Cube-Cola: Recipe as field report

Kate Rich

Kate Rich is an Australian-born artist and trader. In the 1990s she moved to California to work as radio engineer with the Bureau of Inverse Technology (BIT), an information agency servicing the Information Age. Restless at the turn of the century, she headed further east to Bristol UK, where she is now bar manager with Kayle Brandon at the Cube Microplex, an “alternative” cinema. Opposed in principle to the business and environmental practices of the Coca-Cola corporation, the Cube bar has never served Coke. Cube-Cola, an open-source alternative released under the GNU General Public Licence, is a project by Kate Rich and Kayle Brandon.

“We’d tried Pepsi and Virgin Cola and various others too,” says Brandon, “but they weren’t really a positive alternative. They were acceptable, but they weren’t Coke. And people really want Coke.” After conducting various taste tests, they felt the preference had less to do with flavour than the power of the brand. Any alternative they were going to offer had not only to taste almost identical but overcome the incredible pull of Coca-Cola’s marketing. “Given that most of the Cube’s customers come because they like the place’s DIY attitude,” Brandon explains, “one way of doing that was to make the cola ourselves.” (

In setting out to produce our own autonomous cola drink from a non-brand open source recipe found on the internet, it was not immediately clear from the downloaded instructions that to follow them would take two dark years of experiment and struggle. To get the recipe to actually work and to make the process reproducible, demanded for two years in laboratory wilderness, having gained some marginal command of our wilful assembly of cola ingredients, their supply chains, sequences, atmospheric preferences, decay arcs and volatile human effects. Also the equipment requirements such as brand varietals, retail sources, repair and replacement techniques and the gestures necessary to operate them involved an at times a steep learning curve.

We realised our attempts to transfer this now two-year, two-operator repository of contextual knowledge back into a generalisable recipe format would be futile, or at the least require inclusion of a 1:1 scale changelog, longitudinal catalogue of all components (including those that had failed), the GNU/GPL licence, plus at least 10 kg of essential receptacles and tools. We thereby retreated from the idea of recipe to something between software development snapshot, DIY training manual and factory floor plan. The aggregated archive and instructional data are online now.