HVAC wrote:

Many persons think that you can't get a patent on a perpetual motion
machine. That's not true, as these examples show.

United States Patents
May 24, 1932. H. L. Worthington 1,859,643 Magnetic Motor.
June 19, 1934. J. W. Poysa. 1,963,213 Magnetic Motor.
May 19, 1959. Norman L. Dean. 2,886,976 System for converting rotary
motion into unidirectional motion.
July 13, 1965. E. Baumgartner. 3,194,008 Positive Buoyancy Prime Mover.

Dec 20, 1966. Dan K. McCoin. 3,292,365. Power conversion apparatus and
method utilizing gravitational and buoyant forces.
Dec 7, 1971. E. Rutkove. 3,625,089. Gravity Wheel Apparatus.
Jan 27, 1976. David Diamond. 3,934,964. Gravity-Actuated Fluid
Displacement Power Generator.
April 24, 1979. Howard R. Johnson. 4,151,431 Permanent Magnet Motor.
April 24, 1979. Howard R. Johnson. 4,151,432 Permanent Magnet Motor
(Revision?)
January 22, 1980. Leslie R. Hinchman and Robert B. Hinchman. 4,184,409
Hydraulic Engine.
March, 2003. Thomas Bearden. 6,362,718. (A device that extracts
"compressed energy" from the time domain in amount mc2. It draws energy
from the longitudinal electromagnetic waves that fill the ocean of
space-time".) (!)

You can patent time machines too. And the patents exist.
You don't have to prove a patent works to get a patent.

This is a severe problem in emerging fields like photonics where people
patent things that they do not know how to do, but that someone will
eventually figure out how to do. The patent system is largely a
negative force in these fields.