Numlock News: January 16, 2024 • Mean Girls, Fingerprints, KFC
By Walt Hickey
Following the smash-hit success of Wonka, yet another musical has crushed expectations and outperformed at the box office, with Mean Girls — the film adaptation of the Broadway musical Mean Girls, itself based on the 2004 film Mean Girls — making a surprising $33.2 million, another $6.5 million coming in from a few overseas markets, a solid opening for a film made for $36 million. Meanwhile, Wonka crossed the $500 million mark globally, becoming one of the highest grossing musicals of all time. It’s unclear how to read this, as both films were marketed to deliberately obscure the fact that they are, in fact, musicals, but nevertheless over-performed based on their ability to grab an audience that likes musicals.
A new study looked at a type of crustacean called amphipods, some of which actually produce silk, just like a spider does. The genetic analysis found that the ability to produce silk has evolved multiple times over the course of crustacean evolution. The paper argues that the ability to manufacture silk had evolved no fewer than six different times in the vast evolutionary history of crustaceans, among amphipods, ostracods and tanaids, all of which are vaguely shrimpy. Anyway, call Marvel, because I have an extremely gross villain concept for a Spider-Man and Namor team-up.
A small metrics tweak made by Apple has sent the entire podcast industry into a moment of doubt, as the all-powerful downloads number that many podcasts bank on in order to command ad inventory has been seriously impacted by a small technical tweak in September to the way that the Apple Podcasts app handles its queue. Essentially, prior to the shift, the app would automatically download any podcast to which a user was subscribed. Following the shift, automatic downloads were switched off for users who had not listened to five episodes of a podcast within the past two weeks. This led to a massive decline in downloads from users who never actually listened to podcasts anymore, with some severely impacted shows seeing declines as high as 40 percent.
New Orleans Style
In 2022, KFC China sold $100 million worth of New Orleans-style wings, selling 480 million pieces of New Orleans-style chicken per year. The wings — sauces with sugar or honey in vegetable oil with onion powder, garlic powder and a little bit of chili powder — are a massive hit in China. The best part is that New Orleans-style wings were straight-up just invented in China about 20 years ago by KFC, and there is absolutely no such thing in New Orleans. It’s like a reverse General Tso’s chicken, which was invented in the U.S. for American palates by a Chinese chef.
The Afrobeats genre of music has seen significant global growth driven by streaming services, including on-the-ground investments by Spotify in Nigeria to develop the genre’s presence on the platform. From 2017 to 2022, streams for Afrobeats songs on Spotify increased 550 percent, with 2023 seeing 14 billion streams from the genre, up from 2 billion streams in 2017. The genre is seeing success outside of Nigeria, with London and Paris among the top five cities listening. Revenue generated by Nigerian artists from Spotify in 2022 hit $27 million.
Japanese breweries have seen what American drinkers have done for Mexican breweries and would like some of that, please. In general, breweries in the U.S. are cheaper than they have been in recent memory, and that’s led Japan’s Asahi to buy up the contract brewery Octopi Brewing. The company targets 700,000 cases of production (50,800 barrels) of their beers, such as Asahi Super Dry and Kozel, for the North American market. That is well within the powers of Octopi, which is the 13th-largest craft brewer in the U.S. and produced 220,000 barrels of beer in 2022. Japanese beers have generally been a niche within a niche in the U.S., ranking 15th in terms of importers, but Japan’s industry is looking outward as drinking declines domestically. In 2019, 7.8 percent of Japanese people in their 20s drank regularly, down from 20.3 percent the prior decade.
A neural network trained first on 525,000 synthetic fingerprints followed by 53,315 actual fingerprints from 927 people in the hopes of designing a program that could identify other fingerprints from the same person. To test it, they fed the machine a real fingerprint from a subject, and then two other prints, one of which was also from that individual and one of which was from a different person. Given random choice, we’d expect someone to guess it right 50 percent of the time. In a test drawing from the fingerprints of 133 people, the program got it right 77 percent of the time, an indication that it is indeed possible to figure out other fingerprints of a person given a single sample.
Thanks to the paid subscribers to Numlock News who make this possible. Subscribers guarantee this stays ad-free, and get a special Sunday edition. Consider becoming a full subscriber today.