Server Farm

Claire L. Evans wrote a magical piece called Beyond Smart Rocks that took me by surprise. It digs under the silicon and sand that run modern computers to look for sustainable alternatives. A rich corpus of research into mold and memristors shows a fascinating alternative to the relationship we have with computing today. As Evans muses,

Switching from silicon to slime is a transformative idea. For me, the very question feels radically hopeful: might building computers from slime molds and mushrooms transform computing from a sophisticated solipsism into a far more sophisticated expression of our awe-inspiringly complex, interconnected world? Certainly it would change our whole relational experience of computing. It might also be more sustainable, as biological computer systems would consume far less energy than traditional hardware and, at the end of life, be completely biodegradable. ”We can shut down our PC completely,” [Andrew] Adamatzky explains, “but we will never shut down a living fungi or a slime mould without killing it.” Forget planned obsolescence.

The research that Evans dials in on is about Physarum polycephalum, a type of slime mold that has the propensity for limited learning.

Its sensing, searching protoplasmic tubes can solve mazes, design efficient networks, and easily find the shortest path between points on a map. In a range of experiments, it has modeled the roadways of ancient Rome, traced a perfect copy of Japan’s interconnected rail networks, and smashed the Traveling Salesman Problem, an exponentially complex math problem. It has no central nervous system, but Physarum is capable of limited learning, making it a leading candidate for a new kind of biological computer system — one that isn’t mined, but grown. This proposition has entranced researchers worldwide and attracted investment at the government level. An EU-funded research group, PhyChip, hopes to build a hybrid computer chip from Physarum, by shellacking its protoplasmic tubes in conductive particles. Such a “functional biomorphic computing device” would be sustainable, self-healing and self-correcting. It would also be, by some definition, alive.

“A new kind of biological computer system — one that isn’t mined, but grown.” That hits home for some reason. If alternatives like the PhyChip catch on, server farms could actually become more like farms than data centers, where maintenance relies on horticulture as much as it does on traditional IT skills. Imagine growing your parts for your laptop along with herbs for your dinner.

As Evans notes, such experiments only recapitulate the beautiful interconnectedness of our planet. At most we can produce a poor recapitulation — even as our modern world accelerates forward with “innovation” after “innovation.”

[A]s we dream of embedding artificial intelligence into every material surface of our lives, we are at best poorly emulating processes already at play beneath our feet and in our gardens. We’re making a bad copy of the Earth — and, in mining the Earth to create it, we are destroying the original.

However, there is something about the paradigm shift that this alternative creates that is indeed transformative. To go from “a sophisticated solipsism” towards a “sophisticated expression of our awe-inspiringly complex, interconnected world” is a shift I'd want to be a part of.

Best to keep an eye on this and work on my green (slime) thumb.