A1.user: Alexa, turn on X
A2.the command is sent via the relay to the controller
A3. based on the reply you get
OK
Out of range error
Controller offline
etc…
Other implementations do it like this
B1.user: Alexa, turn on X
B2. Alexa replies OK
B2.the command is sent via the relay to the controller, no real feedback to the user
Now, the question is mostly on user preference. While the time it takes to do A2/B3 is based on network speed, the perception of speed for the user is based on the time of the Alexa replies to the command.
When everything works well, the (B) seems better - instant feedback.
But when there are issues (let’s say network is down for the controller)
(A) implementation gives an error - controller down
(B) implementation says OK but does not actually do the command
On the other hand, if network latency is big, then
(A) implementation will run the command but say it timed out
(B) will say OK, as always
One of my biggest irritations is when I get feedback that doesn’t correspond with the actions taken, so scenario B just sounds like a source of frustration.
“Alexa, turn off the sofa lamps”
[blue/white LED spin]
“OK!”
[no lights go off]
If there’s a problem, the tin can on the table should tell me about it,
Hubbie & I prefer to know the status of the command, don’t say OK until after the command has successfully completed. We have a Harmony Hub and sometimes Alexa will say “OK” and nothing happens, that is frustrating. To me, the best solution is for Alexa to reply back OK to mean that she understood the command, and then later if the command is unsuccessful then give status for that also.
I voted for B.
I would prefer the Okay as an acknowledgement of what i asked.
as in “okay i’ll go do that now”
Ideally it would be Okay → action taken = action happens
or
Okay → action taken → didn’t work → “sorry I was not able to make that happen” or something to that effect.
Although I voted for option A, which made the most logic sense to me, I do like the suggestion Mvader makes, where Alexa confirms the understanding of the command and then ONLY further responds IF the action could not be completed. This would seem to satisfy both options A & B without an excess of feedback confirming the successful completion of a command such as turning on or off a light that the user has already seen.
It will be impossible to please everyone all of the time, but if Vera and Alexa can communicate at a level that alerts a user of the understanding of the command and then performs that command without an error, no further communication from Alexa would be necessary. This seems like the best solution I’ve heard so far.
[quote=“mvader, post:5, topic:197189”]I voted for B.
I would prefer the Okay as an acknowledgement of what i asked.
as in “okay i’ll go do that now”
Ideally it would be Okay → action taken = action happens
or
Okay → action taken → didn’t work → “sorry I was not able to make that happen” or something to that effect.[/quote]
I agree this is better, but this does not work with an Alexa home skill (we can only send back one of a number of fixed replies)… yet (?), it would need a new custom skill, that has it’s own set of issues:
“alexa, ask vera to turn on x” instead of “alexa, turn on x”
discovery of devices is only tied to home skill, so what alexa discovers is no longer relevant
worse device name recognition from alexa (since it has to try to understand a device name from an infinity of posibilities, instead of having to understand 1 name from a list of X fixed names from discovery)
I agree–if we want Alexa to act a certain way in general, it should be done at a higher level than at each individual skill, ie the API should support more than one status. Maybe since Vera has a developer interface with Amazon, it could be something to be brought to their attention to implement later.
[quote=“mcv.vlad, post:9, topic:197189”][quote=“mvader, post:5, topic:197189”]I voted for B.
I would prefer the Okay as an acknowledgement of what i asked.
as in “okay i’ll go do that now”
Ideally it would be Okay → action taken = action happens
or
Okay → action taken → didn’t work → “sorry I was not able to make that happen” or something to that effect.[/quote]
I agree this is better, but this does not work with an Alexa home skill (we can only send back one of a number of fixed replies)… yet (?), it would need a new custom skill, that has it’s own set of issues:
“alexa, ask vera to turn on x” instead of “alexa, turn on x”
discovery of devices is only tied to home skill, so what alexa discovers is no longer relevant
worse device name recognition from alexa (since it has to try to understand a device name from an infinity of posibilities, instead of having to understand 1 name from a list of X fixed names from discovery)[/quote]
so are you saying you are already at that fixed number of replies? (curious what that number is?) in my vision, you only need 2 replies. “okay” or “sorry, i couldn’t make that happen”
I would like to note that i would hate having to say “alexa, ask to do something” i really really prefer to just say “alexa turn on the living room lights”
so i’m very happy we don’t have to do that with the vera.
The Ooma skill works like that. I agree it’s easier to go ‘Alexa call X’ than it is to say ‘Alexa, tell Ooma to call X’. I’d hate to have to do that for Vera as I use that skill a lot more.
[quote=“mvader, post:11, topic:197189”]so are you saying you are already at that fixed number of replies? (curious what that number is?) in my vision, you only need 2 replies. “okay” or “sorry, i couldn’t make that happen”
I would like to note that i would hate having to say “alexa, ask to do something” i really really prefer to just say “alexa turn on the living room lights”
so i’m very happy we don’t have to do that with the vera.[/quote]
Off the top of my head, when an Alexa Home Skill requests is received, you can only reply once with (where applicable):
controller offline
light on
dimmer on at percent
+ setpoint
thermostat current sensor temperature
We can’t actually control the text that Alexa is speaking nor can we add 2 replies. Also, we have a time limit the request can ran. If we exceed that time limit (lag spike, laggy internet, etc), even through the command has been sent, Alexa will say there was an issue with the command.
We’ve forwarded some of your opinions and wishes to improve the API to Amazon so we’ll see what the future holds.
The more and specifically more accurate feedback from Alexa / Vera the better.
I vote for A as it tells us what happened but as other have stated sometimes the answer from Alexa doesn’t match what actually happens either. Sometimes it says, “OK” and nothing happens. Other times it says, “I cannot do that” (or whatever…) and it actually does. At other times it just takes too long.
I would like B if it followed up with a status too. Following rules of effective communication, one should always confirm what was asked before answering. Alexa should do the same don’t you think?
I voted (b) [as fast as possible] because I am using Alexa most of the time for its convenience, not to help me troubleshoot. I understand most of us a technical and want to know more info / status / data points all the time. But if I ask Alexa to turn off the light, I will know it doesn’t work because I see the light NOT turning off.
Of course, the edge case is, you want to ask Alexa to turn off the upstairs light while you are downstairs, but honestly I see my Vera’s device status out of sync at times anyways (aka the light is on but my Vera says off), so can we really trust the Vera status at the end?
In any case, I rather have Alexa acknowledges my request to turn off my light right away, so I can move on with my life (to my tech friends out there, think of it as an async call). Most of the time, everything will work anyways so this will give us the fastest performance under normal circumstances.
If something doesn’t work, I will need to go through an endless night of troubleshooting anyway, but I rather not sacrifice my overall faster response time just to plan for an occasional outage…
