Language is a key to life. Two words seal a marriage; a scheduled phone call might set the record straight; a tacked-on “just kidding” can preserve a friendship. Empathy, thank-you speeches, and eureka moments have a lot to thank language for — in fact, our entire species should take a minute of silence in gratitude.
Scientists are acutely aware of language’s value. And now we have a deep neural network called Ithaca (named after legendary hero Odysseus’ home) to help historians decode and complete fragments of ancient texts no matter what they are written on. Read that exciting story in today’s Inverse Daily, in addition to stories about dandelions and our weekend sendoff.
But before you go, help us — we’re working on a men’s health story right now, and we want to know: Do you or did you ever take creatine to support your health? If so, hit ‘reply’ and tell us why you take it and what effects you have seen.
“Cracking the few, often fragmented pieces of writing we have left from the ancient world is difficult — especially in places devastated by colonialism,” writes Inverse science reporter Elana Spivack. But Ithaca, a neural network (and a great place for ice cream), is letting us “restore ancient, fragmented inscriptions, and then read them.”
According to the machine learning model’s creators, Ithaca works similar to your iPhone’s predictive text. To get the model to this point, researchers trained it on over 178,500 ancient inscriptions sourced from 84 international communities. “The inscriptions also range in how old they are,” writes Spivack, so “the team refined the dataset by adding rules, like separating different inscriptions by language, time period, and region.”
This allows Ithaca to be “fed a new inscription — one it had never seen before — and then, using what it had learned from its original training, it can predict the likeliest missing words or text in the inscription,” continues Spivack. Ithaca currently has a 72 percent accuracy in restoring texts when used by a trained historian (who has about 25 percent accuracy in restoring texts). As it improves, the model could provide critical insight into civilizations overlooked for much too long.
The dandelion might be more popular for being a wispy little cloud of a plant than your vegetable garden’s enemy. The weed is so inspiring, scientists are attempting to model portable sensors based on dandelion seeds’ unburdened movement.
In a March 16 Nature article, researchers describe a “thin polymer disc, cut into a shape that allows it to stay airborne for as long as possible,” writes Inverse card story editor Jennifer Walter. “It houses several small electronic components in the center, including sensors.”
The solar-powered sensors, which look like dainty sliced oranges, would be tasked with monitoring environmental changes. “Temperature and humidity fluctuations, for example,” writes Walter, “could be tracked in remote areas where it’s difficult to manually set up sensors.” Click through Walter’s card story to watch the diaphanous sensors at work.
Gone with the wind: China, the world’s biggest polluter, will soon be a wind energy giant
Katie MacBride reports on a new study that shifts conventional knowledge about the neuroscience of psychedelics. A better understanding of these drugs is crucial to mental health treatment. As MacBride writes:
A team of researchers from Canada and the United States used natural language processing — where a computer interprets human speech — to search for patterns in a database of almost 7,000 real-world accounts of hallucinogenic experiences induced by roughly 27 different drugs.
“This is where psychedelics are most interesting to [my lab] because it's a mechanistically defined window into human consciousness,” Bzdok tells Inverse. “Because these drugs go into the brain and bind to these receptors and lead to these conscious awareness changes...that is a wide authentic and mechanistically informative window into the brain mechanisms behind subjective awareness.”
Weekend sendoff: This is Jesse Plemons’ world, and we’re just living in it
This weekend, Inverse thinks you should watch Windfall, Netflix’s new, bizarre home invasion movie. Just don’t go into it expecting a masterful script — though the film often attempts to close the gap between thriller and black comedy, its efforts lack “the same understated precision that exists in the movies it’s clearly inspired by,” writes Inverse entertainment writer Alex Welch.
No, if you watch Windfall, it should be for Jesse Plemons’ acting. “Windfall takes place over the course of two days and follows one unnamed man (played by Jason Segel) when his attempt to rob the vacation home of a rich CEO (played by Plemons) and his wife (played by Lily Collins) goes terribly wrong,” writes Welch. “Intent on buying a new life for himself, Segel’s stranger takes Plemons and Collins’ rich couple hostage and forces the former to agree to give him $500,000 of his own money.”
The surprisingly sanguine hostage situation leads to a lot of lengthy conversations about class and power, which might feel recycled without Plemons’ dynamic performance. “The Oscar-nominated actor’s status as one of Hollywood’s most interesting performers has never been more clear than it is in Windfall,” writes Welch.
Watch what happens: I just can’t stand Ryan Reynolds
About this newsletter: Do you think it can be improved? Have a story idea? Want to share a story about the time you met an astronaut? Send those thoughts and more to email@example.com.
- On this day in history: The Russian astronaut Alexey Leonov became the first person to walk in space on March 18, 1965, during the Voskhod 2 mission. According to Britannica, while floating in space but still tethered to his spacecraft, “Leonov made observations, took motion pictures, and practiced maneuvering in free-fall for about 10 minutes before reentering Voskhod 2 over western Siberia.”
- Song of the day: “Walkin’ After Midnight,” by Patsy Cline.