>>"It was slow, when it worked at all," Williamson said. "The software was riddled with serious bugs. Those problems lie entirely with the original Siri team, certainly not me."
As Steve Jobs said, once you become a VP you can’t make excuses anymore. You have crossed the Rubicon and need to start taking responsibility for the failures of your team.
My impression of Siri is that it was effectively R&D early on by the team that made it. To propose that Apple would purchase a company, and then expect an R&D product to support their level of scale without serious infrastructure changes, is a bit absurd.
So the takeaway here is that it was dumb because was poorly built?
Also, how different is that "Information" report behind the paywall? It would not surprise me to see Mashable rewriting an article but to do so with paid content seems inappropriate.
To say what cheald said in a more concrete way: "now we know why Siri was so dumb for so long" is definitely the title that I am currently seeing. The entire concept that there is a single title that somehow should match across sites is a weird concept made up by Hacker News that doesn't exist in the real world and probably causes more harm than it does good.
You can probably point to a lot of quality failings at Apple as starting with Steve Job's death. Apparently he failed to teach and instill a culture of high quality and instead just focused on micro-managing everything himself. Steve Jobs may go down as both the best and worst CEO Apple ever had.
Not true at all. Look up the story of Apple University. Jobs can't run Apple from his grave. He tried his best to leave Apple with the right culture and leadership in his view. And for the most part, Apple has done really well since he passed. No one is going to get 100% of things right 100% of the time, and Jobs is no different; things would have inevitably gone wrong under his leadership, too.
Exactly this—a particular anecdote from the biography describes his motivation. Apparently when the HP Touchpad failed, Steve Jobs viewed it as a tragedy more than a triumph:
> "Hewlett and Packard built a great company, and they thought they had left it in good hands, but now it's being dismembered and destroyed. It's tragic. I hope I've left a stronger legacy so that will never happen at Apple."
In creating Apple University he was really trying to instill a legacy that would outlast him. That said, when pulling out the "Steve Jobs never would have let this happen" card, it's also important to remember that MobileMe happened on Steve's watch, as did the iPod HiFi.
I still have a working HiFi in my home gym. It’s a solid piece of engineering, especially for its time. I don’t know why people love to hate it so much.
I feel like sentiment toward the iPod HiFi is less "hate" and more "why does this exist". It was a perfectly solid piece of hardware, but Apple tried to enter a market that was already saturated (iPod docks) with a far more expensive product that didn't offer any significant advantages.
Jobs screwed up plenty of times. Look at the whole history of NeXT Computer for example. Software so far ahead of everything else at the time, but totally screwed up getting it into the market.
How was NeXT a failure when it was acquired by Apple and is the basis of macOS, iOS, watchOS, and tvOS? It's currently running on over one billion devices.
I dunno, Apple’s has some pretty terrible CEOs. Jobs might have been a dreadful micromanager and possibly a bit of an arse, but he did disrupt at least four major industries and lay the foundation for the most valuable public company ever.
Whereas Sculley, Spindler and Amelio were a parade of failure from start to finish.
I doubt very much he would have lost his job today for any of his behavior.
Look at what the U.S. President has and continues to do, and he still has his job, and a much more important and public one too.
Politics aside, Jobs' turnaround of Apple into the most valuable company in the world (for awhile, recently) would have ensured he remained at the helm for as long as he wanted.
What behavior are you talking about that would cost him that position today? Jobs wasn't physically assaulting employees. It isn't illegal to be mean or rude to people. At worst it might cost Apple a few workplace lawsuits/settlements.
One thing I feel nearly certain is that had Steve Jobs been alive, Apple would have never under any circumstances released a flagship phone and a flagship laptop simultaneously that, out of the box, had no way of connecting to each other. I still can't believe that happened.
I don't see why. The Power Mac G5 still came with ADC for a year after they switched all their displays to DVI. (And the ADC-DVI adapter was $149.) Admittedly they weren't released simultaneously.
I disagree completely. By the time they shipped with incompatible cables / ports, wireless backup with iCloud, wireless sync with iTunes (who even "syncs" at all anymore...), and even wireless deployment and debugging had all been implemented.
Steve wouldn't have wanted you to connect your phone to your computer in the first place. That's why iTunes got wireless sync back before everything was in the cloud. The only really remaining piece was developers needing to plug in for deployment and debugging, but who cares - they're developers, they'll manage.
Yet, so many people plug their phone into their computer for the simple reason of charging it. Now perhaps wireless charging is coming, but the Apple cable fiasco sure inconvenienced anyone with that habit.
Comparatively I think we could safely say that Microsoft transition from a retiring Bill Gates to Steve Ballmer didn’t go as smoothly as what happened at Apple.
The cofounder/CEO of Siri contests the depiction of the scalibility issues in the original Information article:
> This statement, wholly false, was made by the architect and head of the biggest launch disaster in Apple history, Apple Maps. In reality Siri worked great at launch but, like any new platform under unexpectedly massive load, required scaling adjustments and 24 hour workdays.
After switching from Android to iOS one of the things I missed most was Google Now in comparison to Siri. Before I was far more likely to say "Okay Google" and get surprisingly contextual results.
Something always make me scratch my head. Is there any proof that actual people use and value Google and Amazon assistant more than Siri? Or is the leadership only designed by autoproclaimed experts and funky ad-hoc metrics?
I totally confess that I use Siri for only 3 things:
-Desactivating all wake up alarm during my morning poo
-Setting timers while baking
-Sending a « I’m coming » text to my loved one before riding my bike home
Yes, this is a very dumb and limited usage. But I don’t see Android users around me doing incredible things that I could be jealous of with their voice assistant. I am missing something? Or maybe it’s because I’m not in the US and we still have to catch up with the trend of using voice assistant more?
The things you listed are my primary uses of Google Assistant, but here's some others. What time is it in [location]? What's the temperature outside? Change the temperature inside (nest). Turn off all the lights.
Really the only thing you can't do with Siri that I'm aware of is home automation. But I could be wrong about that.
Siri can do home automation with HomeKit and it actually works really well. The disadvantage here is that HomeKit has less supported accessories because, as I understand it, it's less open/requires certification of the accessory.
That said, for engineers in the crowd I know that I and others get around it by making a HomeKit RPi adapter for things that don't support it...
I thought Apple required HomeKit devices to go through a certification process that includes an audit of device security controls (things like secret management, crypto, default passwords, etc) — and very few of these low-cost IoT devices worry about security.
