Methodology
Overtaq holds two independent judgements about every machine result and never collapses them into one. Verification asks whether the result happened as described. Comparability asks whether it can be set beside the human record at all. A result can be perfectly verified and completely incomparable.
The two axes
Verification
How well evidenced the result is.
Comparability
Whether the machine result can be set beside the human record.
Crossover classification
What Overtaq counts, derived from the two axes above.
Country attribution
Human records carry the nationality the athlete or player represented for that record. Machine results are attributed to the team, lab, company or institution responsible — never to the machine itself — and carry more than one country where the project is genuinely multinational. Overtaq publishes no country leaderboard: ten disciplines is not a sufficient dataset, and ranking multinational projects by country is not yet defensible. The fields exist so that filtering, structured data and the records feed can use them today.
Corrections
Every change to a published record is logged here with a date and a reason. Corrections are submitted through the correction form.
- 23 AUGUST 2026 · 100M
Separated two machine marks that had been merged into one record.
Overtaq initially carried 9.32s and 9.39s as conflicting reports of a single run. They are two different machines: 9.32s by Honor Robot’s Lightning in a pre-event test, 9.39s by an X-Humanoid entrant in competition. Both are now held as separate progression entries.
- 23 AUGUST 2026 · STANDING HIGH JUMP
Comparability downgraded to “not directly comparable”, permanently.
The machine mark is a standing jump; the human record cited against it is a running jump. The entry is retained to document the mismatch rather than removed.