.

We start with the basic reason for allowing a patent system to exist at all. Patents are society's gift to inventors. But society is prudent and gives only when it has no choice. When there is a magical recipe behind a valuable product, for example the alchemy that produces a silver mirror or a mug of cold beer, it's worth granting a short monopoly to extract that alchemy from its inventor. But when the product is wholly self-describing, when society can see exactly what is needed to recreate and copy it, that monopoly is worthless to society.

1. A large number of software patents, and all business method patents, can be excluded on first principles: there is no justification for paying for something you're going to get for free anyhow and anyone who sets the rules to make society pay for things it does not need is engaged in a fraud.

The patent industry replies that the patent is a “reward for innovators”, meaning, “give the inventor his monopoly even when you don't need to, and he's more likely to carry on inventing”. By granting monopolies when they are not justified, we immediately teach the inventor to continue demanding “rewards” for useless disclosures. Real inventors are not motivated by patents at all, it is simply a fair trade for their alchemy.

2. The proper reward for innovation is market advantage. Patents do not drive innovation, they can only, in the best case, drive disclosure of secrets.

The patent industry retort is that patents represent value, like gold, both in the patent offices' databases, and in industry's portfolios. This might be true if only valuable alchemies were recorded for posterity. Software patents, being mostly useless, describe no alchemy except that legalese required to work around the rules which, in theory, ban these things.

3. By allowing patents on software, patent offices have corrupted their own databases, probably beyond repair. They are failing on their very first job – to collect valuable alchemies on society's behalf.

The patent industry points to the large settlements for software patent infringement and says, “see, we created all this value. Give us more freedom and we'll create even more”. Lawyers are the first to see litigation and conflict as “value”, but for society this is a pure friction cost. A good property form should result in fewer, not more, lawsuits.

4. Software patents cause lawsuits because they are often vague, impossible to avoid, and granted for obvious ideas. They are a pure friction cost.

The patent industry finally points to some cases where patents have helped small firms with products (not trolls!) fight off large monopolists. These cases are remarkable for their scarcity. But even then, how can the software industry use patents that are too costly for startups, arrive only after seven or more years, and which last for twenty more?

5. Software patents are too late, too expensive, and last too long. They just don't fit the needs of the software industry. A more suitable protection would be instant, cheap, and last only for a few years, and it would drive the disclosure of trade secrets.

Finally, the patent industry gets annoyed and says, “we don't patent software in Europe at all!” And this is true. What is patented is not software, but a kind of legal fiction 'machine' who's only function is to secure the patent, control the market, and keep the real alchemy as secret as possible.

What is patented by Microsoft is still software, in the same sense that what patented by BASF is chemistry.

Um, yes, that's a tautology. But there are several big differences:

1. Chemistry is limited by the laws of physics, software is not, it is limited by the limits of Turing machines and, because of this, are mathematics and pure thought.
2. Software does not have a specific form, can be represented in an uncountable number of ways and generally can be applied to infinite domains. The chemistry example you have has two parts, a process for synthesis and a final product. You may be able to avoid a patent on a particular chemical synthesis but you will not be able to avoid the design patent on the final product because is has a specific, well-defined form since it is made up of the same units: atoms. The same cannot be said about software. Take for example the "backing store" patent, the algorithm can be expressed in an uncountable number of ways and the final products take no particular form. Although you are not supposed to be able to patent "total concept" this is exactly what software patents grant.
3. Software patents serve as a benefit to only two parties: patent lawyers and wealthy patent portfolio holders. Otherwise they are a needless drain on society because they consume financial resources that has to be eventually be paid for by the market without actually benefiting society. They are black holes: monies pour in and nothing comes out. If you're about to counter this by saying that "something comes out: we got the software", no, we would have gotten the software anyway. The problems that the software solves would have existed anyway, somebody, or more accurately, somebodies would have created one or more variants of the software that solves the problem. Society gets no benefit, software patents should be voided and all patents that smell of software patents rejected as a matter of procedure.

Also, I thought it was illegal to patent things that are OBVIOUS "inventions"

It's not "illegal" per-se, it is just that a patent application for something somehow deemed obvious should be rejected.

what if someone once decided to patent the wheel, or the book?

An Australian patent lawyer in fact did patent the wheel a few years ago to point out that the Australian patent system is broken to the core (he's never claimed anything on his patent). This is known in the "the biz" as "total concept" and is part of the litmus test for patentability. But it is EXACTLY this (total concept) that is being patented in the case of software patents.