Even those that promote patenting algorithms should agree that patents purpose is opening the details of a machine or process in exchange for a monopoly.
That was the purpose of original patent's plans, that everybody could replicate your machine or process with the information provided.
When you want a monopoly and also ban publishing the details of the algorithm or working examples, something wrong is happening.
It has gone too far, with people recommending other people NOT to read patents, because it will be bad for them on court.
The reason that companies tell employees not to read patents is that knowingly infringing a patent is triple damages (as opposed to just accidentally reimplementing something).
I've read a bunch of software patents, with attorneys helping me out with language and advice, but this was under the guise of "How can we do a product like X, but not infringe on the related patents?" The patents in question were very broad, nearly without exception either ridiculously obvious or adoption of techniques that I found in textbooks and papers from conferences. It was almost like the company producing product X had bribed the PTO, it was that bad.
I believe that nearly every piece of software exceeding a few hundred lines of code infringes on someone's bullshit software or business methods patent. And of course, you can't go reading all the patents, so every software project is pre-screwed. Shipping software these days is a matter of ignoring the issue, crossing your fingers and hoping that a patent troll doesn't come knocking.
The PTO largely doesn't seem to care or do any research beyond the bare basics. As far as I can tell they're not really incentivised to do so either.
They collect a whole heap of fees for getting and maintaining a patent (http://www.uspto.gov/learning-and-resources/fees-and-payment...), and even claim a fee for revisiting a patent, but so far as I've been able to find, it doesn't cost them anything to actually invalidate one. It ought to cost them all the money they've taken so far, and then some on top, to provide them with more of an incentive to make sure the patent is correct.
That was the purpose of original patent's plans, that everybody could replicate your machine or process with the information provided.
When you want a monopoly and also ban publishing the details of the algorithm or working examples, something wrong is happening.
It has gone too far, with people recommending other people NOT to read patents, because it will be bad for them on court.