It's a lot of work that could be spent productively doing something else, so I've stopped patenting stuff. I think it just isn't worth it.
We prefer to have things as "commercial secrets" and make them sufficiently obscure that by the time someone reverse engineers them either we have an improvement in place, or we have enough of the market that it's not worth their effort to go through implementation.
If you go the patent route then you have to have effectively full disclosure. To enforce your patent you have to detect infringement, issue "Cease and Desist" orders, then pursue through the courts. If the person infringing is in another country then the only benefactors will be the lawyers. Anyone who infringes will make money, then disappear.
That's for my fields, and YMMV. The only people who have ever cared about patents are those who have money to invest, and don't actually understand the business. They want to see their investment protected, and think patents will do it.
There is now one exception to this, and that's for an innovation that we think can't subsequently be upgraded. You need to follow through the decision tree, with assigned probabilities, and the outcomes. Basically, do some Game Theory.
In practice when big companies file patents they make sure to leave out enough information to make the patents as unhelpful as possible. If you were trying to replicate my work it would be pretty close to useless to use my patents.
I would guess the general opinion in the HN-crowd is that ideas are worthless and that execution counts. Well, patents are "just" ideas. So, are they worthless? If so, why does everyone keep asking about them?
I love open source and patenting doesn't sit well with me. The idea of a'commercial secret' is more appealing because IMHO its more about competition and as a startup you can always out innovate your competitors.