Apple won’t certify them, but there’s not yet a mass market for secure IoT devices (which would necessarily cost more due to the extra engineering required). It’s a bit of a chicken-and-egg problem.
I recently got a certified smart lock, and it's been superb. Any communications outside my network are handled through Apple/HomeKit and I trust them a lot more than some random Chinese IoT supplier.
> That said, for engineers in the crowd I know that I and others get around it by making a HomeKit RPi adapter for things that don't support it...
Yep, they're using HomeBridge. [0]
I'm planning on doing this for my garage door (with a little WiFi dev board) and for a video doorbell (with a RPi with camera). Others have used HomeBridge with a cheap $20 Chinese IP camera to do the same thing, although you have to install an alternate firmware if you don't want the camera to phone home to China... :O
I have a Kwikset Kevo that I enjoy. If I have guests they just install the Kevo app and I can send them temporary keys that expire. I don't use the integration, but apparently it's integrated with Alexa.
Sorry, slow to reply. It's a Schlage Sense. Major selling point for me was that it has a keypad, so I can give non-tech savvy folks an easy in (plus backup in case of flat phone). If you don't want that, there are nicer looking one.
Google's voice recognition is excellent. I trust Assistant to help me throughout my day with a number of things now:
- Cooking & laundry timers
- Reminders (remind me to do X when I get home)
- Directions (Take me to the nearest X, take me home, etc)
- Checking hours of business (what time does X close today)
- Initiating phone calls (Call X)
- Checking weather (is it going to snow today? How cold is it right now?)
- Conversions (what's X ounces to ml)
- Quick stock checks (what's X stock price)
- Playing music around the house (play X on the kitchen speakers)
Additionally, I've found myself using Android's voice recording capabilities to dictate text all the time. It's just so much faster than typing or swiping.
I haven't an iPhone for a few years, so I can't compare with what Siri is capable of today.
This is what I experienced years ago when I last had an iPhone. Google's voice recognition is exceptional. I can mumble words and it still seems to contextually figure out what I was probably trying to say.
Unless you're picking from a list of options, at which point Google becomes utterly useless to me.
I've literally spent 10 minutes in a "SMS, WhatsApp or Hangout" "SMS" "SMS, WhatsApp or Hangout" loop before, trying every inflection I could think of.
How about continuity of communication between Google Home & Phone.
In my kitchen, I can ask Google Home "how far is X" .. and it responds with a distance/time and pops up a notification on my phone to initiate Google Maps navigation to the target. Useful if I'm in getting ready to go there.
Does Apple's HomePod have any such continuity with an iPhone?
This is strictly anecdotal but a friend of mine is an exec at Facebook and we were recently discussing phones. He had an iPhone for a few years and has now had an Android for a couple years. He said that by far the most drastic difference between the two platforms is Siri vs. Assistant (or whatever Google's assistant is called in the phone). He said it is no contest how much more useful and functional Assistant is, compared to Siri. We did not dwell on this discussion and I otherwise find him an objective, un-biased user of the two platforms, so there you have it.
But it's not personified to the degree Siri is. I noted this elsewhere in the thread.
Doing that bit of theatre, with some humor, pop culture relevant things, even just good responses, appears to matter more to people than "it can do it", and "it actually did it" hit rates.
OK Google needs a name.
Google is this big, smart thing. People get that.
But, "Ask Siri" has less friction and appears easier to identify with than giving an order to a corporation is, "OK Google."
Maybe just, "Hey Google" would make a subtle difference.
And, in typical Apple fashion, "Ask Siri" is all you need to know to attempt it.
Further, discovery has error. If that can be best guess, combined with some fun, people will play.
"OK Google" lacks these human friendly connotations.
A name would be a follow, hard for Google, but... If it were me, I would flat out own that, get the name, and run ADS. "We hear you... meet [name not common, familiar sounding, friendly, easy to say.]"
There has to be either an old world name with good connotations, or a variant that is catchy, fun, gender fluid.
Lol, use goog. Mostly kidding, but a tiny bit serious.
Better, much better is out there.
Or, let the user name it. Why not?
"OK Google, can I call you inigo?"
"Sure, I will answer to inigo. And OK Google when I'm not properly introduced, OK?"
I'm not a huge fan of Siri's 'jokey' personality, but I could live with it if it actually worked. As it is, it just feels extra frustrating to have a 'personal assistant' that is extremely dumb and cracks jokes.
Supposedly these devices have low-power chips that can only recognize certain pre-configured names (like "OK, Google"), and then wake up the main CPU to do the rest.
Cant speak for the other person, but as someone who's used iPhones exclusively since the original, Siri makes mistakes with sufficient frequency that it's subjectively just bad. I can't quantify what percentage of requests fail to be understood correctly, but it's sufficiently high that the frequency is annoying. It's even worse when using the Apple Watch (Series 3).
Unexpectedly, this can sometimes be a positive too. "Hey siri, turn my lights <pink/orange/purple/blue/red/white>" has on multiple occasions been interpreted as "Hey siri, turn my lights", which then results in the lights being turned off. This is hugely entertaining when you're demonstrating your new Philips Hue bulbs over FaceTime, and the person on the other end now just sees your camera go black instead.
I often want to set a timer for exactly 50 minutes so I know when my laundry is done.
But siri always interprets "Fifty" as "Fifteen". Pretty much every single time, it always makes the same mistake, no matter how clearly or slowly I try to pronounce it. I have to say "Hey Siri, set a timer for five zero minutes" to get it to pick it up accurately.
Americans think I sound British.
British think I sound American.
Siri fails less often when set to British instead of American for me, but still fails too often regardless.
I'm a native English speaker, though whether I have an accent depends on your perspective! I can do either American or British somewhat convincingly, and I usually don't have any trouble being understood by Siri.
I suspect the problem with "50" is that whatever ML underpins it is heavily tuned/biased towards "15" because a lot of people probably set 15 minute timers all the time. 50 minute timers are much less common, I suspect?
Walt Mossberg has a funny "series" on his Twitter where he screencaps all the blatent failures he has with Siri. Many times, he or a follower will attach a screenshot of Assistant completing the action. It's not scientific by any means, but it's fun to go through and provides some concrete examples. His biggest issue seems to be that Siri is unreliable - there are queries that work, but only sometimes, and that makes the experience of using the product terrible.
If Apple engineers who work on Siri were required to do all of their corporate communication with Siri exclusively, Siri would rule inside a year.
That is assuming they were trusted to work on it instead of report to status meetings, or meetings to their dotted line manager, etc. Apple was built around Steve. It’s going to take a while.
