Acceleration rankings

Open the leaderboard. Choose 0–40, 0–60 or 0–100 km/h, or enter a wheel model. The general table requires continuous GPS confirmation. Model tables default to GPS only; turning that filter off also admits runs with independently validated battery energy. Each table shows one best run per rider and model, the date of the run, timing uncertainty and a marker for firmware modified through LoEUC.

Measure and upload#

Start an explicit acceleration measurement in the app while connected to your wheel. A passive best acceleration in a ride summary never uploads. The app records a standing start, telemetry and GPS; threshold times come from interpolation between wheel frames. Choose Submit to leaderboard on the result screen. A signed copy waits on your phone until internet and your site account are available. An old run remains eligible; there is no upload deadline. A repeat upload does not create another entry.

What the server checks#

The server verifies the device signature, recomputes threshold times and compares the client result. A signature alone does not earn an application badge. That badge requires validated Android Key Attestation or Apple App Attest, including an expiring server challenge. An unsupported device can still submit signed runs without the badge.

For GPS runs the server derives correction from raw wheel speeds and GPS Doppler speeds, not the rider's correction. GPS must cover the entire acceleration and recorded finish tail, with no mock flag or gaps. Doppler speed must agree with movement, the speed integral and finish speed; a significant downhill removes confirmation.

Small position fluctuations are smoothed with a local quadratic fit weighted by reported accuracy, preserving acceleration. Most fixes must have accuracy within 10 m; brief excursions up to 15 m are allowed. Distance comparisons account for position uncertainty and sparse speed sampling. Raw position jumps are checked before accepting the smoothed trajectory. Wheel and GPS speeds are compared over matching two-second windows; the finish also retains a raw speed check. GPS must cover one second after the recalibrated finish, while wheel telemetry must cover 1.8 seconds. Signed recordings remain unchanged. Without GPS, correction can only worsen the recorded time. Energy requires a server-maintained model entry with evidence for reliable battery current and wheel mass; unknown current sources remain under review and cannot enter a ranking. Spinning an unloaded wheel without sufficient energy fails this check.

Implausible frame jumps, clock inconsistencies, signed sequence/time inversions and extreme model outliers go to manual review. Late uploads with lower sequence numbers are allowed when the recorded order is consistent. The protocols and telemetry are open, so these checks raise the cost of fabrication rather than guarantee an authentic ride.

Charts show speed, motor phase current and PWM on a shared time axis. Battery current is retained separately for energy verification. Older recordings lack phase current and show an explanation instead; PWM is available if recorded.

Privacy and reports#

Open a ranking entry to see speed (wheel and GPS), motor phase current and PWM graphs. The public page contains no coordinates, serial number, signing key or private account identifier. Uploaded GPS geometry consists of offsets from the first fix, never absolute coordinates. Sign in and use Report if a run looks implausible. The owner reviews reports and records a reason for accepting or rejecting a run.

Current limits#

Satellite-orbit checks are not available: the upload contract contains neither satellite observations nor absolute positions. The server cannot recover wheel identity or the current sensor's provenance from decoded frames; it records the model and firmware captured by the app. Application attestation requires the operator to configure trusted platform roots and a fresh Android revocation list; absent configuration never grants the badge.