In a way, I think it's even worse than that. I've used Alexa since the Echo was a relatively new product. Back then, I experimented with new phrases and commands often, but was frequently greeted with wrong answers or "Sorry, I don't know how to help with that." Over time, I stopped trying those commands. Skip forward to today--the backend has been improving for years, and many of those commands now work, but it's too late. Their users have already been taught that they don't, so folks stop trying to use those features. Not only can you not discover new commands easily, you might mentally blacklist useful commands permanently.
Perhaps more frustratingly, the "What's New" emails they send out don't help with this. They never say "Hey, we know you tried to ask your Echo to report its volume level before and it didn't work, but it does now." They always say "Ask Alexa to tell you an Arbor Day joke!" -_-
TBH, I don't understand why they couldn't add a circuit that checks if the microphone is being activated by the device's own speaker, and ignore it. Some sort of simple phase inverting circuit from the speaker that gets summed with the microphone input ought to do it. It's not exactly rocket science and they're doing complex phased-array/beam-forming stuff with the microphones already, so it's not like the engineers are idiots?
It's a very impressive infrastructure, and works pretty well too.
"Hey, 3 months ago you asked for the forecast according to the ECMWF weather model, and I couldn't answer you. But now I can, so go ahead and try!".
> "Hey, 3 months ago you asked for the forecast according to the ECMWF weather model, and I couldn't answer you. But now I can, so go ahead and try!".
What might work more effectively is
"We've been improving our weather forecasting and you can now get the forecast in your area according to a range of new models, including the often requested ECMWF weather model! Give it a try now!"
Thinking back to this classic where Target figures out the trick to not creeping out someone they suspect may be pregnant is to shuffle the maternity coupons in with a bunch of other random stuff: https://www.forbes.com/sites/kashmirhill/2012/02/16/how-targ...
What Target discovered fairly quickly is that it creeped people out that the company knew about their pregnancies in advance.
“If we send someone a catalog and say, ‘Congratulations on your first child!’ and they’ve never told us they’re pregnant, that’s going to make some people uncomfortable,” [snip] So Target got sneakier about sending the coupons. The company can create personalized booklets; [snip]
“Then we started mixing in all these ads for things we knew pregnant women would never buy, so the baby ads looked random. We’d put an ad for a lawn mower next to diapers. We’d put a coupon for wineglasses next to infant clothes. That way, it looked like all the products were chosen by chance. And we found out that as long as a pregnant woman thinks she hasn’t been spied on, she’ll use the coupons. She just assumes that everyone else on her block got the same mailer for diapers and cribs. As long as we don’t spook her, it works.”
Sounds much closer to hunting, literally targeted marketing.
I mean, you gotta listen to every word that comes out of the customer's mouth right?
Their Principles are too reductive to even guarantee that copies of other people's ideas are successful. For as much of a genius as he may be, Jeff Bezos has created an organization of automatons.
[obligatory disclaimer: I worked there years ago, but no longer do.]
I'd argue this is the wrong thing to solve for. The best skills take a long time to make and require privledged access. Sonos, Spotify and others have this, and they work amazing. It's the large mass of Alexa skills they were quickly made and that don't have privledged access that are dragging the whole experience down.
And this is entirely Amazon's doing.
It's humbling (and annoying) when you think of yourself as a power user of some application to find out from some semi-new user of it that there's a new feature that would have made your life much easier if you had only known it was added a year or two ago.
There's only so much time to learn about the tools you use, much less the changes in them over time. How do I find out if GNU grep has some interesting new feature in the version that ships with the next version of the distro I use? Or rsync? Or tmux/screen?
This problem definitely seems worse with smaller tools, since there's a lot of them and they are often packages together by some other party. Not every project is as large and used as much but such a large population as Firefox, which arguably does a very good job of advertising new features. But you can definitely tell it takes them a lot of time and effort to do so, and a smaller project may not find much traction if they tried the same strategy.
This makes me shudder. Such a functionality would mean that they stored your failed request in some sort of database, a database that they later used to send you a personalized marketing email. A machine cataloging our voice for later inspection is a very dark future. Alexa should delete and scrub any iota of voice that it doesn't instantly understand.
"Oh, remember last week when I though you were asking me to buy you a pot plant? I now realize you were asking me to buy you some pot from Canada. It will be arriving in two days."
"I didn't understand it at the time, but I now realize that you were yelling at your husband Alex, not Alexa. Your social credit score has been adjusted to reflect this negative interaction."
I don't think this kind of thing would need (or even benefit from) actual voice recordings when it comes to the "new functionality" part of things. I would expect something more like a running tally of the X most common failed interactions after running the transcript through a fuzzy-match filter, and a long list of accounts associated with each.
This alone would make for a much more usable experience, and yet no voice assistant has implemented it yet, which flabbergasts me. I don’t have much knowledge in this particular realm so maybe I’m missing some big blocker that prevents it from being possible, but to me it seems like such an obvious thing to do.
This is a really important point, and both a strength and a weakness.
The iPhone began as essentially a front end to existing services (with visual discovery, which a voice interface inherently lacks). I used to name mine "FEP" (as in Front End Processor) -- a front end to a subset of "real" computing or as I think of it these days: multiple windows into a shared computing space.
A watch (like the apple one) is really a crappy general UI device; discoverability is pretty bad because of the limited area and speed. But it's great in the role the phone had: subsetted interface to a limited number of "real" computing tasks (yes, it has a few of its own tricks too but mainly as data collection for apps on your phone).
Thompson captured this issue by talking about devices and software in terms of "the task it was hired to to". The problem is the voice assistants haven't figured that out yet.
The thing about this approach is it creates pressure to make more functionality available at the edge.
Alexa and google home tried to jump right to the edge in one go, which skips too much phylogeny.
Apple seems to understand this, but gets it wrong in the opposite direction: the iPad hasn't moved far beyond being "most of an iPhone but with a larger screen". And if you have an apple speaker, a phone, and iPad and call out "hey Siri, set an alarm for 10 minutes" you may get three devices chiming in 10 minutes. They don't act like a single device.
Edit: there's also a third-party skill that will let you play old-school adventure/interactive fiction games. https://www.amazon.com/Vitaly-Lishchenko-Interactive-Fiction...