It's how reliable it is when it does it. I use both iOS and Android, and Google Assistant is so much more reliable that I find I use it. With Siri I find I end up having to intervene so often it isn't worth it.
I do have an Australian accent, so maybe that is a factor.
Another one is when Siri tries to update an existing reminder when I say “create reminder”
I have Apple Music and I’ll often use my little Echo Dot when I want to play a song just because I can simply talk to Alexa from across the room and it will hear me.
I’ve started using Siri to control music when I drive. “Hey Siri” followed by “pause music”, “next track”, “go back”, “play music” all work as expected.
All of those Siri can do maybe half the time. In my experience, it's a completely unreliable service. Sort of embarrassing when your voice assistant can't perform the most basic possible task on an iPhone: locking the screen.
I ask Siri for the closing time of different stores all the time. "Hey Siri, when does X close?" works for me. On Siri most things don't work for me but that does.
I've yet to meet anyone who got siri to do anything, let alone useful, on the 5th try. I can mumble microcontroller part numbers to google, or other odd non-standard words/phrases and 90% of the time its correct on the first try; half the time i'm astonished that it even was able to figure out what i said
Android's voice typing has been weird for me. It used to work pretty well offline, then a few months ago all the punctuation stopped working. Now it's working again and the overall recognition is better than ever.
which will, of course, trigger the reminder as you walk down the street away from home, forcing you to go back and get the change of clothes that you forgot.
Interesting. I guess I thought that discussing how to model a software problem would be an acceptable and interesting conversation on HN. Do any of the downvoters care to contribute in a productive way? I'd be curious to hear what people found disagreeable in my question.
Nest doesn't have to be accurate to the minute to be effective. A reminder based on before I leave work would be.
For instance, if I don't know what time aim getting off of work and I wanted a reminder to call my wife when I got to the car, it could do that completely accurately based on phone connecting to the car. It could only guess about doing something before I left work.
I use it for these things, they're ok. I occasionally try to get it to control an app, it usually fails, but that's my dream state: I'd like to be able to say, "Hey Google, open app X and do Y task."
But I don't know if there's enough incentive for apps to add an assistant interface. How easy is it to add commands on your app that it will accept via assistant? Are there common APIs or structures for this to shift the incentives, allow it to proliferate app by app, rather than waiting on the assistant developers to manually do each one?
After that we need to go back to Aza Raskin's project, uh, Ubiquity? Anyone could build little scripts, verbs, that helped build a natural language input from the ground up. That's how we get to stuff like, "Hey Google, take a screenshot of news.ycombinator.com and email it to Joe."
> Are there common APIs or structures for this to shift the incentives, allow it to proliferate app by app, rather than waiting on the assistant developers to manually do each one?
I think once somebody finds a good API for that, it will be a watershed moment.
Amazon Alexa has apps called "skills", but they expose their own voice interface which is always inconsistent.
I always thought that Android's intents could be a good API for voice commands. It actually surprised me that we never really got much deep app integration with voice on Android phones given that API.
I can open app and ask for actions on Google assistant on my Pixel phone. I use things like
- "Play <song> on Spotify"
- "Send message <message> to <someone> on WhatsApp"
- "Take a selfie or screenshot"
Maybe app actions are only available on Pixel devices.
No, they are available on all devices (at least on my Nexus 6P too). It's only a matter of which apps are implementing the "Actions on Google Assistant" and for which operations.
I want every app to start thinking about voice ux, but that's probably down to the individual apps, and you're right that some do. I can play some things on Stitcher, say, though I can't remember if I have to use podcast names specifically or if I can use my playlist names.
Some Google apps don't seem to support it though.
Ok Google, log my weight on Google fit as X? Search results page.
Ok Google, play (my playlist name) on Google Play: plays unrelated radio station inspired by my playlist name, with sometimes comical results.
Sigh. That’s the main reason I won’t move to the s9. I already have an iPhone and Siri just sucks. Why would I want another sucky assistant? It’s like hiring the wrong person... twice.
I really wish the manufacturers would only have a few features, like weather or timer or location (home, work, school) that were truly bulletproof in execution. It feels like they are trying to do too much most of the time when I just want them to not suck consistently so I can actually use them reliably.
Checking hours of the business works well. The integration with Yelp has that done pretty well.
I've actually been super impressed with Siri. You can have it call you an uber, get the leading scorer from your favorite hockey team, contextually ask followup questions about a person you were searching, etc. I haven't used Siri in a long time (besides timers, weather, etc) and it has come so far since I did.
> Is there any proof that actual people use and value Google and Amazon assistant more than Siri?
I found at least one measurement report: As of July 2017 Verto Analytics showed Siri on top[1] with 41.4M unique users per month (but falling), and Alexa with only 2.6M (but rising fast).
Well from my experience my wife never uses Siri on her iphone and always tells me to ask Google on mine even when she has her iphone in hand. I always tell her just ask Siri and her response is always, "you know Siri sucks". I honestly know this but only tell her to ask Siri to annoy her for interrupting me to ask Google something on my phone.
Honestly my own experience makes me wonder how many of those "users per month" either accidentally invoked it by holding down the button on their phones or had Siri just pop up and start yakking because its hotword detection went wrong.
My company had to switch from Microsoft to IPhones. I hear Siri every day. Usually followed by swearing.
When I got mine, I was trapped in some Siri -> Lock Screen -> Siri combination. "Siri go away" and "Siri show me the home screen" didn't help either. I just wanted to get on the home screen but since there is no Back-Button...well it's disabled now.
I recently switched from being an iphone user to a pixel 2, and I can say that the difference between siri and the google assistant is signifiant. Enough so that I rarely used siri and I use the google assistant every day.
It isn't that the google assistant is more featured packed, more so that it just works much better. Especially with slightly more complex queries.
I've had similar confusion based on the anecdata: none of the android users i know (including myself) ever use the Google Assistant. Or if they do, they don't do it in public. most iPhone users i know use siri at least a little bit. So is Siri actually better? Or is there something else going on that makes it more usable even if it's technically not as "good"
And the quantity of articles complaining about siri would seem to back up the idea that siri is getting used a lot more - Google Assistant is certainly not perfect, it's got some glaring flaws [1]. If people were actually using it, they'd be complaining. The lack of press coverage about Google Assistant's weak points is not a good thing.
I use google voice a ton. For dictating messages while in the car, and for setting reminders and calendar notifications while walking. Probbably adds up to around an hour of speaking a week.
