How to Read a Patent | Hacker News Reader

How to Read a Patent

adlervermillion.com

202 pointsteachingaway29 comments

Open articleView on HN

A slightly more humorous take on the same thing:

http://www.danshapiro.com/blog/2010/09/how-to-read-a-patent-...

wow. that is better than my version.

Kind of like reading research papers:

1) Read the abstract 2) Read the conclusion 3) If any questions or doubts remain, read the middle.

Former patent examiner here. This is the truth. Only thing I'd add is, read the claims broadly. Anything that could possibly be meant by the words in the claim.

If it's broad enough to contemplate what you're doing, look through the spec, correspondence (the "file wrapper"), and general knowledge at the time for arguments to restrict interpretation of scope. Also, talk to a lawyer :(

edit: linked to this in another comment, and it also belongs here: "Broadest Reasonable Interpretation" http://www.uspto.gov/web/offices/pac/mpep/s2111.html

Would you say it's more or less impossible to write your own patent claim? I.e. is there some book or some resources that could teach you to do it or is the knowledge required too big?

Mostly impossible. There is a fairly useful book - the USPTO's "Manual of Patent Examining Procedure" (or MPEP). Official - http://www.uspto.gov/web/offices/pac/mpep/ My copy - http://adlervermillion.com/MPEP/

Its a huge book though, and not worth reading it unless you want to work with patents every day (as a patent agent, patent examiner, patent lawyer).

Also, its a reference book, not really something you can read start-to-finish.

Of course it's possible for anyone to write a patent claim. However, it's unlikely that the claim you write will be of any value.

I went to law school, passed the patent bar, and worked at the patent office. Even I wouldn't want to write my own claims. There are too many technicalities to deal with in the process. At the very least find a patent agent to do it (agents can draft and manage an application; attorneys can do that plus opine on validity etc.).

Only the claims define the power to block others from using the ideas in the patent.

In theory.

But remember that the patentee (that's the one with the patent) can always keep a "continuation" pending. With a continuation, the patentee can rewrite the claims however he wants. He can wait until your product is on the market and rewrite his claims to follow your product exactly. He could even sue you, lose for invalidity or noninfringement, and then rewrite his claims to more closely match your product and sue you again.

The only limit on claim rewriting is that the new claims have to somewhat reflect what's in the specification and have to be reexamined. Reexamination, like examination, is more a coin flip than an engineering evaluation of the patent claims. Reflecting the specification doesn't have to follow too closely what was in there originally but it does place some limit on the breadth of claim rewriting.

And rewritten claims have the same long ago priority date against prior art as the original claims. Your product does not count as prior art just because it was widely known when the new claims are written.

One common confusion people have when reading patent is the word "comprise," which appears a lot in the claim language. This probably comes from the confusing definition of the word in the legal dictionary, http://legal-dictionary.thefreedictionary.com/comprise, which implies there are more elements than the ones specified.

Comprise always means all or nothing in the claim conditions. All conditions have to be satisfied for the claim to be valid against a product.

"Comprising" means having at least all of the listed elements.

"Consisting of" means having all of the listed elements and nothing more.

"Consisting essentially of" means having all of the listed elements, with more stuff that doesn't materially change the character of the invention (that's where the lawyers come in).

http://www.uspto.gov/web/offices/pac/mpep/s2111.html

Having read this article, I begin to suspect that only the claims define patent rights, and not any other section of the patent.

Its the number one mistake people make... but maybe I should tone down the repetition a little?

I'm a patent litigator and I agree it's the number one mistake laymen make. I say keep it.

No. Keep it the way it is. The number of times patent stuff shows up on HN and people go read the abstract or description, or just the title, and then go get outraged is (hehe) outrageous. This needs to get hammered into people's heads.

sadly, this won't stop them, as I'd wager more than a few of them know that the abstract has no legal weight, but are always up for an excuse to bash their particular company/individual of choice who files software patents.

agreed. a close second is the confusion between a published application and an issued patent.

The repetition is a little amusing, but it's also highly effective. I like it! An article of this sort is successful if people walk away remembering one concrete thing, so I think you should keep it.

Assuming of course that it's a smart idea to be reading that patent at all...

>>> Check to see whether the prior art reference explains every element of a claim in the patent. If so, the reference “anticipates” the claim, and the claim is invalid.

I'm not a lawyer, but I have over a dozen patents, and have participated in other IP related activities such as invalidation. Here is how I understand it, as explained to me:

It's assumed by default that the patent is granted in light of the information referred to by cited works. In other words, the examiner has declared that nothing in the references anticipates the claims. If you're looking for new prior art, it has to be outside of those references.

Of course anything can be challenged, but instead of "look, we found some prior art," you have to argue, "look, the examiner missed something."

OP, does that make sense?

A couple other thoughts: First, the body text may be gibberish, but the body of patent A can be used against the claims of patent B, if A is not cited by B. I've used this successfully, thanks to a few solid days of Google searching. Or, A and C can be combined to show obviousness.

Second, body text can contain useful information in its own right. I have a product under development right now for my side business, using an electronic circuit described in an expired patent.

Interesting read, thank you for posting it.

One suggestion is to copy/move the warning "If you have an important patent question, hire a patent lawyer" to the intro.

A lot of people skim through the conclusion section and may just know enough now to be dangerous, like you said :)

Q: How To Read a Patent A: If you work in a company, then the answer is very simple: "Don't!" - Don't even try to pretend that you've heard it, or that it exist.

Only claims define the intellectual property rights, I get that. But patent jargon is often so abstruse and vague that you need to read much of the supporting text to even understand the claims. Sometimes when I look at claims for patents that I'm an inventor on even I don't know what IP is being staked out based on the claims section alone.

Agree. Patent language is way too confusing. Hopefully the Supreme Court's ruling in Nautilus v. Biosig will force lawyers to clarify their patent language. From Justice Ginsburg:

> we hold that a patent is invalid for indefiniteness if its claims, read in light of the specification delineating the patent, and the prosecution history, fail to inform, with reasonable certainty, those skilled in the art about the scope of the invention.

https://medium.com/patents-technology-law/good-news-in-the-f...

One trick I find very useful is to read the claims out loud. It forces me to slow down and concentrate on the words.

Also, some patent lawyers have a "said fetish": every component is referred to as "said component" so you get sentences like "A widget consisting of a foo, a bar and a finangle, with said foo afixed atop said bar and said finangle to make an assembly where said assembly has said foo..." etc. The presence of the "said"s makes the claims weirdly unreadable, as they grab scarce attention from the lay reader, so just blip over them when reading.

I wish someone would write "how to read an IETF specification" or "how to read an ECMA standard."

Here's a link to a patent on Google Patents for example: https://www.google.com/patents/US5255452

You can see how Google emphasizes the independent claims on the right.

I think we killed it :(

Here is a cache of the homepage though...

http://www.salar.ly.nyud.net:8080/

I've needed this for a while.

Reply on news.ycombinator.com