Updated 16 Jan 2022.
Patents Need to be Detectable.
This may appear to be an obvious statement, but people pour money into patents where infringement simply cannot be detected. Here’s a case where undetectable software claims were thrown out of federal court.
When looking at an invention (or an issued patent), one of the first questions should be: “Can I tell that my competitor is using this invention?”
If there is no way to tell that a competitor is performing the same method or manufacturing the same product, how could that patent ever be enforced? The answer: it cannot be enforced and is therefore worthless.
I don’t mean “not as valuable,” I mean completely, utterly worthless. In fact, the patent may have a negative value to a company.
Most software patents are awful and not just because they protect software. Most software patents are awful because the patent owner could never tell that a competitor uses their invention.
Consider a gloriously complex and innovative artificial intelligence method for analyzing a bunch of data and returning a result. The algorithm runs deep inside a competitor’s datacenter, where we will never have access.
Could we ever tell that the competitor is using the exact same algorithm as in our patent? No. That patent is worthless and the invention would have better been kept as a trade secret.
Method of manufacturing patents are worthless
Another classic example is method of manufacturing patents. Methods of manufacturing are notoriously difficult to detect. You will never have access to your competitor’s factory, so you will never know if they are using your process.
I looked at a company who had made enormous advances in their manufacturing system. They were able to run their equipment ten times faster than their competitors. Their main competition came from China, and for this US manufacturer, their process improvements were the key that made them competitive.
Shortly after this patent was published, the Chinese manufacturers started selling at a price much lower than ever. We suspected that they started implementing the trade secrets found in the patent.
However, we had US patents – but the competitor was overseas.
Method patents are infringed *when* and *where* the methods are performed.
Because the methods were performed overseas, we could never enforce the patents. Even so, we would never have access to their manufacturing facilities. We could never walk through their factory, check their temperatures and pressures of their equipment, and verify that they were following our process.
Sadly, writing a patent for this invention eviscerated the value of the company. They gave away their competitive advantage *because* they got a patent.
Undetectable infringement makes a patent worthless. However, sometimes things are detectable but there are different levels of difficulty.
Different Levels of Detectability
One end of the scale are inventions that are so compelling to a consumer that a competitor would advertise that they have the feature you patented. These features are so important – and so valuable – that a competitor advertises when they infringe.
The next level down might be something that you can tell through casual inspection. Take a product in a retail store. If we can visit a store and see the product on the shelf and know – without any further work – that it copies our invention, then it is easily detectable.
Sometimes, it may be still possible to tell, but it takes a bit of work.
For example, a patented circuit on an integrated circuit chip can be verified by slowly, methodically slicing each layer off the chip and tracing every single circuit. This is the other end of the scale, where it is extremely difficult but theoretically possible to tell when someone copies the invention.
The question about detectability is a deal killer when the invention is completely undetectable. It is an absolute, unquestionable, deal killer. That patent is worthless. In fact, it is worse than worthless because you essentially gave away that innovation and got no enforceability in return.
If an invention is difficult to detect, but you are sure you can detect it, then the invention just needs to have enough value to make it worth the trouble. In the silicon chip example, that painstaking process of reverse engineering the chip might be brutally expensive, but for very high value inventions, it may be worth the expense.
Part of the problem with detection is that you need to know where and when to look. Many times, a startup is deeply engaged in its market and knows all its competitors. They go to the same tradeshows, talk to the same customers, and knows their products inside and out.
One example might be downhole oil field devices, such as drill heads. Even though they are impossible to see when you drive past an oil rig, there is a tight-knit community where everyone knows everything a competitor is doing.
 There are tricks for dealing with this situation, such as writing a patent around the detectable effects of the algorithm, or focusing on the interfaces within a software product (an API, user experience, administrative user interface, etc.)