I have a related anecdote. After being on Android since version 1.5 I’ve recently switched over to an iPhone (8) due to privacy and security concerns. Thats when I begun using the voice assistance for the first time. For me, it was a trust issue - I simply had an easier time trusting Apple’s Siri with my voice. On Android I would disable every non-essential Google feature that I could. I don’t feel like I need to do that on IOS.
Would say your friend is unusual. People that have GA I know use it all the time including myself. I also carry an iPhone and never use Siri like never.
My wife and I have an iOS and Android phone, and an Alexa home thingy.
- Android definitely reads our voice better. Alexa is a close second. iOS has to be helped in loud or non-ideal situations.
- iPhone maps/GPS are way worse too.
Often times I'll just hand my wife my phone in the car because Apple maps is taking forever or, giving INSANE directions that take us through tons of backroads to save .12839083 minutes on our drive when the highway is a clear direction that any human (or good AI) would have recommended.
As for the Alexa, we use it every day and it works great. "Alexa set alarm for 5." "AM or PM?" "AM." Boom. Done.
And I'm NO Android zealot. I have tons of problems with it. The app store seems to be way less damaging to your phone for iOS. Like, the first result for any phone problems for Android is "delete some apps." Ugh, the entire point of having a smart phone is apps. So if "removing all your apps" is the solution to the only advantage of having a smart phone, then your entire phone architecture is insane. That'd be like saying "Hey, my car keeps stalling." "Well, it only stalls when you drive, so don't drive it then." "Okay."
So far unmentioned in this discussion is the only reason I use Android voice assistant: Android Auto. When I'm driving my car I obviously can't type on my phone, so I use my voice for everything: changing the navigation destination, calling for music to be played, dialing the phone and sending text messages, amusing children by asking it stupid questions, etc.
I carry an iPhone on my person but I use Android in my car. I don't ever find myself using Siri, but I use Google Assistant a lot. I tried using CarPlay a few times but the voice recognition just wasn't there. I distinctly remember trying to get it to navigate to a restaurant called Da Nang Quan and it just was stumped. Google can recognize these kinds of things. I've also noticed it has no problem recognizing the spoken commands of my young children which generally speaking Siri can't figure out at all.
So I just tried to replicate the query for the restaurant Da Nang Quan (which is in Oakland California) and this time it actually makes a sensible interpretation of my question ("Danang Quan") once out of five times, and got close another time ("Danang Kwan") and got complete nonsense the other three times. Unfortunately all three of the results it tried to give me for the nearly-right questions were in Vietnam which again is not anything like the answer I needed.
There's an Ike's Sandwiches like literally 30 feet from my office in San Jose, and I always like to ask Siri when it closes. To this day, I've probably tried on a dozen occasions and it has never once gotten it right. I'm a native English speaker with no accent.
I tried again just now, this is what I got:
Me: "What time does Ike's sandwich close?"
Siri heard: "What time does Eichs sandwich close"
Siri responded: "Hmmm... I don't see any hours for Rick's Sandwich Shop on Elkins St, Boston"
Things like this is why I've all but stop using Siri.
We bought an amazon device to use as a kitchen timer. It sits right next to where everyone charges their iPhones. Nobody uses Siri. It’s sad. I want to love her, but I honestly can’t remember the last time she was useful, she only works with apples’ ecosystem (and I refuse to use maps it’s tried to kill me on multiple occasions), and I can’t honestly remember the last time I was impressed with her progress forward as a tech.
If the sum of 20 years of voice technology is a glorified stopwatch, you got to wonder...
OT but what do you mean by "English speaker with no accent"? Clearly every person has an accent, and every system that understands language must be able to understand the various accents that people have.
In the US, people say "no accent" to mean what the rest of the world calls an American Accent (even though American Accents change with region, just like any country). It's a very US-centric way of looking at things, but means no harm.
I've lived here in the US for 25 years and I notice the phrase every single time. :-)
Right, you caught me :) what I really meant to say is American English speaker with no strong regional markers (i.e. Southern, New York). Maybe slightly Midwest, but I haven't been back in decades so it's pretty weak.
More importantly, I was distinguishing because while Siri should be able to to understand various accents, it's struggled with that. When it launched, there were only two English variants (US and UK), so if you were a native Indian or Australian English speaker (for example) your experience would suffer. Now, Siri has support for 9 different English variants so they've clearly come a long way, but I remember it wasn't always the case.
There is no such store as "Ike's Sandwiches". There's "Ike's Place", and there's "Ike's Love and Sandwiches". You might have better luck if you use it's actual name.
That was part of my point though. Siri should be smart enough to understand my meaning (and Apple heavily marketed that aspect of it). If I put "Ike's Sandwiches" in Google (or even Apple Maps) they both know what I mean. Siri somehow thinks I'd be interested in a restaurant over a thousand miles away.
My wife clicks the shutter on her iPhone and without touching an additional button walks later into our family room and ask for fine detail in photos and they appear in 4k.
We also have a 4k Chromecast and a Google home.
Just buy, plug in and log in. She already used Google photos.
One of my favorite things that Alexa can do, which Siri still can't, is integrate with my Logitech Harmony and SmartThings hubs. This lets me control all my AV devices and Lutron Z-Wave lights using voice commands.
I hear now there's an unofficial bridge (runs as a NodeJS service) called Homebridge which might get Harmony working with Siri, and Lutron has newer lights not using Z-Wave that are Homekit compatible, but when I was shopping for all this stuff several years ago Siri /HomeKit support was abysmal.
It also helps that Echo's microphones and speech recognition is much better at far distances, with background noise, than "Hey Siri" (and assuming I don't have my phone in my pocket). At this point, I pretty much only use Siri when I'm in the car, and that's the environment where it seems to perform the worst (long delays to start listening, frequent misunderstandings, request failures).
Smartthings is the platform to beat for Apple. It’s simple, easy and incredibly powerful. I can’t understand why Apple doesn’t apply their magic to the model and just knock it out of the park. They have all of the pieces and the ux experience to make it exceptional.
Siri is something that's occasionally useful, if you think to use it.
Alexa and now Google Home can integrate themselves into your whole life. Alexa is the primary interface for the home automation at my home and my girlfriend's. Her 9 y/o kids can use it. Our 67 y/o mothers can use it. We've got a baker's dozen of Alexa-enabled speakers between us, covering all of our living spaces and half the bathrooms.
They get used constantly.
Many of our use cases are "trivial" but having a voice assistant everywhere instead of tied to phones / tablets greatly changes their utility.
