[throwawaykf03]

I encourage you to experience a patent prosecution yourself to test these claims. Fortunately, you can do it vicariously completely for free! The entire prosecution history of a patent, i.e. the arguments back and forth between the applicants and the examiner is available for free on the USPTO public PAIR website. Go to that site, enter an application number, then click on "File Wrapper".

It will be immensely boring reading. But you will see all the work that goes behind a patent application. The File Wrapper will have tons of documents, mostly boilerplate, so you want to look for documents titled Office Actions, (Non-Final Rejections, Final Rejections, Allowances) and Applicants Arguments in response.

You will see almost everything, including the evolution of the claims from application to issue, the search strategy the examiner uses, the prior art references presented, the rejection issued, and the applicants responses on 1) how the prior art does not apply or 2) amending the claims to sidestep the prior art.

There obviously is variance in quality of examiners, but typically they're a tough bunch to get anything past. I would echo everydayman's sentiment, that while they are they not all technical experts, they are very good at search and they do find stuff pretty well on average. They do have pressure to get rid of cases ASAP, but their default is to reject. I have seen more frivolous rejections than I've seen frivolous allowances.

[beagle3]

I have been granted patents that most people on HN (myself included), would reject based on obviousness or lack of novelty, to anyone skilled in the art. There was one back-and-forth before grant.

I did not write the patent myself - I described it to our patent-writing-guru-for-hire, and he wrote it in patentese (which is compatible with, but slightly more readable than plain legalese). He also echos yours (and everydayman's) sentiment that those guys are tough. But he also has a 95% success rate, with rarely more than two back-and-forth iterations.

[everydayman]

> he also has a 95% success rate, with rarely more than two back-and-forth iterations

Just to be clear, this is actually 2 distinct data points: 1) 95% success rate 2) rarely > 2 rounds

First on your #2: any software patent[1] that was obtained in the last ~10 years AND was granted with <= 2 rounds is either: a) truly innovative and from out of left-field (READ: extremely rare) b) the original Claims were very narrow c) the Claims were significantly narrowed during prosecution

Now combine this with your #1 and what we can likely ascertain is that your "patent-writing-guru-for-hire" either doesn't try to get a decent patent for his clients, doesn't know what he's doing, or allows his customer to dictate that time-to-completion outweighs quality.

When you write code, do you get credit for how fast you can get the thing compiled and out the door - or do you take pride in the end result is not only to spec AND well thought out AND thoroughly QA'd AND provides advantages in running it? Yet, with "95% success rate" and "<= 2 compiles" is basically asking your "patent-writing-guru-for-hire" to sacrifice quality for speed in getting the job done. Make no mistake, the english words used in patent Claims is as serious and deliberate as the writing of any code - a wrong placement of a comma, period, semi-colon, choice of words, or ordering is on the same level as a misplaced bracket, for-loop, choice of data structure, etc... At some point, they'll likely come to bite.

Take coding as analogy - anyone can write code that compiles AND then claim he's got 95% success rate at compiling - but the question is whether the end result is worth the time, effort, and expense spent. Just as there is a justification for top coders to be paid upwards of $150k/yr despite the existence of $5k/yr alternatives: QUALITY.

If you are a startup with real money poured into true R&D (e.g. where many trials and experiments and dead-ends were required to achieve the innovation) that can then be easily duplicated and you determine that you need solid protection - then investors demand you to get some defensible attribute - patents offer this. I suggest you simply won't get that with someone who claims a 95% success rate in <= 2 iterations. Impossible.

If you still don't believe me, then give me your patent number and I'll detail you some simple workarounds that competitors can (and will if need be) use to completely avoid your patent - aka workaround.

[1] technically, there is no such thing as a software patent

[beagle3]

All agreed.

Except the company for which this technique was developed (and patent was written) was acquired, partly on the basis of this cluster of patents (which the buyer believed would give them ammo against the leaders in the field). So the guru was right in the path he took.

I will not give the patent number because that will expose me (no thank you), and I'm aware of workarounds. The thing is, That's true of every software patent[1] I've seen that is not mandated by some standard. Do you have counterexamples?

[1] there's no such thing as a software patent, of course.

[everydayman]

Ok good to hear - the patent(s) facilitated investment (e.g. acquisition of the startup) as intended.

There are workarounds to just about every way of doing something. The question is how feasible and practical the workaround is.