Slashdot Mirror
A Look at the US Patent System
cheesedog writes "The LA Times published an interesting editorial on the current state of our patent system. From the article: 'on many levels, the U.S. patent system is profoundly flawed. Too many patents are issued for 'innovations' that are obvious, vague or already in wide use.' Online reaction has been mixed, with PatentHawk striking out in defense of the patent system, and Right to Create providing some support for the LA Times editorial."
As I wrote some time ago (I looked for it so I could just put the link in here but couldn't find it on SlashDot), one of the things that keeps getting left out of this debate, is the fact that because the PTO is being bombarded by thousands of patents each year it needs to computerize the way it works. The steps I see are:
1. Create a dictionary of all words used in applying for a patent. This is rather obvious because a good spell checker is needed anyway. But this goes beyond that. It allocates a unique id number to each of the document's words. This allows you to reduce the overall size of the document quite a bit. (After all, if you use a four byte word that gives you around four billion words and the largest dictionary only has a couple hundred thousand words in it.) Legal jargon usually uses more than four letters in a word and thus the document would be smaller overall.
2. A second dictionary of terms which are equal to each other. This dictionary would grow over time. Basically, things like "flashlight", "Light emitting device", "a device with a lamp in it which projects a beam", and "hand held light device" are all the same or similar. Thus, when a term which is unrecognized comes up in a document it can be added to one of the lists and from then on it is associated with that term. (And yeah, they should be able to add, remove, etc... from the list.)
3. The program should have already scanned all previous patents and created the above two dictionaries. Then when a new patent comes through (since they have to be submitted electronically now anyways) it is passed through the program which determines how closely a given document comes to other patents. Note that this is different from "are the sentences the same" or "are the sentences in the same order". The program should not care what order anything appears in - just do a search like Google and find how many words are the same or similar (remember they could replace all words of "flashlight" with "hand held light emitting device" via word processor).
3a. Since the patent system is divided up into various areas (ie: Games, Construction, etc...) the program should scan across all boundaries to ensure that something from one area is not now being patented in another area.
3b. All entries should be listed (just like with Google) in a descending order of revelance. So a patent which was given out in, say 1816 (The Stirling Engine) isn't re-patented as "The Audacious Engine" simply because all of the places where it says "Engine" in the orginal patent are replaced by "a non-internal combustion device".
4. All applied for patents should be kept on file so they too can be checked against. Notes on why the patent was denied should also be kept on file so they can be referred back to.
People may say we can't do this. Google has to handle over a billion web pages yet it can do it in a matter of a few seconds. There are only a couple million patents. The PTO should be able to handle this really easily. Hire the guys from Google to set things up. (And no - I don't work for Google.)
As for graphical pictures showing how something works - it depends. There are software packages which can compare one item/picture to another but all it would take is to accidentally send the picture reversed, rotated slightly so it looks different, use different colors, shades, shadows, etc.... You can look for similarities but that is about it.
In any event - it is nice that the powers that be are trying to fix the problem (or at least suggest changes) but it would be more realistic to try to automate the whole process so the patents can be throw out faster and faster. Which is why both good and bad patents are needed and both should have their own set of dictionaries. You need the bad patents in there as a way to say "Hey! Here are examples of why you can't have a patent!" Further, the bad patents could be used just like the good ones to show how someone tried once before to get something passed. And just like the good search informa
Check out my website: Playfully Clever
Patenting is really a boring issue unless your directly involved with its consequences but im happy the issue is starting to come up in mainstream media.
Like if you use any technology whatsoever?
Don't waste your vote! Vote for whoever you want, unless you live in a swing state it won't matter anyways