Claim Check
A claim gets repeated everywhere, but whether it holds depends on the half nobody says out loud. Each entry unpacks one: what was claimed, what actually happens, where the gap is, and when it closes.
Claim Check · B
"Open source" — which layer is actually open?
The most common phrase on a spec sheet, and in DIY circles it is read as "I can repair it, modify it and replicate it"…
Unitree G1 · SO-101 arm · LeRobot Humanoid · Dynamixel and the ROBOTIS ecosystem · 2026-09-28
Claim Check · B
More degrees of freedom is better — the most popular claim, and the easiest to falsify
Degrees of freedom is the easiest number to compare on a spec sheet and the easiest to charge more for. It is also the…
Unitree G1 · SO-101 arm · LeRobot Humanoid · 2026-09-28
Claim Check · B
How much of a LIBERO score measures the environment instead of the model
A benchmark is doing two jobs at once: measuring capability and being gamed. When randomisation is thin, the second on…
LIBERO benchmark · OpenVLA · 2026-09-28
Claim Check · B
'The simulation score is high, so it will work on hardware': where that breaks
Treat simulation as a filter, not a promise. A policy that scores well in sim is worth trying on hardware; do not copy…
MuJoCo · LIBERO benchmark · OpenVLA · 2026-09-28
Claim Check · A
'This servo does 20 kg·cm': can the datasheet figure be a design figure?
The design move: size by a third of the claimed figure, and use the peak only to sanity-check transients. Add a column…
SG90 and MG996R servos · Feetech STS series servos · Robstride joint motors · 2026-09-28