Help me learn to read patents!
December 1, 2016 10:48 AM   Subscribe

I have a new-ish job where I increasingly need to glean information from patents. Despite many years of reading other technical literature, I find them to be very difficult to understand, and I would like to find suggestions/resources for getting better at finding information in them more quickly.

The patents in question are biomedical, and are well within my area of expertise. I don’t usually care about the claims (there’s a legal department for that), but rather I’m trying to understand/use data from within the patent. I have a really hard time looking at a figure and finding the relevant description in the text, figure legend, and methods section. When I read a scientific paper, I usually read the abstract, look at the figures to understand the major findings, and read the text selectively depending on how important/relevant to me the data seems. I would like to find an analogous workflow for patents, but (a) they are SO LONG and (b) they have tons of jargon and legal verbiage that I can’t tell if I care about.

If it matters, I am generally looking at the patents using google patents or pdfs from freepatentsonline, and I am absolutely open to other ways of viewing that might be more helpful.
posted by juliapangolin to Technology (7 answers total) 4 users marked this as a favorite
 
I subcontracted briefly as a technical consultant on patent infringement cases, making test cases to show that a product was infringing on a patent. In those cases, my supervisor taught me that the three most important things are
  1. Be very, very sure about the meanings of words used in claims. Use a dictionary for everything.
  2. Confirm that the independent claims all actually claim different methods.
  3. Ignore the dependent claims.

posted by infinitewindow at 11:10 AM on December 1, 2016


What sort of data are you talking about? You could try a word search for various key words. Also, try searching for the inventors on PubMed, etc. to see if they've got related published work that you can refer to.

I write patents primarily to support the claims and don't particularly care if they are otherwise useful, nor does the patent office. So if you can find another source for the same information, it might be easier to deal with.
posted by exogenous at 11:20 AM on December 1, 2016


Sadly, there isn't a particularly well agreed upon set of writing standards for patent drafters, besides "you must have the required sections." There are some rules of thumb that I have been taught and I strive to produce, but I've read lots of patents that don't follow them. For example, the Detailed Description should start with an overview of the whole workings of a thing, and that should ideally also be FIG. 1. The figures should then be discussed in order, highlighting subsections of the thing.

Patents tend not to have "major findings" in the sense of an academic paper, outside of the claims (which, yes, can be dense, but often you can figure out what is important in a patent from reviewing the claims. Here's a primer. Patent abstracts are generally terrible (they aren't supposed to contain claim language, but they always do, sometimes they are just claim 1 rewritten as sentences).

On thing that may help (and also confuse) is that it's fairly common to see multiple patents in a "family" where the most of the text and the drawings (generally called the "specification") are identical, but only the claims are different. (there are technical terms for these relationships, but I'll skip them). This and This are patents that are related in that way*. If you have a stack of patents or applications to read, try to sort out what the relationships are, and you can save yourself some reading.

One clue for picking out legalese is to look for paragraphs without numbers. The numbers are reference labels for the drawings, so the "boilerplate" tends not to have it, as it's not invention specific. In the examples I linked above, this last paragraph is an example: "Although particular preferred embodiments of the invention have been disclosed in detail for illustrative purposes, it will be recognized that variations or modifications of the disclosed apparatus, including the rearrangement of parts, lie within the scope of the present invention"

Finally, if you're trying to figure this all out on a screen, consider killing some trees while you get your bearings. U.S. patents are all formatted in the same two column style which makes it easier (for me, at least) to catch similar paragraphs. Also, when I print patents, I will staple the drawings in a separate bundle from the front page and the rest of the text. That makes it easier for me to flip back and forth. I do appreciate that Google Patents makes text searching/copy-paste easier, but the reflowing is pretty much the worst (especially if you have to cite anything ever).

If I'm really trying to understand something, I break out the highlighters. If Gear A of the widget seems important, I will highlight the references to Gear A in the text, and then highlight Gear A in the figures. Multiple colors get involved pretty much every time.

If you have a particular patent that you'd like to see broken down a little bit, feel free to memail me.

*IAAPA, IANYPA, TINLA, example patents have nothing to do with my practice.
posted by sparklemotion at 11:28 AM on December 1, 2016 [3 favorites]


IAAPA, IANYPA, TINLA. I write patents with the intent that they be correct and useful to several audiences (as well as satisfying the modest requirements imposed by the USPTO and patent examiners, as alluded to by exogenous). But based on the prior art that I review in other parts of my work, I'm in the minority there -- generally speaking, it doesn't matter very much, and in addition, few of the patent attorneys I've met actually have any engineering chops, so they're not well equipped to write the kind of disclosure that would be useful to practitioners in the field.

All of which is to say, patent applications are poor sources of general information about inventions. Perhaps you need a better target for what you want to get out of the applications? If it's "how many milligrams of X do you mix in?" or "what temperature should the reaction be done at?" then you can probably find that. But "why does this compound have this effect?" or "what are the possible implications of this pathway?" are unlikely to be discussed in any useful way.
posted by spacewrench at 12:01 PM on December 1, 2016 [2 favorites]


You might find this guide helpful as far as how to move through the sections of a patent.

Do you ever need to compare different patents or do a survey of patents that all come from one inventor or are on one topic?
posted by purple_bird at 12:03 PM on December 1, 2016 [1 favorite]


You could try a word search for various key words.

Be very cautious in doing this; patents often use intentionally obfuscatory language, meaning you may not find something by searching for what it's usually called.

There's a reason for this. The bargain in patents is that the patentee gets a legal monopoly on the invention for a fixed period of time; in exchange, they have to tell everyone exactly how they did it. Once the patent term is up, not only is anyone else allowed to make/do the thing, they know precisely how to do it, because it says so right there in the patent.

So, that's one of many differences between academic papers and patents: in an academic paper, the author wants the reader to know how they've done it. In a patent, the author might not want the reader to know how they did it, but they are legally required to tell them in order to get the monopoly. So they try to get the best of both worlds by using obscure, obfuscatory language making the patent difficult to find using the most obvious search terms, and making the relevant parts within the patent difficult to find if you're just doing Ctrl-F for the obvious terms, and possibly making it difficult to understand even if you can find it.

True story: I was once reading a patent, which described (as one step among many) a process involving DNA, some oligonucleotides, repeated application of an enzyme which could extend DNA, etc.

After about the third time I read that part, I realized that what was being described in that step was PCR. (For those not in the field, PCR — the polymerase chain reaction — is an astoundingly common technique in molecular biology. If you watch crime procedurals on TV, you've probably heard it mentioned in the context of forensic DNA identification, but the forensic application of PCR is just one among many many uses of the process.)

The words "PCR" or "polymerase chain reaction" never appeared in the patent. Had you been searching only for those, you never would have found it.
posted by DevilsAdvocate at 1:37 PM on December 1, 2016 [4 favorites]


Thanks, folks. I marked sparklemotion's answer as best because printing the figures and descriptions thereof has been the biggest single help in trying to manage/understand the information!
posted by juliapangolin at 11:08 AM on January 1, 2017


« Older ̶R̶o̶m̶n̶e̶y̶   |   Long term travel with young children Newer »
This thread is closed to new comments.