Most ordinary people I know favor Siri despite it's differences.
The "character", theatre, framing, if you will, with Siri is superior.
Few people in my circles use the Amazon device at present.
Somehow, people like Siri. Siri is an agent in their minds and that is how they treat Siri.
The "OK Google" is one too, but not as well personified.
If I were to hazard a gut guess, based on an anecdotal sample, personification, as well as relevance to pop culture are big draws and can make up for technical gaps.
Personally, it's no contest. I use Google voice constantly. Works well enough for me to continue to invest in the experience.
> "Is there any proof that actual people use and value Google and Amazon assistant more than Siri?"
This is a common confusion among iPhone users. I suspect the reason is that if all you're used to is Siri (and it's myriad disappointments), it's hard to conceive of the the quantum leaps that is Google Assistant in product polish and usefulness.
First and foremost is the near flawless voice recognition. I, with my India accent even, can just say things to my Android device (and now, my new google home) it just ... gets it. It's spooky.
Then, after it recognizes the question/request correctly, it is very likely to take the correct action with a high enough probability that it's now almost a surprise to hear "Sorry I don't now how to do that yet" when it does happen.
Outside of the active queries, there are things it does automatically that are at the level of what a particularly attentive human assistant might do for you.
It'll check your calendar and remind you that you should start driving for your 4pm appointment because the traffic is unusually heavy on 101 today.
Or that your flight tomorrow morning has been cancelled. It knows about your flight because of the confirmation email in your gmail account.
Remember, you have to take no action for this sort of reminders. Just have to opt in when assistant is setting itself up on your new phone.
There are myriad other such examples. It has saved my bacon numerous times. We're truly living in the future.
You nailed it. People think this technology is not that useful because they have only used Siri and Alexa. Use the Google Assistant and you will see it is very helpful.
I have used all three and can say from personal experience the Google Assistant is head and shoulders in front of the other two in UX.
I use Google Assistant all day, and know people who do the same. It's actually amazing, and although Apple was the first one to add any decent voice assistant, Google was the first one that understood my accent at all ("Indian English"). When we had that, it was a shame to even ask any question to Siri. Now when Siri has an Indian English version as well, I still compare it all the time with my friend's latest iPhone, and it still sucks. It understands me every time.
As far as I can tell, the only advantage Siri has over Google Assistant is its integration with Wolfram Alpha. If you ask Siri "What is the angular diameter of Mars?" it will tell you the current angular diameter of Mars. Google Assistant can't do that. Apart from the Wolfram Alpha connection, Siri is inferior in every way, IMHO.
Agree completely. Siri works near perfectly for my basic uses: Alarms, timers, reminders, weather, stock prices, music, calls, texts etc. I concede that Google and Amazon's assistants are more advanced, if the measure of advanced is the number of gimmicky joke responses and Easter eggs they can say during a social gathering.
Personally, I get the most mileage out of Google's voice assistant when I'm driving.
It's a situation when I can't (well, shouldn't) use the phone's UI, so it's great to ask it to:
-find a place and check its hours of operation
-navigate to said place
-find a gas station along the way
-etc.
On an occasion, I've asked it to add things to a calendar.
As a lazy person, I prefer to do things with the minimal number of steps. Scheduling things in a calendar requires an inordinate amount of clicks. On the other hand, it's something that a CLI could be well-suited for. For such things, voice control is natural.
I see us at the beginning of a path that grants magic powers, controlling everything with our voice, or even our intention.
That requires machine learning, which requires data, which requires data collection, and the closest one to Google there isn't Apple or Amazon, it's Facebook.
It's sort of sad to read a story about a story, and not be able to figure out which problems were created in the source vs. the retelling.
Mashable: "Right now, it looks like Siri won't be blown up and a [sp] rebuilt". Yeah, well, that'd most likely be a huge mistake[1] depending on your definition of "blown up" and "rebuilt", and also Siri's already been rebuilt twice.[2]
The Information: "One of Apple’s original plans was to launch its speaker without Siri included, according to a source." Of course — there are always contingency plans. To me, this is a much more damning comment about the Siri team than Apple's "vision".
Regarding [1], the move to Mesos was not a rewrite. Some core components had to be improved, but [2] was primarily an infrastructure upgrade that happened to make Siri considerably faster by ditching hypervisors. It was a massive feat that was accomplished by relatively few engineers.
There were always many components under review for reengineering; keep in mind that Siri was released before WebSockets was a thing, let alone RFC.
The snowball effect of Siri's failures originated from iOS Program Management pushing us to expand into as many other languages as soon as possible from the beginning, thus diverting resources from implementing or polishing original "Siri, Inc." features and various "nice to have" improvements to en-US usage.
Source: I worked on Siri from the moment it was acquired, and no longer work at Apple. Richard Williamson's recent comments are pathetic.
Yeah, claiming that moving to Mesos was rebuilding Siri is stretching it a bit. The beauty of the Mesos transition was that most services needed only minor modifications to be moved to the new scheduler, and still get the benefits of better resource utilization.
Regarding building a speaker without Siri, the origin story I read from Bloomberg was that the idea of a great-sounding speaker that would adapt to room acoustics was dreamed up by Mac audio engineers, before Echo was even announced, so a speaker with a smart assistant built-in probably wasn't even on their radar at that point. I didn't necessarily interpret it as a lack of faith in the Siri team to deliver.
Team dysfunction seems to be a part of it. The 'head' guy for Siri, Dag Kittlaus left Apple to start Viv Labs in about a year since Siri was released. Their assistant Viv, looked pretty okay[1], and was since acquired by Samsung[2].
I feel if he had stayed on, Siri would have been better.
Samsung seems to be trying pretty hard to get people to use it[3]. It's relatively new so I'm interested to see where it goes, especially since it has the lowest market share.
Not to mention Williamson shit-talking the Siri team to this reporter:
"Williamson denies the accusations that he slowed Siri development down and instead cast blame on Siri's creators.
'It was slow, when it worked at all,' Williamson said. 'The software was riddled with serious bugs. Those problems lie entirely with the original Siri team, certainly not me.'"
I remember that Viv got positive press when launched. But how good did it actually become? It seems to me we'd be hearing more about Samsung's voice assistant if Viv did really live up to the hype.
How/for what do people actually use Siri, or Alexa for?
I mean I can't think of having seen anyone use them for a long time for anything but to either test them out for curiositys sake, or for demoing their new Alexa controlled drapes.
