Novices don't need precise control over musical elements, they want to enjoy playing and connect with others. Blaine & Fels argue such interfaces should be simple enough to approach in minutes, while offering layers of complexity to uncover over time.

Weinberg describes interconnected playing as a musical ecosystem. We chose a process-centred network, focused on the players' experience rather than the composition, where every node influences the others and playing together becomes a discovery of the system itself.

Piano Cubes, glass jars tilted to shift a constantly playing arpeggio, heavily inspired our interaction. Polymetros showed how people play together at public exhibitions, and how confusing it gets when players can't tell their contribution apart from others'.

We asked how a human-machine ecosystem of interdependent nodes could empower non-musicians to create music, and what methods people use to map such a system's inputs to its outputs.

Each shape holds a Puck.js Bluetooth sensor broadcasting its data to a central computer. Receiving the Bluetooth signals directly on the computer proved unreliable, so I added a development board that only listens for them and passes them on as MIDI, giving us fast, robust communication. Interference from other Bluetooth devices still caused some latency during testing.

We built the soundscape in VCV Rack, a modular synthesizer software. Each object produces its own sound or alters the sounds of the others, all from a predetermined set of musical patterns.

Building on Piano Cubes' turning and flipping, we designed four shapes. The pyramid plays a different drum pattern on each side, the cube plays melodies, the tall cylinder filters the drums and background music when rotated or rolled, and the short cylinder sets the tempo of the whole ecosystem.

In the first tests the objects were small and light, so people picked them up instead of moving and rotating them. We made them significantly larger to invite the intended interaction.

The cube repeated a single note on each side, which some testers found annoying. We changed it to play a predefined melody once each time it is flipped.

The short cylinder originally changed the density of the background music, which most testers didn't notice. We remapped it to control the tempo of the whole system.

We ran seven tests with five non-musicians, two of them in pairs. Each participant got minimal, intermediate or full instructions, explored freely for 10 minutes while thinking aloud on video, and then took part in a semi-structured interview.

Everyone started carefully, one shape at a time, and only began composing once they understood each shape. Disorganised exploration like shaking and tapping far outnumbered systematic flipping and rolling, 46 to 11 occurrences, and people picked the objects up 36 times.

Participants liked making music from preset patterns, though the simplicity made them lose interest after a while. The shapes invited intuitive motions but confused people when the sound didn't match the shape, and collaborating helped them discover interactions together.

The cylinders were easy to map, as rolling and rotating give immediate, continuous feedback. The cube and pyramid only change when flipped to another side, and that discrete response made them noticeably harder to understand.