Drifter begins from a simple inversion: the instrument does not stand outside the environment it measures. It is carried by the same current, wind and waves that it attempts to register. The resulting trace is therefore not a neutral map but a situated observation, produced from inside movement. Drift becomes both condition and method: the object records the sea while the sea continuously rewrites the position from which that record is made.
At sea, visibility depends on infrastructure. A receiver leaves coverage, a battery weakens, a packet arrives late, or a position becomes older than the map suggests. Drifter keeps these states distinct. A missing transmission is not silently repaired into a continuous trajectory; the gap remains part of the evidence. Coverage therefore becomes another geography, one that decides which movements can be seen and which temporarily disappear from the technical field.
A drift model can propagate possible trajectories from a position through environmental forcing, while a physical drifter produces one material path through conditions that can never be represented completely. Their relation is comparison, not equivalence. Agreement may support an interpretation; divergence may expose missing forcing, object-specific behaviour or incomplete observation. The widening field of possible positions is not treated as failure, but as an honest description of uncertainty accumulating over time.
The Central Mediterranean is crossed by logistics, rescue operations, fishing, military and border infrastructures, migration routes, weather systems and overlapping jurisdictions. Much of what becomes knowable depends on technologies owned by states, scientific institutions and commercial providers. Drifter asks what changes when a small, reproducible instrument can produce a local observation outside those infrastructures, without pretending that such an observation can replace them or become a complete view of the sea.
Environmental sensing often promises control: measure a territory, reduce uncertainty, predict behaviour, make the field manageable from a distance. Drifter keeps the technical capacity while refusing the authority of total vision. The device does not need to identify everything around it, preserve every signal or convert every absence into an inference. Its usefulness comes from limiting its claim, carrying the material conditions of its observation rather than hiding them behind the appearance of a comprehensive system.
Battery life, enclosure, calibration, buoyancy, radio range and delayed reception shape what the device can know. These are not external engineering problems separated from the artwork; they are the material form of its perspective. A failed transmission exposes dependency on infrastructure that remains invisible when it works. In this sense failure becomes a method: it reveals where observation depends on power, maintenance, connectivity and the continued ability of the object to remain in the world.
Open hardware redistributes a limited capacity to produce and inspect maritime observations. Reproducibility does not make the resulting knowledge neutral, but it allows another person to rebuild the instrument, question its calibration, alter its assumptions and compare traces. Drifter therefore approaches civic sensing as a distributed practice rather than a miniature scientific authority. Several partial observations can coexist without being forced into a single omniscient viewpoint, and their differences can remain visible as part of the research.
Drifter becomes practically useful when its record can enter a larger evidence architecture without losing its identity. In SeaCommons, a sensor packet may sit beside incident reports, vessel positions, environmental conditions and computed drift. The relation between them is provenance. A packet remains a packet, a model remains a model, a report remains a report. Correlation can strengthen a picture, but it should not erase the different conditions through which each piece of evidence was produced.
Drifter moves between artwork and infrastructure without allowing one role to close the other. The artistic project keeps the politics of sensing open before utility becomes the only criterion; the research system tests whether those questions survive operational demands. To be carried is different from navigating. To disappear from a network is different from ceasing to exist. To estimate a future position is different from observing one. The sea remains larger than the system designed to describe it.