Load the exp_*.jsonl recording from each phone. The viewer
reconstructs who was linked to whom at every moment of the run, and checks the mesh
size that was actually up against the size each step claimed.
Experiments on this server
For a run that has not been uploaded yet. One file per phone — select them all together so links can be matched between devices.
Measured distances — enter for any experiment id, before or after the run
Type the id the phones will use (the wizard's experiment id), pick how many phones, and fill one number per pair, in metres. Phones are numbered by join order. Saved on the server, so entering them from a phone at the field is enough. This beats GPS: at 30 m spacing a phone fix carries ±6–29 m of error — the size of the thing being measured.
Mesh size over the run — claimed by the step vs actually up
Topology at the scrubbed moment
Each phone’s own step right now
One line per phone: order, x, y.
Saved in this browser per experiment, so you enter them once. With positions
set, the graph is drawn to scale and pair distances appear in the uptime table
— which is what lets link uptime be read against real range.
Pair uptime within a block — share of the block each pair held a link
The route one packet took
A packet’s id survives relaying — it is the dedup key — so every node that saw the same packet recorded the same id, and a relay record names the neighbour it came from. That makes a real path, not a guess from timing.
Every sighting, in order
The mesh while it travelled
Every step in this run