Joel Spolsky: Circle of Honor for Ask Patents
Nowadays, many people complain about lousy software patents. And I say: "Stop complaining and start to destroy them." It took me about 15 minutes to prevent one Microsoft cheating patent confirmation. Have 15 minutes? You can do it too.
In a minute I will tell you this story. But first - the background of this case.
Software developers, in fact, do not invent much. The number of actually new, not obvious inventions in the software industry, which in some areas deserve a state-provided monopoly, is probably equal to two.
The remaining 40 thousand of strange software patents issued each year are mostly garbage that any working programmer can “invent” three times before breakfast. Most issued software patents are not “inventions,” in the sense in which most people understand this word. These are simple things that any freshman Java student must do as a given two-hour homework.
')
However, many companies, large and small, realized that patents cost money and they try to register as many patents as they can. They believe they can get a large pile of patents, as an inexpensive byproduct of the work (which still needs to be done) by the Research and Development (R & D) department, just sending a few lawyers to run around the rooms and ask programmers what they are working on, and then try everything it's patented. Almost everything that they find is either obvious, or already done before, and thus not subject to patenting. But they use low tricks to get it through the patent office.
The first technique they use is to make the language of the patent statement confusing and confusing, as far as possible. This really makes it difficult for experts to identify a prototype (prior art) or to evaluate if the invention is obvious.
A pleasant side effect of making an obscure patent is that it rather works as a trap for patent infringers. Numerous patent owners, especially “patent trolls,” do not really want their patents to be infringed. Very often they want you to violate their patent and build a big business based on it, and only then discover it, and thus find themselves in the worst, from a legal point of view, position and could be successfully blackmailed by them. The harder it is to read a patent, the more likely it will be violated.
The second technique for obtaining bad software patents is the use of a thesaurus . Often, patent applicants invent new terms to describe well-known things. Most experts search for prototypes using search tools / systems. They have to. No expert can be aware of more than (rounding to the nearest integer) 0% of prototypes that can reject an application.
Since the experts rely heavily on keyword search, if you apply for changing some of the keywords in your patent so that they differ from all others already used, you can pass your patent even if it is explicitly based on the existing one. [someone else's] prototype, because using unnatural invented terms you make the search for prototypes much more difficult.
Now about the third technique. Have you ever seen a patent application covering a wide to ridiculous area? ("God, they are trying to patent a moving agent!"). And that's why. The application is deliberately “stretched”, that is, there is a desire to get the most “wide” patent, knowing that in the worst case the expert will narrow their claims to those that they will receive in any case.
Let me illustrate this as simple as I can by example. The patent is based on a list of claims: things that you claim to be invented by you, and which you will get a monopoly in, if the patent is confirmed.
For example, submit a simple application with these three claims (aka the patent formula otherwise):
1. Vehicle.
2. Vehicle of claim 1, where the engine is connected to the wheels.
3. The vehicle of claim 2, where the engine runs on water.
Please note that clause 2 refers to clause 1 and narrows it down. In other words, he claims a limited set of things from clause 1.
Now, suppose you invented a car that runs on water. When you apply for a patent, you can do it in the above manner, even knowing that there is already a prototype (prior art) for the “vehicle”, and you really can't claim it as your invention. The theory is that
(a) hey, you might get lucky!
(b) even if you are unlucky and the first point is rejected, the rest, more specific, will remain.
What you observe is a risky game (however, in case of a win, giving a big score), and you have to look at more specific claims in order to understand what they really expect to receive. And you never know if a patent examiner falls asleep at the wheel, and - BOOM, you get the right to extort money from everyone who makes, sells or uses vehicles.
So, anyway, a large number of lousy software patents are issued annually, and the more they are issued, the worse it is for software developers.
The patent office was a little annoyed by this. The America Invents Act changed the law by allowing the public to state prototype examples (prior art) during the consideration of a patent application. That's why the US Patent and Trademark Office asked us to create Ask Patents , a section of the Stack Exchange website where software developers like you can give examples of prototypes to prevent the issuance of lousy patents.
Seems difficult, right?
Honestly, at first I thought it would be difficult. Will we even be able to find vulnerable applications? The funny thing is that when I chose at random from a pack of patent applications for software, I came to the conclusion that they were all lousy, and this made our work much easier.
Take the patent application US 20130063492 A1 filed by Microsoft. One of the users of Ask Panents made a request for a prototype on this patent on March 26.
I tried to find a prototype for this patent application, just to find out how difficult it is to do. I read the application first. Well, to be honest, I just ran through her eyes. In fact, I missed the essay and description, and went straight to the claims. Dan Shapiro wrote an excellent blog post , How to Read a Patent in 60 Seconds , which taught me how to do this.
The text was typically confused, and terms like “pixel density” were used to call all the other programmers in the world “resolution”, either by accident (because Microsoft’s lawyers were not programmers), or, more likely, because confusion complicates Search.
