Impossible Research

It's actually an understatement that some compare software patents to a minefield. Detecting a landmine is dangerous, yet it is feasible. The reliable identification of conflicts with software patents is simply impossible.

The European Patent Office has already granted about 30,000 software patents. That number is the result of various full-text researches performed by software patent critics. Even as the European Commission presented a proposed directive in the February of 2002, it admitted the existence of 20,000 European software patents.

If you write a program and want to ensure that you have no conflict with any of those numerous patents, you have to read every single patent. The law says that everyone has the obligation to know what's in the patent register. If you say that you didn't know you violated one of those patents, all that they'll tell you is that "ignorance is no defense".

It's not possible to focus on a smaller selection of patents. The exorbitant number of software patents would not only be half as bad if you could at least narrow your search down. If you write a particular type of program, it would be better to only check on conflicts with a selection of maybe 100 patents that are relevant to your area. Unfortunately, that doesn't work.

There is no consistent terminology that allows you to search for patents with keywords. Of course you can search for certain words, but you won't get all of the patents that you may have a problem with. Your database may violate a patent that only contains words that have to do with programming languages. Your e-commerce software may infringe upon "telecommunications" patents, and so forth. So if you wanted to be certain that you have no problem, you'd have to study them all. Each patent is many pages written in a very hard-to-read language that programmers don't use in their day-to-day work.

In certain other fields than software, a so-called nomenclature makes it easier to search for patents. For instance, if you find out that a certain chemical agent cures a certain disease, then you can run a patent search based on the unique Latin names of the agent and of the disease. Those names are consistent around the globe. In the field of software, that is unfortunately not possible.

Small and medium-sized enterprises surrender at the sight of tens of thousands of software patents. Knowing that they'll never be able to research them all, they give up. Consequently, patents are an uncontrollable business risk to them.

Large corporations don't check on all patents either. They may do some selective patent searches to monitor what some of their competitors are doing. However, even the largest corporation can't ensure that none of its products infringes upon any of countless patents. It would be too costly to try, and it wouldn't really help. Even if you look at all published patents, there will always be new ones that are in the application process and have not been published yet (that only happens after 18 months).

As a large corporation, you simply count on your legal department. If any problems come up, they have to solve them. If any company that has products of its own brings up a patent infringement claim, then the large corporation sees how it can use its own patents against the other company's products, and ultimately they'll agree on a cross-licensing deal. If the large company gets attacked by a patent profiteer who has no products, and if that patent profiteer has a strong legal basis, then they work out some payment that doesn't really hurt the large corporation and takes care of the profiteer.

Click to read why software is an intellectual achievement that should not be patented