Hacker News Reader

Comment by tgflynn | Hacker News Reader

It seems to me that one of the biggest problems with patents is that it has become possible to patent the "what" instead of the "how". In most cases the "what" is obvious, it's the "how" that's hard.

An example would be the Amazon One-Click patent. That should never have been granted because the "what", ie. the basic idea, is obvious and once you have that the implementation is trivial.

To get a patent you should at least have to describe a method for solving a non-trivial problem and prove that it actually works.

View on HN

I thought "how" was the whole point. Imagine I invent an air conditioner that's 100x more effective or efficient based on a new mechanical process I invented. I damn well better be able to patent that "how" despite a century of prior art in "what".

Yes, but that's the opposite situation to many software patents. If air conditioners didn't yet exist you shouldn't be able to patent just the idea of an air conditioner without coming up with an effective design for one.

The One Click patent would be like patenting just the idea of an air conditioner because once you have that idea any software developer worth his salt could implement it.

Except it wasn't: Amazon--like everyone else filing a patent--had a duty to bring the idea to market within the time frame that US patent law required, which they then did. _That_ is the how.

Demanding the patent goes into the specific details on the "how" gets us things like "they implemented it in JS but we used TS so that's legally distinct, this patent doesn't apply to us" (ignoring how idiotic software patents are, and how stupidly long they are allowed to be active for) or "they made their machine using sheet metal and distinct PCB components, we used plastic and an FPGA, this patent doesn't apply".

If you get a patent granted, and you sit on it, you lose that patent. If you make real the ideas/things described in your patent within the required time frame, your patent "kicks in" and you get to sue others for copying your idea, even if their specific realization of that idea differs from yours.

Reducing for how long patents are granted, disallowing "tweaking-a-thing to renew the patent", and either drastically cutting down the duration of, or entirely canning, software patents, would be a nice move though.

> Except it wasn't: Amazon--like everyone else filing a patent--had a duty to bring the idea to market within the time frame that US patent law required, which they then did. _That_ is the how.

That's not proof of non-obviousness. And even if it were, it's still not a reason to allow 'what' patents (which the law actually doesn't allow! But they get granted anyway).

> Demanding the patent goes into the specific details

You are arguing a strawman. A patent must describe the invention in sufficient detail for someone skilled in its art to reproduce. Incidental details shouldn't be (and I believe aren't, legally. What the USPTO actually does in practice is a different matter) grounds for calling an alternate implementation non-infringing.

What you're describing is a hideous perversion of the patent system. It allows patenting any obvious [1] idea so long as you are first to file [2]. Alternately, if the idea is novel and non-obvious, it grants patent protection without requiring disclosure of implementation. It basically grants patent protection to trade secrets.

[1] A test for non-obviousness: does an implementor care to look at your patent or reverse-engineer your implementation, or is a description of what the invention does sufficient?

[2] And eventually implement.

> [1] A test for non-obviousness: does an implementor care to look at your patent or reverse-engineer your implementation, or is a description of what the invention does sufficient?

This seems like a wonderful test of obviousness to me.

However, the "what" could be: "A method of reducing the number of online sales that are attempted but somehow aborted".

The "solution" of having a single click to purchase, in the context of cached purchase info and permission, might not be obvious to everyone working on that problem. (Even though it might be obvious in hindsight - which is not a barrier to patentability.)

That "solution" is specific enough to not block other's from solving the same problem (avoidance of aborted purchases) in via different methods.

That would be the approach I would use to defend single-click against your test. (Not a lawyer, no special legal expertise, just working through the logic as I can see it.)

> the "what" could be: "A method of reducing the number of online sales that are attempted but somehow aborted".

That's a good point. But it does open the door to patenting so many things it would make business impossible - store layout? Ad composition? Sales timing? Employee treatment?

Fortunately the law does not seem to have such a broad view of what is patentable: https://en.wikipedia.org/wiki/Patentable_subject_matter#Unit...

The other problem is the 'could' in your hypothetical - as far as I know, that's not what Amazon put on the patent application. Probably because they knew it is not patentable subject matter.

There's no requirement that you actually implement a patent to keep it. The patent troll industry runs to a large extent on patents that aren't being practiced.

You may be thinking of trademarks, which can be granted on the basis of an intent to use the mark in the market with a time limit thereafter to keep them.

"had a duty to bring the idea to market within the time frame that US patent law required, which they then did"

Do you have a reference for that ? I've never heard of any such "duty". There are many patents that have never been implemented yet remain valid.

Beyond that my point is not that they didn't show the "how" but that the "how" in that case was absolutely trivial and completely undeserving of patent protection.

Reply on news.ycombinator.com