” Given the coming prevalence of AI, it’s worth spending six hours with this course (the estimated time needed to complete it). You can audit Generative AI for Everyone for free, and watch all of the lectures at no cost. If you would like to take the course and earn a certificate, it will cost $49.

Generative AI for Everyone will be added to our collection, 1,700 Free Online Courses from Top Universities.

Computer Scientist Andrew Ng Presents a New Series of Machine Learning Courses–an Updated Version of the Popular Course Taken by 5 Million Students

Stephen Fry Reads Nick Cave’s Stirring Letter About ChatGPT and Human Creativity: “We Are Fighting for the Very Soul of the World”

by OC | Permalink | Make a Comment ( 15 ) |

Discover the Mikiphone, the World’s First Portable Record Player: “Fits a Jacket Pocket; Goes into a Lady’s Handbag” (1924)

in Design, History, Music, Technology | November 27th, 2023

The iPod shuffle recently enjoyed a bit of a comeback on TikTok.

Can the Mikiphone be far behind?

The invention of siblings Miklós and Étienne Vadász, the world’s first pocket record player caused a stir when it was introduced a century ago, nabbing first prize at an international music exhibition and finding favor with modernist architect Le Corbusier, who hailed it for embodying the “essence of the esprit nouveau.”

Unlike more recent portable audio innovations, some assembly was required.

It’s fair to assume that the Stanford Archive of Recorded Sound staffer deftly unpacking antique Mikiphone components from its cunning Sony Discman-sized case, above, has more practice putting the thing together than a nervous young fella eager to woo his gal al fresco with his just purchased, cutting edge 1924 technology.

A period advertisement extols the Mikiphone’s portability …

Fits in a jacket pocket

Goes in a lady’s handbag

Will hang on a cycle frame

Goes in a car door pocket

Ideal for picnics, car jaunts, river trips

…but fails to mention that in order to enjoy it, you’d also have to schlep along a fair amount of 78 RPM records, whose 10-inch diameters aren’t nearly so pocket and purse-compatible.

Maison Paillard produced approximately 180,000 of these hand-cranked wonders over the course of three years. When sales dropped in 1927, the remaining stock was sold off at a discount or given away to contest winners.

These days, an authentic Mikphone can fetch $500 and upward at auction. (Beware of Mikiphonies!)