The majority of the patents I've looked at read to me as outlines of research projects rather than detailed descriptions of how to make something. (I am a person of ordinary skill in the art; I have a degree in the field)
Typically the components cited are "motors" or "controllers" or "actuators". Now, there are two issues here:
Broad claims that establish the territory that is claimed.
Enablement of those skilled in the art to make the invention.
This language is appropriate for (1), but it is way too vague to satisfy (2).
I would consider reciting "a motor" without further details, to automatically mean "undue experimentation", owing to the huge number of conceptual types of motors with widely differing properties, not to mention the sheer number of different components with "motor" in their name that may be bought. (For example, Google patent search gives at least a 1000 matches for US utility patents with "motor" in the title.)
So, clearly, the "undue experimentation" is an enormous grey area.
My question is: why does the patents office errs, in this grey area, on the side of the inventor, rather than on the side of the public?
Looking for answers here: http://www.uspto.gov/web/offices/pac/mpep/s2164.html (2164 The Enablement Requirement)
Some relevant excerpts:
The test of enablement is not whether any experimentation is necessary, but whether, if experimentation is necessary, it is undue.
This can be interpreted by the examiner to say that the inventor shouldn't hold back the details. It's easy to disclose them, right? And if you, the inventor, don't disclose them, the experimentation will be considered undue. (This interpretation is within the English meaning of "undue".)
Time and difficulty of experiments are not determinative if they are merely routine.
It evades me why this statement was considered acceptable to include. It opens up the system to abuse, for example if I specify that "to determine parameter X, the practitioner should try all possible combinations of [some mathematical structure that has very many combinations]" in some routine way, where I have already done this, and know the parameter X. Preferably, the law should compel the inventor to disclose this "shortcut". (Note that this is not the same as best mode, because they will arrive at exactly the same answer.)
And the example that follows the above quote:
$50,000 and 6-12 months
is considered not undue, but
many months or years
Here is a court ruling that enablement has not been met because
“an unreasonable amount of work would be required to arrive at the detailed relationships appellant says that he has solved.”
This is similar to what I'm saying above: the inventor is making it more difficult for the public to practise the invention, by withholding details that they clearly know.
But again it highlights the greyness of the grey area, by its use of the term "unreasonable" rather than quantifying what that means in terms of dollars and months.
This is my complaint repeated in the official text:
Thus, the teachings set forth in the specification provided no more than a “plan” or “invitation” for those of skill in the art to experiment using the technology in other types of cells.
Once again they avoid facing the issue of how much:
No exact numerical standard has been fixed by the courts, but the “amount of required experimentation must, however, be reasonable.”
Note the word "reasonable" rather than numbers of dollars and months.
Some numbers are quoted, however:
- 4 hours — reasonable
- 1 to 2 years — “a clearly unreasonable requirement”
I guess, given the "6-12 months" above, this means that 1 year is our standard.
And this seems too high to me. Why does the patent office not stipulate this to be, say, 1 week?
This would be in the public interest, improve the quality of patents, and reduce patent abuse. Would it have any negative consequences? I can't think of any.
Another take on this is that the effort (which we can count in the common denominator of dollars, or months of work for one person; which are similar), the effort to create a working solution can be seen as a set of steps, some large, and many small. The law seems to require that each of these steps has a predictable bound, but is reluctant to specify an actual quantitative bound. What is the logic behind this? What does such a shy strategy achieve? Especially in light of the fact that the inventor, if enablement is taken into account, must be in full possession of the exact recipe for at least one embodiment. So to require them to disclose it does not seem to be undue from the public's point of view. After all, a patent is a contract which exchanges a monopoly for a full disclosure, and such full disclosure is easy to make.
This is not the same as asking the inventor to provide a working prototype (clearly, this is often unreasonable), but they should be able to provide a parts list, and a means they would use for obtaining each part (e.g. citing a distributor of a component). So while it may be beyond their means to actually implement a prototype, the test for "invention" rather than a "plan" should be that a specification refers to actual examples of parts and methods (while claims should be allowed to refer to generic parts). In case of software patents, source code should be provided (into the public domain) as part of the specification, while abstract high-level pseudocode used in the claims. I'm saying this from a commonsense perspective. Clearly, the established convention disagrees. And the question is why?