All fields are broadly similar, and any two fields are also
different.  Because fields are different, patents have different
effects in different fields.  Thus, it is easy for patents to be
beneficial on the whole in one field and harmful on the whole in
another field.

One way in which fields vary quantitatively is in how a patentable
technique relates to a usable product.  Software is at one extreme;
patents nearly always cover small subunits, many of which are combined
in a new way to make a new and useful program.  At the other extreme
we find pharmaceuticals, where a patent covers an entire chemical
formula as a whole.  A patent on a chemical formula is much closer, in
its effects, to software copyright than to software patents.

If the aim is to promote progress, we should treat each field in a way
that promotes progress in that field.  What other fields require is
beside the point.  I don't know whether patents are beneficial in
other fields which I have never practiced, though The Economist thinks
so.  But we need not know the answer to this question to decide that
software patents cause trouble.