Without delving into the text, I realized that this patent essentially tries to say: “Suppose you have a picture that you want to bring to another resolution. And suppose you want to have different versions of a picture for different resolutions, then you can choose one of the versions of the picture that is closest to the resolution you need and scale it. ”
It was not new to me. I was sure that the Win32 API already has this functionality. I remembered that it was commonplace to provide different icons for different resolutions, and, in fact, I was convinced that the operating system could choose an icon based on the screen resolution. So, I spent about a minute on googling and, ultimately (bingo!), I found this interesting document “Developing a DPI-independent Win32 application [PDF]” , written by Ryan Haveson and Ken Sykes from, by coincidence, Microsoft.
And it was written in 2008, while Microsoft’s new patent application tried to state that this “invention” was “invented” in 2011. Boom. Prototype found and specified.
The total time spent may have been 10 minutes. One of the Ask Patents participants indicated that the patent application refers to something called a “scaling series”. I was not sure what this meant to mean, but I found a characteristic section of an earlier Microsoft document that demonstrated this “invention” without using these words. So, I edited my answer a bit to indicate this. Here is my complete answer to Ask Patents .
Incomprehensible, but whoever this person placed the prototype request for, he pressed the “Accepted” button on the Stack Exchange. We thought it might have been a patent examiner, but he acted with a regular account.
At that moment I immediately forgot about it, until May 21 (two months later), when I received this letter from Micah Siege (Micah works as our full-time patent expert):
Micah showed me a document from the US Patent Office, confirming that they had rejected a patent application, largely based on the document that I found. This was the first "confirmed kill" from the Ask Patents, and it was surprisingly easy. I did not do the hard work of examining the patent application and thoroughly confirming that there is a prototype - the expert did it for me.
This, however, has an interesting lesson. Software patent applications are uniformly of poor quality. It is surprisingly easy for them to find a prototype. Ask Patents can be used to reject them with very little effort. And this is partly the individual extermination of a single patent application, which can make a crack in the mountains of bad patents at a time.
I dream that when big companies hear how damn easy it is to block a patent application, they will use Ask Patents against their competitors. How great it will be if Apple, Samsung, Oracle and Google get on the Ask Patents in the "Mexican standoff" . If each of these companies allocates three or four engineers for a few hours a day to disassemble with applications for patents from their competitors, the number of patents granted to these companies will be halved. Would it be something!
Have 15 minutes? Open Ask Patents right now and see if one of those requests is related to a topic that you know a little about and write about examples you can find. They are hidden in a flat list; Most of the prototypes can be found on Google. Have a good hunting!
In a minute I will tell you this story. But first - the background of this case.
Software developers, in fact, do not invent much. The number of actually new, not obvious inventions in the software industry, which in some areas deserve a state-provided monopoly, is probably equal to two.
The remaining 40 thousand of strange software patents issued each year are mostly garbage that any working programmer can “invent” three times before breakfast. Most issued software patents are not “inventions,” in the sense in which most people understand this word. These are simple things that any freshman Java student must do as a given two-hour homework.
')
However, many companies, large and small, realized that patents cost money and they try to register as many patents as they can. They believe they can get a large pile of patents, as an inexpensive byproduct of the work (which still needs to be done) by the Research and Development (R & D) department, just sending a few lawyers to run around the rooms and ask programmers what they are working on, and then try everything it's patented. Almost everything that they find is either obvious, or already done before, and thus not subject to patenting. But they use low tricks to get it through the patent office.
The first technique they use is to make the language of the patent statement confusing and confusing, as far as possible. This really makes it difficult for experts to identify a prototype (prior art) or to evaluate if the invention is obvious.
A pleasant side effect of making an obscure patent is that it rather works as a trap for patent infringers. Numerous patent owners, especially “patent trolls,” do not really want their patents to be infringed. Very often they want you to violate their patent and build a big business based on it, and only then discover it, and thus find themselves in the worst, from a legal point of view, position and could be successfully blackmailed by them. The harder it is to read a patent, the more likely it will be violated.
The second technique for obtaining bad software patents is the use of a thesaurus . Often, patent applicants invent new terms to describe well-known things. Most experts search for prototypes using search tools / systems. They have to. No expert can be aware of more than (rounding to the nearest integer) 0% of prototypes that can reject an application.
Since the experts rely heavily on keyword search, if you apply for changing some of the keywords in your patent so that they differ from all others already used, you can pass your patent even if it is explicitly based on the existing one. [someone else's] prototype, because using unnatural invented terms you make the search for prototypes much more difficult.
Now about the third technique. Have you ever seen a patent application covering a wide to ridiculous area? ("God, they are trying to patent a moving agent!"). And that's why. The application is deliberately “stretched”, that is, there is a desire to get the most “wide” patent, knowing that in the worst case the expert will narrow their claims to those that they will receive in any case.
Let me illustrate this as simple as I can by example. The patent is based on a list of claims: things that you claim to be invented by you, and which you will get a monopoly in, if the patent is confirmed.
For example, submit a simple application with these three claims (aka the patent formula otherwise):
1. Vehicle.
2. Vehicle of claim 1, where the engine is connected to the wheels.
3. The vehicle of claim 2, where the engine runs on water.
Please note that clause 2 refers to clause 1 and narrows it down. In other words, he claims a limited set of things from clause 1.
