Does a patent on an invention which does not work (but granted as if it does) count as invalidating prior art?

Question

Suppose Alice (a fictional name for this example) comes up with an novel invention idea for a machine to do very useful work, but which requires as part of the larger machine a component perpetual motion machine as its energy source. (The perpetual motion machine was patented by Bob, who never built it but licensed the patent to Alice.) Alice applies for and receives a patent without demonstrating any working model, and is never subsequently able to demonstrate a working model. The invention is never commercialized and nobody obtains the benefits of the work Alice's Marvelous Machine is not actually able to produce. Alice and Bob's patents can be assumed to be from a different country if that helps the premise seem more plausible.

Independently and slightly later, Claire comes up with a similar machine intended to accomplish the same goal, but it uses rechargeable batteries instead of the perpetual motion machine as an energy source. Claire's machine actually works and can provide the benefits that Alice's machine could not. However, substituting the energy source may be seen as relatively obvious, and the effect on its feasibility as an energy source might be seen as a reasonably predictable consequence of the swap-out. Other differences between the machines are immaterial to performance (just a result of independence of invention). Claire is applying to the USPTO.

Does Alice's patent on something that does not work prevent Claire from obtaining a patent on something that does?

Can this strategy be used to prevent patenting of an invention someone thinks would be more useful if open, even if that patenter can't quite get such an invention to work? Does Bob's patent block anybody else from patenting a perpetual motion machine?

Edited to add specific better example: Can someone patent a machine providing what they (incorrectly) think is a solution to solve an NP problem in P steps (i.e. claiming P=NP), even if not in the US, and have that block any later inventor from patenting a real solution that works? (If P=NP, such a machine would be one of the most valuable inventions to have been invented; patent rights could be very important on this.)

Answer 1

Using a better example. Postulate that Alice's invention would never work because it assumes something that is untrue. It is a device that moves you directly from any arbitrary stage of sleep to another arbitrary state of sleep by making a specific but undefined soft sound. It is not a perpetual motion machine but current science has found no way to do this and the experts in the field are sure it can't be done. Clare finds a specific soft sound, that she characterizes in detail, that will raise the stage of sleep to a more aroused state than the current state 90% of the time. Maybe she has done experiments and presents data. Claire's is a special case of Alice's and is therefore not even novel let alone non-obvious. Past patents and publications must be enabling in order to qualify as usable prior art. Alice's patent is not actually enabled but is presumed to be enabled because it was granted. To get past this, Claire will need convincing affidavits from experts in the field that, at the time of Alice's filing, it was flat out scientifically impossible to do what she claimed her device does. True story.

On the topic of perpetual motion machines specifically - they can't get a patent because they fail the "usefulness" criteria since it is an impossibly. However you can get a patent on a novel and non-obvious mechanical device that spins around for a very long time as long as no energy is being extracted from it. It is an amusement device that is fun to watch spin. I think there is a class/subclass just for this.

Answer 2

I think your example is contrived. The way you explain it, Alice's invention actually would function given a working power source. No one, in my experience would patent such a device with claims specifying Bob's perpetual motion machine if another power source would also work. You would claim a generic power source and then maybe have a dependent claim specifying Bob's device.

In any case, Alice's patent is indeed prior art for Claire's invention. The fact that Claire may have started first is of no importance since Alice filed a patent first. Also irrelevant is whether Alice commercialized her invention. The simple fact is that Alice has a patent. Claire must either avoid infringing it or else negotiate a license. Also, I doubt that Claire can obtain a patent for her device if it is functionally the same as Alice's except it uses batteries. Substituting a power source would likely be deemed obvious by the examiner.

As George White points out, it is possible to invalidate a granted patent. It just is usually hard and expensive to do so.