The right for patents to exist is to stimulate innovation and innovation is believed to benefit the majority of the people. The stimulation of innovation is even generally believed to outweigh the negative effects of the monopoly granted. But why are patents thought to stimulate innovation?
If an individual or small enterprise makes an invention the inventor might need a bigger company to actually produce it. But if he tells about it, somebody else might just build it without giving anything back to the inventor. Thus there is no incentive to publish an invention. This is one thing patents try to change.
On the other hand, if the inventor produces the invention himself, a bigger company might take it and produce the invention as well, but sell it for a much lower price because the investment of building up a mass production is lower for a bigger company. This is the other thing patents try to prevent.
So there are good reasons to believe that patents on real inventions stimulate innovation. But with software things are different. If someone has written a piece of software he owns the copyright on that software. So he can go and sell copies of it and nobody else is allowed to do that. Not only is the software protected by copyright already, the investment of building up a mass production is zero. Copying software does not cost money. Thus there is no need to grand an additional monopoly through patents here.
Furthermore for software there is no difference between a prototype and mass production. The whole investment to be done is that of writing the code, which is something that person can do himself. If the idea how to write the code is not obvious nobody else can write a similar program as long as the source code is kept secret and only the binary versions are published.
There is absolutely no benefit of making software patentable. Therefore it should not be done.

It is true that software can be decompiled and analyzed, but this often is not significantly less of an effort than developing the same software from scratch, thus it does not lead to any special disadvantages for the original developer, when compared to the situation where two developers independently find the same solution to a certain problem.

This is crap: "If the idea how to write the code is not obvious nobody else can write a similar program as long as the source code is kept secret and only the binary versions are published."
Binary code can be analyzed / decompiled / reverse engineered to recreate similar source code.