Judge Posner admits he is no expert on what the fixes should be and his tentative suggestions for fixing the system are, in my view, decidedly mixed on their merits (e.g., specialized adjudications before the USPTO - remember when it was suggested that a specialized appeals court would improve the patent system and the result was a court that has been so maximalist in its approach to patents that it has in itself become a significant part of the problem).
So where to begin?
Legally, it has to go back to fundamentals and, for me, this has to go back to the scope of patentable subject matter and whether this should be defined to include software at all.
The Patent (and Copyright) Clause of the Constitution (Article I, sec. 8, cl. 8) provides that the Congress shall have the power "to promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries." Note that, in defining this as one of the enumerated powers of the federal legislative branch, the Constitution does not mandate that the legislature provide for patent protection of any sort. It merely permits the exercise of such a power within constitutionally prescribed limits. Thus, any legitimate exercise of patent authority in the U.S. must come from Congress and must respect the constitutional bounds that any grant of patents be for "limited times" and be done in such a way as "to promote the progress of science and useful arts." Legally, then, any patent system in the U.S., if adopted at all, must be authorized and defined by Congress with a view to promoting the progress of science and, implicitly, must employ "limited times" consistent with what it takes to promote scientific progress.
The first issue, then, is whether patents are needed at all to promote the progress of science. In the U.S., in spite of philosophical arguments to the contrary by Jefferson (http://news.ycombinator.com/item?id=1171754), this has never been seriously in dispute. The industrial revolution was already well in progress in 1789, when the Constitution was adopted, and the federal authority, though generally regarded with great wariness at the time, was seen as vital to protect the rights of inventors and to reward them with limited monopoly grants in order to encourage the progress of science. In the first U.S. Patent Act (Act of April 10, 1790, 1 Stat. 109, 110), Congress implemented its constitutional authority to sanction patent monopolies by defining patentable subject matter very broadly, to include "any useful art, manufacture, engine, machine, or device, or any improvement therein." Congress amended the Act in 1793 and then again in 1952, so that today it reads as to the idea of "patentable subject matter" as follows (35 U.S.C. sec. 101): "Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title."
Thus, patents in the U.S. can be granted for any original invention that fits within the definition of patentable subject matter and that also meets the other conditions of the patent act (i.e., that is useful and non-obvious). Note, though, that the 1952 definition of patentable subject matter significantly expanded the scope of such subject matter in the name of bringing the patent laws up to date with developments in then-modern technology, all in the name of promoting the progress of science. It did so by defining patentable subject matter to include any "new and useful process" as well as any "new and useful improvement" of any original invention. Over time, "process" has come to embrace business methods and also software. And the protection of "useful improvements" made clear that new uses of existing machines or processes could be patented notwithstanding older Supreme Court decisions such as Roberts v. Ryer, 91 U.S. 150, 157 (1875) ("it is no new invention to use an old machine for a new purpose").
To promote the progress of science, then, Congress in 1952 allowed patents to be granted for any inventive process and for any inventive new use for any such process. In my view, this generally made sense for what was essentially the continued playing out of the same sort of industrial revolution that animated the original forms of patent protection granted in 1790. Looking at that language at that time, one could readily make the case that patentable processes and improvements thereon could and did promote the progress of science. Discrete inventions tended to be sharply differentiated and tended to involve significant development effort in time and resources. An inventor could keep a process secret and not patent it but the grant of a limited monopoly gave a decided inducement to disclose it to the world and, hence, to expand the broad pool of scientific know-how available to society.
Then came the digital revolution and, with software, a new or improved process can amount to an almost trivial variation on prior art amidst a seemingly endless stream of improvements developed in ever-rapid succession and with little or no capital investment beyond what developers would be motivated to do for reasons entirely independent of gaining monopoly protection for the fruits of their efforts. Moreover, there is little that is closed about such innovations: a wide knowledge base is in place, known to an international developer community that is basically scratching its collective head asking why it should be restricted legally from using techniques and processes that represent common knowledge in the field.
The main question, then, concerning software patents, is whether the existing framework makes sense as one that promotes the "progress of science" insofar as it grants patent protection to process inventions in this area. Congress needs to seriously ask itself that question. A second question, also tied to constitutional authority and assuming that it is legitimate to grant some form of patent for such inventions, is whether a 20-year period of exclusivity makes sense in an area where innovation occurs at blazing speeds and with not too much capital investment tied specifically to any given discrete invention. Is that necessary to promote the progress of science? That too is a question that Congress needs to consider.
Thus: (1) there is nothing magical about the current definition of patentable subject matter and Congress can adapt this to suit the needs of the time in promoting the progress of science, (2) process patents are in themselves a fairly recent phenomenon (at least in any large numbers) and it is no radical change to curtail them in areas where they make little or no sense in light of the constitutional purpose for why patents even exist in the first place, and (3) legitimate patent reform needs to go far beyond procedural fixes around the edges of the system and needs to focus on the realities of modern technology and whether the patent laws further or impede the progress of science as applied.
The policy debate can and will go all over the board on this but, if it is framed in light of the constitutional foundation for having patents in the first place, it can be shaped in a way that puts the focus on the fundamentals of what needs to be fixed as opposed to lesser issues that do not get to the heart of the problem. The main problem today is the blizzard of vague and often useless patents in the area of software. These are effectively choking all sorts of innovation and are benefiting mainly lawyers, trolls, and others who do not further technological development by what they do. It is a mistake, in my view, then, to swing too broadly in trying to fix things (as by advocating abolition of all patents) or to be so timid about the issues that reform is marginal at best and ineffective in dealing with the current crisis of an over-abundance of essentially worthless patents. Congress embraces the patent system as a whole and shows no hostility to its fundamentals. Reform must be shaped in light of those fundamentals but it must, at the same time, be meaningful to eliminate the main garbage from the current flawed system. Judge Posner has pointed the way generally and proponents of reform ought to follow his lead, with the focus being (in my view) on software.