Hopefully this wasn't too off-topic. Maybe HN should have an ability to create a poll and attach them to a thread. ;-)
I use an Echo Dot on my nightstand to control my LifX lightbulb, adjusting its brightness between 1% and 100%, as my wife and I prefer extremely dim lights right before bed. When my daughter was younger, I also used the same Dot to set timers when we were trying to give her a few minutes to get herself back to sleep when she'd wake up crying.
I use a full-size Echo in the kitchen to set kitchen timers, to populate shopping lists, and to check the weather occasionally as we're going out the door. I also sometimes play music on it.
Recently, I've started using the Google Assistant on my phone to start music in the car (so as to avoid physically touching my phone while driving), and I've started to decide that it's easier to say, "Hey Google, navigate to 123 main st" instead of using the touchscreen, even when I could use the touch screen.
One thing my wife and I use the Google home for constantly is grocery shopping. We had previously used an app called our groceries that synchronizes a grocery list between multiple people. That's a map where to plug in for Google home where you can just say things to it and it adds it to your synchronized grocery list. We use it all the time when we notice we run out of something.
Having two newborns has made me come to appreciate Alexa in a whole new way: my hands are almost always full with one (or both) babies and being able to use my voice to control the lights, play music for them, set reminders or alarms for myself, has been a godsend.
While my wife was bedridden after her C-Section surgery, she loved being able to use Echo's ability to call other Echo's in the house (or outside the home) when she needed something. Kind of like an intercom system. The problem with the iPhone is that it was usually either on the charger, or battery was dead, or it was left on vibrate. The Echo was always 100% reliable for her, which is really critical when you can't even get out of bed.
It's what happens at large orgs, esp ones with leadership problems - they can't move fast to improve products, even when something is obviously not working.
The only thing I use Siri for is on the desktop for starting apps - I've hooked it up to a keyb shortcut. I find its a bit faster than fumbling around spotlight or the dock. Alfred would possibly be even faster.
Google assistant has the best knowledge graph for generic questions, and much better machine learning models for voice recognition and language parsing. I primarily use it for route finding, finding out opening times, currency conversions, sometimes calling people.
Alexa for almost only music, and timers, since it's hooked up to my kitchen speakers. It's terrible for general knowledge questions.
Siri is frustratingly bad/limited. It’s been kind of impossible to deny for a while now. I wonder what they’re going to do, because they had better do something. And the Viv people are at Samsung now so they can’t just buy that.
The title seems poorly worded to me. I read it as "Why Siri underlies Google Assistant and Alexa," as if Siri were the backing technology. Should say something more like "Why Siri is lagging behind Google Assistant and Alexa."
> "It was slow, when it worked at all," Williamson said. "The software was riddled with serious bugs. Those problems lie entirely with the original Siri team, certainly not me."
Wow. It those were his original words, I don't doubt a bit that this guy should be responsible for Siri being left behind.
I know, I couldn't believe the chief of a major software product at Apple would talk smack about the previous team. It's what a newbie might do when he has screwed something up. Classic dodging.
Conversely you get handed a pile of steaming hot shit they call legacy code(and it lacks documentation and the variables lack names that hint at any meaning) and they're telling you it's completely your fault that you couldn't get something to work properly. Getting handed a huge amount of technical debt is a very real problem.
The industry does not lack programmers who can clean up technical debt or outright rewrite a legacy system. At least, I’m pretty sure that Apple doesn’t lack such programmers.
A VP’s job is to find those people and give them the correct incentives. The VP has the rights to hire new people and restructure teams. The VP should use those leverages.
At the VP level, how a particular codebase or team is doing simply doesn’t matter, as the VP has all the leverages for organizational shifts. Blaming a codebase or team only makes sense if the whole company cannot produce quality code, and it’s impossible to hire new employees that can. I am pretty sure that this is not the case for Apple.
A hopefull adequate comparison would be this: you are a technical lead of at a company, you get really really high pay and you have a lot of fellow engineers under your command. You get handed a legacy system, and there are all all kinds of problems with it:
1. The system uses a database that is not duplicated.
2. The system has no monitoring.
3. The web frontend is written in PHP.
What do you do?
To fix 1, duplicate the freaking database.
To fix 2, add some freaking monitoring.
3 is presumably the hardest to fix. But remember that you get really high pay and you have a lot of engineers. What did the engineers at Facebook do? They fixed PHP.
Do you blame the previous team that made this horrible system? No. No. No. You own the system now. There are valid engineering approaches to fix. So fix it. It can take a while, but eventually you can fix it. In the worst case, you will have to rewrite. But that's still a feasible thing to do.
I agree with this, the one thing I'll add is that you may also have to contend with the competing priorities of every other VP. Specifically in that Business Development VP may be saying, "We don't have time to do any of those things! We need X done now otherwise we won't get our customer!" And the technical VP response might be, "If we don't fix this now then we won't have _any_ customers." But in the end the financial state of the company will make the decision for you. If you're on the edge and you immediately need more customers to survive then you're going to forgo doing 1 & 2 and instead implement X that BD was requesting. Hoping that you can get enough breathing room in the near future to actually do 1 & 2.
Does Apple have the resources for this project in particular? I really don't know. However in the general context, I think you lack understanding of two things:
1. You cannot simply put your standard programmer(or any number of them) with undergrad (even with superb knowledge) of combinatorial algoritms in the former position of someone with a graduate degree that focused on algorithms engineering who didn't give a damn about even the smallest aspects of software engineering like version control or documentation. You have even less ability to do this on an embedded device for numerically sensitive algorithms. I've been in this position (except with a far broader background in mathematics and algorithms engineering) at least a couple times with other engineers with graduate degrees in the requisite areas and (I'm not exaggerating) it is truly a fucking nightmare.
2. Just because you are a VP doesn't mean your managers are willing to approve the amount money that goes into a number of algorithms engineers fixing or rewriting a product that just barely functions well enough. I am incredibly close to a couple people that are VPs at companies with ~10,000 people and you would be surprised with how little power they can have on hiring sometimes. Companies are not bottomless pits of money and the people at the top usually treat it as such.
So Forstall and his direct report were responsible for two botched product launches, Siri and Maps. This is a big blow to the camp suggesting firing Forstall was a mistake.
Siri may have stalled now but it definitely wasn’t a botched launch. And Apple maps may not work great in all cases but it did succeed in forcing Google to provide decent maps on iOS. Before Apple Maps the vector maps were an Android exclusive.
Apple Maps is like Bing: almost as good as the #1 service, but almost is not enough to make anyone care.