Now, suppose you invented a car that runs on water. When you apply for a patent, you can do it in the above manner, even knowing that there is already a prototype (prior art) for the “vehicle”, and you really can't claim it as your invention. The theory is that
(a) hey, you might get lucky!
(b) even if you are unlucky and the first point is rejected, the rest, more specific, will remain.
What you observe is a risky game (however, in case of a win, giving a big score), and you have to look at more specific claims in order to understand what they really expect to receive. And you never know if a patent examiner falls asleep at the wheel, and - BOOM, you get the right to extort money from everyone who makes, sells or uses vehicles.
So, anyway, a large number of lousy software patents are issued annually, and the more they are issued, the worse it is for software developers.
The patent office was a little annoyed by this. The America Invents Act changed the law by allowing the public to state prototype examples (prior art) during the consideration of a patent application. That's why the US Patent and Trademark Office asked us to create Ask Patents , a section of the Stack Exchange website where software developers like you can give examples of prototypes to prevent the issuance of lousy patents.
Seems difficult, right?
Honestly, at first I thought it would be difficult. Will we even be able to find vulnerable applications? The funny thing is that when I chose at random from a pack of patent applications for software, I came to the conclusion that they were all lousy, and this made our work much easier.
Take the patent application US 20130063492 A1 filed by Microsoft. One of the users of Ask Panents made a request for a prototype on this patent on March 26.
I tried to find a prototype for this patent application, just to find out how difficult it is to do. I read the application first. Well, to be honest, I just ran through her eyes. In fact, I missed the essay and description, and went straight to the claims. Dan Shapiro wrote an excellent blog post , How to Read a Patent in 60 Seconds , which taught me how to do this.
The text was typically confused, and terms like “pixel density” were used to call all the other programmers in the world “resolution”, either by accident (because Microsoft’s lawyers were not programmers), or, more likely, because confusion complicates Search.
Without delving into the text, I realized that this patent essentially tries to say: “Suppose you have a picture that you want to bring to another resolution. And suppose you want to have different versions of a picture for different resolutions, then you can choose one of the versions of the picture that is closest to the resolution you need and scale it. ”
It was not new to me. I was sure that the Win32 API already has this functionality. I remembered that it was commonplace to provide different icons for different resolutions, and, in fact, I was convinced that the operating system could choose an icon based on the screen resolution. So, I spent about a minute on googling and, ultimately (bingo!), I found this interesting document “Developing a DPI-independent Win32 application [PDF]” , written by Ryan Haveson and Ken Sykes from, by coincidence, Microsoft.
And it was written in 2008, while Microsoft’s new patent application tried to state that this “invention” was “invented” in 2011. Boom. Prototype found and specified.
The total time spent may have been 10 minutes. One of the Ask Patents participants indicated that the patent application refers to something called a “scaling series”. I was not sure what this meant to mean, but I found a characteristic section of an earlier Microsoft document that demonstrated this “invention” without using these words. So, I edited my answer a bit to indicate this. Here is my complete answer to Ask Patents .
Incomprehensible, but whoever this person placed the prototype request for, he pressed the “Accepted” button on the Stack Exchange. We thought it might have been a patent examiner, but he acted with a regular account.
At that moment I immediately forgot about it, until May 21 (two months later), when I received this letter from Micah Siege (Micah works as our full-time patent expert):
The United States Patent and Trademark Office rejected Microsoft's request for the "Resizing Images" patent!
The expert referred specifically to the Prototype (Prior Art) cited in the answer by Joel (“Haveson et al.”).
Here is the document that rejected the patent. Starting on page 4 and beyond, basically quoting cancels the application, obviously, taking into account Haveson.
Micah showed me a document from the US Patent Office, confirming that they had rejected a patent application, largely based on the document that I found. This was the first "confirmed kill" from the Ask Patents, and it was surprisingly easy. I did not do the hard work of examining the patent application and thoroughly confirming that there is a prototype - the expert did it for me.
(If you want to see the refusal, go to the Public Pair and look for the publication (like the publication number) number US 20130063492 A1. Click on the “Image File Wrapper” tab and look at the non-final disclaimer from 04/11/2013. Microsoft, needless to say, appeal this solution, and maybe this lousy patent will float to the surface again.)
This, however, has an interesting lesson. Software patent applications are uniformly of poor quality. It is surprisingly easy for them to find a prototype. Ask Patents can be used to reject them with very little effort. And this is partly the individual extermination of a single patent application, which can make a crack in the mountains of bad patents at a time.
I dream that when big companies hear how damn easy it is to block a patent application, they will use Ask Patents against their competitors. How great it will be if Apple, Samsung, Oracle and Google get on the Ask Patents in the "Mexican standoff" . If each of these companies allocates three or four engineers for a few hours a day to disassemble with applications for patents from their competitors, the number of patents granted to these companies will be halved. Would it be something!
Have 15 minutes? Open Ask Patents right now and see if one of those requests is related to a topic that you know a little about and write about examples you can find. They are hidden in a flat list; Most of the prototypes can be found on Google. Have a good hunting!
Source: https://habr.com/ru/post/188786/
All Articles