A patent must publicize what is being claimed: you are given a explanation of the design or process. You just can't make it and sell it. However, by knowing someone's design, even when patented, it can inspire someone else to build something even better.
Have you ever read a software patent ? The ones I have seen have been written in a very strange way (ie use of non-standard terminology for technical concepts) which seems aimed at obscuring the nature of the ideas being patented rather than clarifying them. If patent descriptions were really intended to convey useful technical information they would read more like journal articles rather than some strange dialect of American legalese.
I'm sure that some innovation occurs through efforts to work around existing patents however I doubt that effect comes close to matching their effect of preventing ideas, even if independently discovered, from being incorporated into new technical solutions.
As for the notion that people are going to read 20 year old expired patents for their ideas that seems pretty unlikely to me given the pace of technological development.
Say you come up with a great new idea in software. And you want to make an agreement with someone to implement this idea. Without patents, they can have "their uncle's company in China" to do the idea (think CrunchPad), rename it to avoid any trademark issues, and screw you. How would you protect against this? This issue faces almost every inventor of an idea, as they will have to make business agreements with someone. Thus, although you would be encouraging free flow of information within the tree, you would be cutting off a large source of it -- individual inventors.
Patents won't be going away because it's hard to distinguish between the above example, and a patent troll.
Personally, speaking as a "little guy' who is interested in advancing technology and making a good living doing it, I am far more worried that some corporation will be able to claim patent "protection" over my ideas. This seems far more probable to me than the inverse situation of someone else independently discovering my invention and beating me in the market. Also even if that happened I'd much rather compete in the market than in a court room.
As for the specific case you mention, I think, especially in software, that it's becoming far easier for an individual inventor to take his invention to market rather than needing to sell it to a large corporation. Also this argument seems largely based on the notion, which I think is most often incorrect, that most value lies in a simple idea rather than in the execution of that idea. In other words that you have some idea that can be stated in a few English sentences that has tremendous value and that no one else has thought of. You only need divulge that idea and anyone on the planet will be able to beat you in the market. Such ideas, if they exist at all, are exceedingly rare, maybe once in a generation occurrences. All recent examples of highly successful innovation: Microsoft, Google, Facebook relied far more on the quality of execution - the construction of solutions from many small ideas - than on one overarching genius discovery. And it is exactly this form of value creation which patents impede.
That's why unless we figure out how to distinguish this effectively, either patents get abolished in software or they persist.
Personally I did invent something that's relatively easy to copy. How would a big company ever agree to buy what I made when they can just copy it?
As for your second example, where you're trying to sell your idea to a company, I wonder if you aren't overestimating the amount of protection the patent will give you ? Are you considering the asymmetry in legal resources between yourself and the company ? There are many grounds for finding software patents invalid and if the idea is valuable enough, the corporation may well litigate you all the way to the Supreme Court. Do you have the resources for that ?
The question is do patents actually promote the progress of science and technology ? I suspect that they do not because while they create additional monetary incentives to pursue innovation at the same time they restrict the free flow of ideas which is the driving force behind scientific and technical progress.
The free exchange and use of information is probably the source of the exponential progress in "information technologies" that Ray Kurzweil has discovered. It seems likely that impediments to the exchange and use of information, such as patents, can only reduce the rate of exponential growth.
Perhaps if patents had never existed our technology would be far more advanced than it currently is.
If it were ever possible to prove conclusively that patents impede rather than promote progress, they would become unconstitutional in the US.