Just to make sure no one gets the wrong idea, it should be noted that the law requires:
The specification shall contain a written description of the
invention, and of the manner and process of making and using it,
in such full, clear, concise, and exact terms as to enable
any person skilled in the art to which it pertains, or with
which it is most nearly connected, to make and use the same,
and shall set forth the best mode contemplated by the inventor
of carrying out his invention.
(35 USC 112). So, even though filing counts as constructive reduction to practice, that doesn't mean you can just come up with a vague idea, write it up with no clue how to actually make the device, and get a patent. (Well, you aren't supposed to be able to--the examiners sometimes slip up!).
You don't actually have to have built a working prototype, but you have to be telling people in the patent how they can actually build the thing.
The examiners often slip up. That is really the nub of the problem. The examiners are incentivized to close cases, not to make the right decisions. So if you badger an examiner enough you can get them to sign off on just about anything, e.g. U.S. patents 6368227 and 7126691.
