What we call “cloud computing” these days is an umbrella term for a bunch of technologies that pool together processors, storage, databases, networking, etc. and makes them easy to use, deploy, and scale. The magic is mostly in the software that clusters together massive amounts of hardware, which is housed in huge, geographically dispersed datacenters. “The cloud” feels like a single abstraction of what we used to call servers, even if there’s no single cloud but a bunch of disparate services. The whole system is pretty remarkable and makes just about every part of modern computing possible.

The Solar Protocol project inverts a lot of the thinking of cloud computing and is building a platform with rather different goals:

This website is hosted across a network of solar powered servers and is sent to you from wherever there is the most sunshine.

So much of modern computing infrastructure is all about optimizing for speed, either by physically locating hardware as close as possible to end-users or using AI to reduce the nanoseconds it takes to route packets. The Solar Protocol is kind of an inversion of these principles, picking servers based on where it happens to be daylight — you don’t need machine learning to derive the algorithm of where the sun is, after all. From the manifesto:

If intelligence is the capacity to synthesize knowledge as logic and apply that logic to make decisions, then the Solar Protocol platform relies on an intelligence that emerges from earthly dynamics: specifically that of the sun’s interaction with the Earth. Our lives have always been directed by a range of natural logics that emerge from the intermittent dynamics of our shared environment. Weather, seasons, tides and atmospheric conditions all dictate our behavior, enabling and constraining our movements, food production and cultures. Solar Protocol uses these logics to automate decisions about how the network operates and what content is shown at different times of the day. How can we learn or relearn to design with natural intelligence?

The present day imagination for the internet has been enabled by an energy regime that relies on lethal fossil fuels. And the result? An online culture that valorizes speed, self expression through ever larger media and data-driven intelligence that is requiring more and more energy. Machine learning for example, requires enormous datasets typically collected through private efforts in online digital surveillance that collect every click, keypress, view and page scroll you might make. This data is then used to train models used to automate decisions about what content to show you. These energy hungry technologies are only made possible by extractive energy systems and labor practices.

(Happily, the software is open source and the hardware is relatively cheap and easy to acquire.)

I was reminded of a similar project by Low Tech Magazine a from few years ago (the Solar Protocol credits Low Tech Magazine for the inspiration) that also involved a solar-powered web server, with the limitation that sometimes the server was just not online. It’s a radical notion, in an age when everyone expects any piece of information to be available at any moment, to simply not be there. A sleep cycle for data.

There’s something of an ongoing debate around cryptocurrency at the moment, driven in part by the sudden meme of the NFT art market, about the massive environmental impact of blockchain technology. These protocols that allow for a distributed, decentralized internet are exciting, but they come at a massive cost and unsustainable carbon output. They could learn from a little natural intelligence.