In days of yore, mechanical inventions had physical embodiments and patent infringements were visible to the eye.

Fast forward to the age of information where methods and devices to solve problems with computers and information, where infringement is not readily visible. For example, the infringement upon a hypothetical method-claim, that is developed for a 3-letter government, agency probably does not have infringement detection 'visibility': think of it like a target having a radar cross section or radar signature.

That being said, if a claim does not have infringement detection 'visibility', does the claim ultimately contribute to the value of the patent? Are there any best practices \ strategies regarding the subject matter to improve visibility?

  • Infringement visibility does add value to a claim and whoever drafts and prosecutes the application should keep that in mind. One way to look at it is, if infringement of at lease one of the claims is detectable outside of the black box, other claims that require looking inside do add to the value of the patent.
    – George White
    Commented Aug 20, 2018 at 6:17

3 Answers 3


Please don’t consider the below legal advise. But I think there are three issues here:

  1. Is the damage caused by publication (of the patent) offset by the potential to obtain money by licensing/suing

  2. How should the claim be drafted

  3. How does one detect infringement in “hidden” cases.

Re 1. I would ask if you are sure that only that three-letter entity will infringe. Otherwise, there may be more value in maintaining (and selling) as a trade secret. Such licensing could include provisions which require them to reveal the use of the idea. (Will they fulfill it?). How much do you expect to be paid?

Re 2. It is worth considering if the output has a visible signature. For example, does the method have a particular form of output, including mistakes, or doe sit need a particular form of input or consumption of resources. It may then be possible to claim these more visible, but less idea-specific items. As an example from beer making (another process which may be kept secret), if you can find a signature of the process in the beer as some small amount of contaminant, then claim beer with with that contaminant.

Re 3. How long will the infringement go on for? The patent can be in force for over 15 years and one can sue 6 years back. Is it reasonable to expect that in all that time there will be no whistle blowers or inadvertent admission or perhaps a worker who left the agency or perhaps a disgruntled employee (all ways of getting hidden information)?

So, to be more specific about strategy. Which, of course, is very fact specific.

(A) See if trade secret protection is an option

(B) try to identify consequences, both upstream and downstream and see if you can claim these (should be done in any case)

(C) If your s/w will not be modified by client, consider artificially generating such consequences.

(D) Try to generate some contractual obligations to provide any type of information.

  • For this exercise assume that three-letter agency's contractors are motivated to infringe. Visible signatures are limited and expected to approach zero as new AI technology is developed.
    – gatorback
    Commented Nov 19, 2018 at 18:15
  • If a tree falls in the forest and nobody is there to hear, does it make a sound? If the infringers can commit to keeping their infringement a secret, there is nothing you can do. After all, even if you sue and get discovery, they could still lie. That is why the question is very fact specific. The strategy would be to charge them for the idea, not just its implementation. And you might as well file a patent, since you might find other users, who are not so dedicated to steal and hide.
    – tilnow
    Commented Nov 20, 2018 at 5:09

A true innovation can be claimed via both a process and an apparatus as independent claims. Very few apparatuses are so embedded that they have absolutely no external visibility whatsoever. Often an apparatus claim converts a process step to be a functor (i.e., converts a verb in the process-step to a noun in the apparatus-structure). Then that functor can be claimed via all-encompassing-enough language to claim all foreseeable alternative implementations of that could implement that functor/function-block. The top-priority danger is claiming an apparatus that is all or nearly all functors/function-blocks, which is disallowed. A lesser danger is claiming a means-plus-function, which will result in a patent of impaired scope and thus impaired value.

Claim-drafting strategy: Focus on claiming an apparatus that is anchored (analogous to the edges of a spider's web) to the externally-observable portions of the apparatus, then focus on working inward from those external-observable via functor/function-blocks so that it is the amalgamation of the collective that is the big innovation. Even the most esoteric function block is to some degree externally observable. The big obstacle to overcome is whether that amalgamated apparatus is itself innovative or is merely a rehashing of pedestrian existing practice. If the latter, then keep refining the functors/function-blocks until the truly-breakthrough innovation is captured in the collective of functors that made this invention possible and different from the prior art that came beforehand.


I'm not a lawyer so this answer is based on observation. It is true that in many cases with software the patent holder can't see the logic of an implementation without access to the source code. That said, if you provide software that accomplishes the same task as the patent describes you may simply be sued anyway. Then, they would be required to demonstrate you infringe. If your algorithm is different, that may seem to be easy, but litigation is expensive and many people would opt to just pay a license fee. There is an example of this from this Ask Patents question.

  • 1
    I think the burden of prove lies with the patent owner if suing for infringement, so technically you don't have to prove you don't infringe. But there's discovery to find evidence and you don't really want someone going through your stuff.
    – user18033
    Commented Aug 19, 2018 at 17:02
  • @DonQuiKong You make a good point. I've edited my answer a bit. I've had one of my patent attorneys state that once you have a jury trial it is pretty much a crap shoot no matter how clear you are from infringement since juries are mostly technically ignorant.
    – Eric S
    Commented Aug 19, 2018 at 23:05
  • Although my invention is not solely dependent on source code, it is possible to reverse engineer the source code from an executable (assuming you had access to it: not the case with three letter agencies) and compare it to a patent
    – gatorback
    Commented Aug 20, 2018 at 21:19
  • @gatorback One of the disadvantages of patents compared to trade secrets is you are required to disclose the invention if you use a patent. Google's PageRank, I believe, is kept as a trade secret.
    – Eric S
    Commented Aug 20, 2018 at 23:46
  • @Eric Shain Method for node ranking in a linked database patents.google.com/patent/US6285999B1 I think specific parameters are frequently changed. Only the "best mode" at the time of filing needs to be disclosed, not later optimizations.
    – George White
    Commented Aug 21, 2018 at 22:31

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .