Offline-first field capture: how CRDTs solve the sync problem
Connectivity is a luxury in the field. A surveyor in a remote catchment will spend hours offline, and the moment two officers edit the same feature — a shared parcel boundary, an attribute on the same asset — you have a distributed-systems problem dressed up as a fieldwork problem. The naive fix, last-write-wins, is the one that quietly destroys a morning's work.
Why last-write-wins is a data-loss policy
Last-write-wins sounds reasonable until you follow it through. Two officers edit the same boundary offline. Both come back into coverage. The one who syncs second overwrites the first, whose edits vanish with no warning and no record they ever existed. In a survey context that is not a merge conflict, it is lost evidence.
CRDTs, in one paragraph
A Conflict-free Replicated Data Type is a structure designed so that concurrent edits on different replicas can always be merged into the same result, regardless of the order they arrive. There is no central referee and no "who wins" prompt — the merge function is mathematically guaranteed to converge. That property is exactly what an offline-first field tool needs.
Geometry and attributes get different treatment
Not all edits are the same shape, so Field Sync uses two mechanisms. Geometry edits go through a state-based CRDT that merges vertex-level changes, so two officers refining different parts of the same boundary both keep their work. Attribute mutations carry a vector clock, which lets the system detect genuine concurrency and either merge automatically or surface a true conflict for a human when — and only when — both officers changed the same field to different values.
- Geometry: state-based CRDT merges concurrent vertex edits deterministically.
- Attributes: vector clocks distinguish genuine concurrent edits from stale ones.
- Real conflicts are surfaced to a person; everything else converges silently.
What this feels like in the field
For the surveyor, the machinery is invisible. They work offline, they come back into coverage, their edits sync, and nothing is lost. The only time they see a prompt is a true conflict — the same field set to two different values — which is exactly the case where a human should decide anyway.
See Airfree Geospatial in action
Enterprise cadastre, LiDAR, earth observation, and AI on sovereign spatial infrastructure. Book a demo tailored to your data and jurisdiction.
Request a demo →