Microbes, Robots, and Ambition

Sourdough features a microbial starter connected to strange, fictional music. For the audiobook, Sloan used SampleRNN, an open-source tool implemented by Kyle McDonald, to learn from Croatian klapa singing and generate short, unfamiliar musical passages. He describes the result as a machine learning system’s imperfect but interesting transformation of its input.

Sloan says robotic arms interest him both for their mechanics and their appearance, though he has not operated one himself. Kitchens pose a challenge for automation because their objects, surfaces, and layouts are irregular; he calls cracking an egg a telling example. He likes combining new technology with older crafts instead of treating them as competing choices.

Sloan sees room for artists to explore machine learning beyond practical commercial applications. He describes a text-completion plugin for the Atom editor that lets writers review and choose among generated continuations while composing. For him, a useful creative tool should give people room to learn and shape its output.

Sloan lost the source code for his Fish app, and later iOS updates made it unusable for newer users. The experience raised a practical question: how can artists preserve digital work without committing to endless maintenance? He contrasts software’s dependence on changing platforms with the lasting legibility of printed books.

Sloan says the shift in public attitudes toward technology affected how he revised Mr. Penumbra’s 24-Hour Bookstore and wrote Sourdough. He wanted Sourdough to reflect the pressures of contemporary Silicon Valley, including difficult work at a robot factory, without simply condemning technology. He also values characters and narrators readers want to spend time with.