Apple Maps is remarkably good today, doing nearly all of the important things well. Here in Australia, it's on par with Google Maps for driving (slightly worse for novel route planning, slightly better for guidance) and when travelling to major world cities I've found Apple Maps is frequently better than Google Maps for navigating on foot and public transport.
when in major cities in China, my experience is that Google Maps is almost unusable (not due to the censorship, as I use a roaming SIM which avoids those problems, but rather due to wildly incorrect data, missing metro lines, etc), and Apple Maps is at least on par with the quality that Google Maps achieves in other countries, and often much better.
I find it interesting that Google doesn't care enough to add the most recent several metro lines in (for example) Beijing or Shanghai, considering that many people try to use Google Maps in China when visiting and usually seem to end up switching to Apple Maps as a result of the complete trash the Google Maps experience is there. I wonder if they switch back when they go home.
In reality Google Maps is farther ahead of its competitors than most people realize.
Google Maps now provides wholly personalized points of interest and attention affordances in its UI. It's also well on its way to 3D scanning and annotating POIs in every structure in the world, including very large pedestrian-only interiors. It also has realtime transit and traffic feeds, and more pedestrian trails and venue detail than anyone else. All of these are areas where its competition is not just behind, but falling further behind. But the personalization combined with superior POI search really illustrates how Google's competitors don't even have the platform to properly conceptualize the type of product Google is building, much less build it.
I don’t know if I’d say most of these things matter enough at the end of the day, just to cherry pick, Apple has 3D, interiors, traffic, transit and pedestrians. Not everywhere, but it’s a problem which time and money can fix. Both of which apple has in abundance.
Also anecdotally I’m not sure if I can agree that Google’s POIs are vastly better, mostly just different but of a similar low quality (which is to say neither pleases me).
It still feels years behind any dedicated maps though(like TomTom), at least in those countries that I've tried it(UK, Spain, Poland, Germany). It's just consistently wrong about which roads are connected with each other, it doesn't know that certain roads are one-way or that you can't make a left/right turn out of a certain street. I've reported dozens of mistakes and they just don't get fixed. Every single morning on my way to work it tries to route me left on a no-left-turn intersection - I've literally filed a map correction report for this 3 or 4 times, still nothing.
TomTom just don't have this issue, I've been using it for 5 years now and it's never done anything like that. Also TomTom Traffic is vastly better than Google Traffic, consistently beating it for time and route prediction(I've been driving with Android Auto and TomTom 5000 set up side by side for a while, and if TomTom suggests a detour it's usually a good one - take the one google suggests and you're getting stuck on some small street that google thinks you can drive through but you actually can't).
You might be right, but most of those reasons don’t actually matter. To be frank, the only things I miss when not using Google maps is their highlighting of high activity areas.
Both maps have similarly good cartography where it matters. In most countries, both maps are similarly accurate.
When you set hard deadlines for software release dates because you want to have tentpole features for IOS releases, and you mostly refuse to do widespread beta tests because then it won't be surprising, you're creating the maps debacle. Now I'm sure maps was harder than it looked, but I'd also bet lots of people on the inside knew that. They should have let the date slip.
I honestly think Jobs thought that no problem is hard enough that you can't cajole or threaten staff into just working harder. You see it all through his career.
We are hitting territory where there are now problems so hard that it will now take time to create the software. He never got that.
This is all moot with respect to Jobs. He resigned due to health reasons 20 months before Apple Maps was launched and died almost a year before it was launched.
They couldn't let the date slip, because the deal with Google was set to expire and not having a built-in maps application at all was not an option.
The question you need to ask is, did a pushy management style make Apple Maps better than it would have been otherwise or worse. I think it's hard to argue that pushing the team hard didn't produce a better product than they would have had otherwise.
As an engineer who reported to Williamson, I’ve got to say I was far more intimidated by him than I ever was of Forstall. He rarely smiled in a meeting and his old school heavy handed management style always took the fun out of the work.
Honestly, I don’t know the full details, but I still hold that Forstall took the fall for the poor execution of Williamson in the Maps debacle.
>Honestly, I don’t know the full details, but I still hold that Forstall took the fall for the poor execution of Williamson in the Maps debacle.
Well that's not true because Eddy Cue fired Williamson [1]. Maps might have been the straw that broke the camel's back, but Forstall was fired also because he couldn't work well with Bob Mansfield (who came back out of retirement after he was gone [2]) and Jony Ive.
Then please set up a new account and tell us about it. I really like Forstall direction, and I dont think Maps was that much of a problem given it is still pretty poor after he left.
"@Jessicalessin This statement, wholly false, was made by the architect and head of the biggest launch disaster in Apple history, Apple Maps. In reality Siri worked great at launch but, like any new platform under unexpectedly massive load, required scaling adjustments and 24 hour workdays."
Apple fell behind in voice on Jobs' watch and prior to the Siri acquisition. In fact, it's probably more accurate to say that Apple was behind on iOS progress and facing a ho-hum 4s launch. Hence, the quick 8-figure Siri buy and hurried integration that put them behind the 8-ball on voice ever since.
The broader context is that Apple doesn't seem to get: "It's the 'AI,' stupid!"
As an aside (OT), it's interesting the way Mashable essentially republished paywall-protected content. If there was new reporting there, I missed it. But that's the nature of modern media.
If the Siri infrastructure is so bad I don’t understand why Apple doesn’t already have several skunkwork teams trying to create a fresh alternative for the past few years.
Siri is very dumb. I use for for reminders and alarms.
I speak fr-ca as opposed to fr-fr, and some conversations go like this:
Siri, call X on cell
Sorry, can’t call X
Siri, call X on mobile
Calling X on cell
In fr-ca, we talk about cell phones. In the contact app, the mobile phone is registered as cell phone, but in fr-fr, they use mobile.
I also sometimes have this conversation:
Siri, rappelle-moi de mettre des mouchoirs sur la liste de matériel du camp
(Siri, remind-me to put tissues on the unit’s list of material)
Do you want me to create a list?
I only have one list, on purpose. I have no way to tell Siri that I never want to create new lists.
It appears to me as if some words have more importance than others, such as the word « list ».
Those are my biggest annoyances with Siri. I will continue to speak and use my iPhone in fr-ca, to promote my language and improve the software by providing data points about which languages are used.
This is my same experience. The only things I can get Siri to do reliably is to route me home, or to do simple math problems. Also very simple short term reminders.
Eh, I'm not sure. The percentage of iPhone users worldwide who can speak some English is probably pretty high. As a non-native speaker myself, I'd certainly prefer a reliable assistant in English over an unreliable one in my native language.
