Siri will be open to 3rd party devs IMHO. User will say to Siri, "[App Name], [command]."
"New York Times app, read me the 10 top stories."
"Pandora app, play Sugarland station."
Each developer can list a series of actions = verb + object. When the user installs the app, Siri will pick up all the actions and they will be available to the user. Again, it must be initiated by the user saying the app name first. I can see two main usage cases:
1. Launch app into particular action - for example, "Pandora app, play Sugarland station." This opens Pandora and plays Sugarland station. Pretty straightforward. Devs just determine the verb + object and what Siri does with the action. In this case, Siri launches Pandora and sends over the verb "play" and the station "Sugarland," and Pandora knows what to do with it.
2. Read/show data within Siri. So, in this usage case the app doesn't launch but rather gives Siri data. For example, "New York Times app, read me the last 10 stories." Siri sends "read" (verb) and "last 10 stories" (object) to New York Times app. New York Times app gets that action and does an API call to the server to return the 10 top stories and gives it to Siri to read. Siri reads them to the user.
#2 usage case is pretty mind-blowing if Apple can pull it off. Because it's taking apps and pulling off its cover and stripping them to it's core services. Users don't have to launch the app to get certain data. App become more services that Siri can access. Of course, the user can still launch the app and do other things in the app via touch. But the app has a voice interaction interface and Siri is the master of that.
Again, this solution seems very practical on paper. It sounds perfect. But...
This involves Siri learning app names. Don't forget, under all the magic is voice recognition. Is Siri going to understand "Instagram"? How will it know the difference between "fourquare", 4x4 (four squared) and "four square"?
First of all, wether or not Apple decides to open an API to siri and wether or not it's possible are two separate things. None of his arguments are very convincing to me. Managing a 3rd party ecosystem is a hard problem but not that hard; especially for Apple who really can call the shots however they see fit. Here's a scenario: all 3rd parties can plug into the siri API but they need approval and can be kicked out for whatever arbitrary reason Apple picks. That would suck for developers (see marco's post earlier) and yet most developers would jump through hoops to get into there.
I don't see the point in registering with apple a particular phrase. Wasn't part of the reasoning behind new hardware being required was that much of the translation is being done ON the phone, not in "the cloud"? Or is this now wrong?
I'm going under the assumption that Siri translates words to text/commands on the phone device. It then determines whether that command is "local" or "remote." If local, it does whatever it needs (create an alarm, add a calendar event). If remote, send it to the Siri servers for them to process.
In the case of apps, they'd all be local. The app could remotely access data required on it's own, not needing the Siri servers for that. So, just like the new notification preferences, Siri gets a preference that gives you a list of all commands, and for those commands you can select which application responds to it. Obviously there may be better ways to handle the interface, Apps may have their own preferences and you can tell the app to register it's commands. Regardless, there are better minds thinking of this stuff than my own.
So, in this particular scenario, there's no need to "phrase squat" or anything. The whole idea of this article seems ridiculous given how I interpret how Siri functions. Which again, could be wrong.
The only major issue I can see is that Apple favored apps will always take first pick.
If you want to say "Tweet blah blah", and use something that isn't the official Twitter client, you're probably shit out of luck. You'd have to say the app's name or something of the sort.
Another example is that Reminder will take precedence over any other third party todo app.
Otherwise I think there are ways around the whole 3rd party issue with a strict approval process and having developers somehow interpret their own text commands and provide some standard feedback about what Siri's response should be.
9 comments
[ 1.7 ms ] story [ 31.0 ms ] thread"New York Times app, read me the 10 top stories."
"Pandora app, play Sugarland station."
Each developer can list a series of actions = verb + object. When the user installs the app, Siri will pick up all the actions and they will be available to the user. Again, it must be initiated by the user saying the app name first. I can see two main usage cases:
1. Launch app into particular action - for example, "Pandora app, play Sugarland station." This opens Pandora and plays Sugarland station. Pretty straightforward. Devs just determine the verb + object and what Siri does with the action. In this case, Siri launches Pandora and sends over the verb "play" and the station "Sugarland," and Pandora knows what to do with it.
2. Read/show data within Siri. So, in this usage case the app doesn't launch but rather gives Siri data. For example, "New York Times app, read me the last 10 stories." Siri sends "read" (verb) and "last 10 stories" (object) to New York Times app. New York Times app gets that action and does an API call to the server to return the 10 top stories and gives it to Siri to read. Siri reads them to the user.
#2 usage case is pretty mind-blowing if Apple can pull it off. Because it's taking apps and pulling off its cover and stripping them to it's core services. Users don't have to launch the app to get certain data. App become more services that Siri can access. Of course, the user can still launch the app and do other things in the app via touch. But the app has a voice interaction interface and Siri is the master of that.
This involves Siri learning app names. Don't forget, under all the magic is voice recognition. Is Siri going to understand "Instagram"? How will it know the difference between "fourquare", 4x4 (four squared) and "four square"?
I don't see the point in registering with apple a particular phrase. Wasn't part of the reasoning behind new hardware being required was that much of the translation is being done ON the phone, not in "the cloud"? Or is this now wrong?
I'm going under the assumption that Siri translates words to text/commands on the phone device. It then determines whether that command is "local" or "remote." If local, it does whatever it needs (create an alarm, add a calendar event). If remote, send it to the Siri servers for them to process.
In the case of apps, they'd all be local. The app could remotely access data required on it's own, not needing the Siri servers for that. So, just like the new notification preferences, Siri gets a preference that gives you a list of all commands, and for those commands you can select which application responds to it. Obviously there may be better ways to handle the interface, Apps may have their own preferences and you can tell the app to register it's commands. Regardless, there are better minds thinking of this stuff than my own.
So, in this particular scenario, there's no need to "phrase squat" or anything. The whole idea of this article seems ridiculous given how I interpret how Siri functions. Which again, could be wrong.
If you want to say "Tweet blah blah", and use something that isn't the official Twitter client, you're probably shit out of luck. You'd have to say the app's name or something of the sort.
Another example is that Reminder will take precedence over any other third party todo app.
Otherwise I think there are ways around the whole 3rd party issue with a strict approval process and having developers somehow interpret their own text commands and provide some standard feedback about what Siri's response should be.