Developing a large and complex program means combining many ideas, often hundreds or thousands of them. In a country that allows software patents, chances are that some substantial fraction of the ideas in your program will be patented already by various companies.

Perhaps hundreds of patents will cover parts of your program. A study in 2004 found almost 300 U.S. patents that covered various parts of a

single important program. It is so much work to do such a study that only one has been done.

Practically speaking, if you are a software developer, you willusually be threatened by one patent at a time. When this happens, you

may be able to escape unscathed if you find legal grounds to overturn the patent. You may as well try it; if you succeed, that will mean one less mine in the field. If this patent is particularly

threatening to the public, the Public Patent Foundation may take up the case; that is its specialty. If you ask for the computer-using community's help in searching for prior publication of the same idea, to use as evidence to overturn a patent, we should all respond with whatever useful information we might have.

However, fighting patents one by one will never eliminate the danger of software patents, any more than swatting mosquitoes will eliminate malaria. You cannot expect to defeat every patent that comes at you, any more than you can expect to kill every monster in a video game: sooner or later, one is going to defeat you and damage your program.

The U.S. patent office issues around 100,000 software patents each year; our best efforts could never clear these mines as fast as they plant more.

Some of these mines are impossible to clear. Every software patent is harmful, and every software patent unjustly restricts how you use your computer, but not every software patent is legally invalid according to the patent system's criteria. The software patents we can overturn

are those that result from "mistakes," where the patent system's rules were not properly carried out. There is nothing we can do when the

only relevant mistake was the policy of allowing software patents.

To make a part of the castle safe, you've got to do more than kill the monsters as they appear -- you have to wipe out the generator that

produces them. Overturning existing patents one by one will not make programming safe. To do that, we have to change the patent system so

that patents can no longer threaten software developers and users.