Please just read the claims
Hi, everyone. Let’s talk about how to read patents again.
There’s a raft of headlines today claiming that Facebook has a patent on secretly turning your phone mic on when it hears a signal from a TV. The story appears to have picked up in Metro UK, which ran the headline “Facebook wants to hide secret inaudible messages in TV ads that can force your phone to record audio.”
Here’s Gizmodo, in classic Giz style: “Facebook patent imagines triggering your phone’s mic when a hidden signal plays on TV.”
Here’s Engadget, which personally breaks my heart since I used to do all the patent stories there: “Facebook patent turns phone mics on to record reactions to ads.”
Here’s Esquire, for some reason: “Facebook’s new patent can turn on your microphone and secretly record you.”
Anyway, all of these headlines are wrong. Facebook did not apply for a patent on turning your phone microphone on when a hidden signal plays on a TV. I know this because I simply read the patent claims, which do not have the word “phone” or “microphone” in them at all. And the claims are the only part that truly matter.
A patent is composed of several parts. There’s the title, which is usually gibberish, the abstract, which broadly describes what’s in the patent, the specification, which explains how to build or use the invention being patented, and then there are the claims, which is the actual subject matter of the patent. If you are not a patent lawyer in patent litigation and you are reading anything but the claims, you are getting it wrong.
This is so fundamental to reading patents that law students are taught “the name of the game is the claim” in order to remember it.
THE PATENT CLAIMS DO NOT CONTAIN THE WORDS “PHONE” OR “MICROPHONE”
So what are the claims of Facebook’s patent application? You can open it here - it’s number 376,515 - and follow along. Let’s look at the first claim, which is typically the broadest and most important - the following claims tend to narrow the first claim in different ways.
A computer implemented method comprising: receiving, by an online system from a client device, a data set comprising an identifier of an individual associated with the client device, an ambient audio fingerprint representing ambient audio captured by the client device during a broadcast by a household device in a vicinity of the client device, and time information indicating a length of time of the ambient audio captured by the client device; identifying a user profile associated with the online system based on the identifier of the individual received from the client device; identifying a content item associated with the broadcast based on the ambient audio fingerprint; determining that there was an impression of the identified content item by the identified user associated with the client device in response to the length of time of the ambient audio exceeding a detection threshold; logging the impression of the identified content item in association with the user profile in a data store of the online system; determining if impression data in the data store exceeds a threshold impression value; and responsive to determining that the impression data in the data store exceeds the threshold impression value, sending an instruction to a content provider to increase broadcasting frequency of the content item on the household device.
This is a lot of words, but they’re not hard to figure out: it’s a system that receives a user ID and an audio fingerprint, matches that audio fingerprint to some content, sees if that content was played for a certain duration, and then checks a counter to see if that content has been played a specific number of times.
IT IS BASICALLY THE BACKEND OF SHAZAM FOR ADS
That’s it. It is basically the backend of Shazam for programmatic ads. Again, the words “phone” and “microphone” do not appear in any of the patent claims. There are some parts of the specification that describe triggering microphones with a silent-to-humans sound, but again, the specification does not matter. Just the claims.
Now, could you read these claims and the specification and infer Facebook needs such a backend so that it would be able to silently enable your phone’s microphone and keep track of the ads on your TV? Sure, if you wanted. But you know what other kinds of devices listen for triggers to enable microphones and send audio data to the cloud? Smart speakers. Like the one Facebook has been rumored to be working on for some time.
The claim does mention that one device will capture audio from another device, citing “ambient audio captured by the client device during a broadcast by a household device in a vicinity of the client device,” but again, there’s no claim about how that “client device” would get automatically turned on without your knowledge. And there’s already an example of a smart speaker listening to silent commands from a TV: Alexa devices can hear a special signal Amazon embeds into Alexa commercials so they don’t accidentally trigger commands.
Once you put an always-listening microphone in people’s houses, the next logical step is thinking about what other kinds of triggers could set them off, and what they might do. And if you’re a big tech company, you file as many patents as you can to block competitors.
Facebook gave Engadget a statement to exactly that effect:
However, Facebook says it has no intention of ever implementing the technology described in the patent. In a statement, Facebook VP and Deputy General Counsel Allen Lo told us that the patent had been filed “to prevent aggression from other companies,” and noted that “patents tend to focus on future-looking technology that is often speculative in nature and could be commercialized by other companies.”
Lo went on to say that the technology in this patent has not been included in any of Facebook’s products, “and never will be.”
So Facebook filing a patent application on a backend ad-recognition system doesn’t mean really much of anything except Facebook’s lawyers are busily patenting every piece of smart speaker tech they can think of, just as Amazon and Apple and Google are doing. (And this is a patent application, which means it is written as broadly as possible to survive patent examination.) It does not mean Facebook is listening to you. And it certainly doesn’t mean Facebook has a patent on secretly enabling your phone’s microphones.
