1,000 operating hours can be impressive.
It is still not an uptime percentage.

Runtime is accumulated operating time. It tells us how much activity was recorded under the publisher's rules. Availability asks what share of scheduled operating time the system could perform its required function.

To calculate a meaningful availability percentage, we need the scheduled interval and a defined downtime total for the same system, task and observation window.

Suppose one robot is scheduled for 100 hours. Under a stated rule, it is unable to perform the task for 8 of those hours. Its measured availability for that window would be 92%. This is a hypothetical example, not a result for any robot in the database.

Now suppose a company only reports 92 operating hours. We still cannot calculate availability: we do not know whether 100, 120 or 200 hours were scheduled, or which pauses counted as downtime.

A deployment that lasted a year was not necessarily scheduled around the clock. A robot may work one shift, selected days or only when material is available. Calendar duration therefore cannot replace scheduled operating time.

The same caution applies to fleet totals. One thousand hours accumulated across ten robots is different from one robot operating for one thousand hours, even when both totals are accurate.

HumanoidUptime records a published runtime total when its meaning and source are clear. It records availability separately and leaves it undisclosed unless a qualifying source provides the required value or inputs.

That is why a record can contain substantial runtime and task output while availability, intervention rate, MTBF and MTTR remain unknown.