Originally Posted by
barturtle
I believe there are some other losses: some of these high-power lights have been known to cause interference with wireless cyclocomputers, so either magnetic or RF fields are created, causing some of the lost power. I'd also guess that the lumen/power input ratio is not a linear ratio.
The RF noise power loss is real, but miniscule. As for the lumens/power ratio not being linear -- you're right, it's not.