All you have to do is read the claims.
There’s a raft of headlines today claiming that Facebook has a patent on secretly turning your phone mic on when it hears a signal from a TV. The story appears to have picked up in Metro UK, which ran the headline “Facebook wants to hide secret inaudible messages in TV ads that can force your phone to record audio.”
Here’s Gizmodo, in classic Giz style: “Facebook patent imagines triggering your phone’s mic when a hidden signal plays on TV.”
Here’s Engadget, which personally breaks my heart since I used to do all the patent stories there: “Facebook patent turns phone mics on to record reactions to ads.”
Here’s Esquire, for some reason: “Facebook’s new patent can turn on your microphone and secretly record you.”
Anyway, all of these headlines are wrong. Facebook did not apply for a patent on turning your phone microphone on when a hidden signal plays on a TV. I know this because I simply read the patent claims, which do not have the word “phone” or “microphone” in them at all. And the claims are the only part that truly matter.
A patent is composed of several parts. There’s the title, which is usually gibberish, the abstract, which broadly describes what’s in the patent, the specification, which explains how to build or use the invention being patented, and then there are the claims, which is the actual subject matter of the patent. If you are not a patent lawyer in patent litigation and you are reading anything but the claims, you are getting it wrong.
This is so fundamental to reading patents that law students are taught “the name of the game is the claim” in order to remember it.
THE PATENT CLAIMS DO NOT CONTAIN THE WORDS “PHONE” OR “MICROPHONE”
So what are the claims of Facebook’s patent application? You can open it here - it’s number 376,515 - and follow along. Let’s look at the first claim, which is typically the broadest and most important - the following claims tend to narrow the first claim in different ways.
A computer implemented method comprising: receiving, by an online system from a client device, a data set comprising an identifier of an individual associated with the client device, an ambient audio fingerprint representing ambient audio captured by the client device during a broadcast by a household device in a vicinity of the client device, and time information indicating a length of time of the ambient audio captured by the client device; identifying a user profile associated with the online system based on the identifier of the individual received from the client device; identifying a content item associated with the broadcast based on the ambient audio fingerprint; determining that there was an impression of the identified content item by the identified user associated with the client device in response to the length of time of the ambient audio exceeding a detection threshold; logging the impression of the identified content item in association with the user profile in a data store of the online system; determining if impression data in the data store exceeds a threshold impression value; and responsive to determining that the impression data in the data store exceeds the threshold impression value, sending an instruction to a content provider to increase broadcasting frequency of the content item on the household device.
This is a lot of words, but they’re not hard to figure out: it’s a system that receives a user ID and an audio fingerprint, matches that audio fingerprint to some content, sees if that content was played for a certain duration, and then checks a counter to see if that content has been played a specific number of times.
IT IS BASICALLY THE BACKEND OF SHAZAM FOR ADS
That’s it. It is basically the backend of Shazam for programmatic ads. Again, the words “phone” and “microphone” do not appear in any of the patent claims. There are some parts of the specification that describe triggering microphones with a silent-to-humans sound, but again, the specification does not matter. Just the claims.
Now, could you read these claims and the specification and infer Facebook needs such a backend so that it would be able to silently enable your phone’s microphone and keep track of the ads on your TV? Sure, if you wanted. But you know what other kinds of devices listen for triggers to enable microphones and send audio data to the cloud? Smart speakers. Like the one Facebook has been rumored to be working on for some time.
The claim does mention that one device will capture audio from another device, citing “ambient audio captured by the client device during a broadcast by a household device in a vicinity of the client device,” but again, there’s no claim about how that “client device” would get automatically turned on without your knowledge. And there’s already an example of a smart speaker listening to silent commands from a TV: Alexa devices can hear a special signal Amazon embeds into Alexa commercials so they don’t accidentally trigger commands.
Once you put an always-listening microphone in people’s houses, the next logical step is thinking about what other kinds of triggers could set them off, and what they might do. And if you’re a big tech company, you file as many patents as you can to block competitors.
Facebook gave Engadget a statement to exactly that effect:
However, Facebook says it has no intention of ever implementing the technology described in the patent. In a statement, Facebook VP and Deputy General Counsel Allen Lo told us that the patent had been filed “to prevent aggression from other companies,” and noted that “patents tend to focus on future-looking technology that is often speculative in nature and could be commercialized by other companies.”
Lo went on to say that the technology in this patent has not been included in any of Facebook’s products, “and never will be.”
So Facebook filing a patent application on a backend ad-recognition system doesn’t mean really much of anything except Facebook’s lawyers are busily patenting every piece of smart speaker tech they can think of, just as Amazon and Apple and Google are doing. (And this is a patent application, which means it is written as broadly as possible to survive patent examination.) It does not mean Facebook is listening to you. And it certainly doesn’t mean Facebook has a patent on secretly enabling your phone’s microphones.
All you have to do is read the claims.