FWIW the assistants probably would not recognize most of these people's accents. Not to speak of whether people would be inclined or not to make the effort to speak a foreign language to their assistant.
I'm French, have a good level in English, all my OSes are set in English. I used to use Siri in English but stopped cause it was weird. The ability to select a different language for Siri than the system language is a blessing.
Not being able to flag requests that Siri failed to parse tells me that the Siri team isn't serious about making Siri better.
Just adding a small "That isn't what I asked" button or letting me say something like "Hey Siri, you got my last request wrong" would give Apple a mountain of usable data.
There's actually a way to do this, for a long time. In older versions, you'd see a wavy underline on certain words it was less confident about, letting you choose alternatives. The current version has a "Tap to edit" link.
Problem is discoverability. And having a mountain of "this request was wrong", without knowing what the answer should have been, would require too much manual sifting.
In my opinion, the problem with these assistants is that interacting with them using voice give you the impression that they're intelligent, but it's just a trick.
When it works, it can be useful for certain tasks, but when it doesn't work, you lose trust in the system pretty quickly, especially if you're in a situation where you expect and need the assistant to work, like when driving.
All the known techniques for NLP and NLU are mediocre. From Apples perspective, it makes a lot of sense to not invest so much money for such little gain. Google Assistant is supposedly better, but I don’t notice/care. Google/Amazon pour a lot of money into their personal assistant tech, but the end result is hardly noticeable or useful.
285 comments
[ 2.7 ms ] story [ 185 ms ] threadAs Steve Jobs said, once you become a VP you can’t make excuses anymore. You have crossed the Rubicon and need to start taking responsibility for the failures of your team.
This guy has a lot to learn about leadership.
Also, how different is that "Information" report behind the paywall? It would not surprise me to see Mashable rewriting an article but to do so with paid content seems inappropriate.
"So this is why Siri is way behind the Google Assistant and Alexa".
The submission title should be changed.
Edit: Actually it happens on every refresh in Chrome as well.
I wonder if/when they'll start slipping in machine-generated headlines for a small % of the audience to train clickbait bots.
"How Apple let Siri fall behind the Google Assistant and Alexa"
"Why Siri is so dumb"
> "Hewlett and Packard built a great company, and they thought they had left it in good hands, but now it's being dismembered and destroyed. It's tragic. I hope I've left a stronger legacy so that will never happen at Apple."
In creating Apple University he was really trying to instill a legacy that would outlast him. That said, when pulling out the "Steve Jobs never would have let this happen" card, it's also important to remember that MobileMe happened on Steve's watch, as did the iPod HiFi.
Whereas Sculley, Spindler and Amelio were a parade of failure from start to finish.
Look at what the U.S. President has and continues to do, and he still has his job, and a much more important and public one too.
Politics aside, Jobs' turnaround of Apple into the most valuable company in the world (for awhile, recently) would have ensured he remained at the helm for as long as he wanted.
The no-poaching collusion between Apple et al probably cost Apple a nine-figure sum to settle (I'm not sure what the exact breakdown between the defendants was): https://en.wikipedia.org/wiki/High-Tech_Employee_Antitrust_L...
Both of these were quite serious issues, in my opinion.
Those set the tone for products and the company with users and Apple hasn’t been nearly as effective at this without him.
Steve wouldn't have wanted you to connect your phone to your computer in the first place. That's why iTunes got wireless sync back before everything was in the cloud. The only really remaining piece was developers needing to plug in for deployment and debugging, but who cares - they're developers, they'll manage.
And now even that's solved!
> This statement, wholly false, was made by the architect and head of the biggest launch disaster in Apple history, Apple Maps. In reality Siri worked great at launch but, like any new platform under unexpectedly massive load, required scaling adjustments and 24 hour workdays.
https://twitter.com/Dagk/status/974047743596404737
I totally confess that I use Siri for only 3 things:
-Desactivating all wake up alarm during my morning poo
-Setting timers while baking
-Sending a « I’m coming » text to my loved one before riding my bike home
Yes, this is a very dumb and limited usage. But I don’t see Android users around me doing incredible things that I could be jealous of with their voice assistant. I am missing something? Or maybe it’s because I’m not in the US and we still have to catch up with the trend of using voice assistant more?
Really the only thing you can't do with Siri that I'm aware of is home automation. But I could be wrong about that.
That said, for engineers in the crowd I know that I and others get around it by making a HomeKit RPi adapter for things that don't support it...
Apple won’t certify them, but there’s not yet a mass market for secure IoT devices (which would necessarily cost more due to the extra engineering required). It’s a bit of a chicken-and-egg problem.
> That said, for engineers in the crowd I know that I and others get around it by making a HomeKit RPi adapter for things that don't support it...
Yep, they're using HomeBridge. [0]
I'm planning on doing this for my garage door (with a little WiFi dev board) and for a video doorbell (with a RPi with camera). Others have used HomeBridge with a cheap $20 Chinese IP camera to do the same thing, although you have to install an alternate firmware if you don't want the camera to phone home to China... :O
[0]: https://github.com/nfarina/homebridge
For all other scenes I usually use the physical hue dimmer button because I’m standing.
- Cooking & laundry timers
- Reminders (remind me to do X when I get home)
- Directions (Take me to the nearest X, take me home, etc)
- Checking hours of business (what time does X close today)
- Initiating phone calls (Call X)
- Checking weather (is it going to snow today? How cold is it right now?)
- Conversions (what's X ounces to ml)
- Quick stock checks (what's X stock price)
- Playing music around the house (play X on the kitchen speakers)
Additionally, I've found myself using Android's voice recording capabilities to dictate text all the time. It's just so much faster than typing or swiping.
I haven't an iPhone for a few years, so I can't compare with what Siri is capable of today.
I've literally spent 10 minutes in a "SMS, WhatsApp or Hangout" "SMS" "SMS, WhatsApp or Hangout" loop before, trying every inflection I could think of.
In my kitchen, I can ask Google Home "how far is X" .. and it responds with a distance/time and pops up a notification on my phone to initiate Google Maps navigation to the target. Useful if I'm in getting ready to go there.
Does Apple's HomePod have any such continuity with an iPhone?
The Google alternative is, from what I understand, something that just works.
Doing that bit of theatre, with some humor, pop culture relevant things, even just good responses, appears to matter more to people than "it can do it", and "it actually did it" hit rates.
OK Google needs a name.
Google is this big, smart thing. People get that.
But, "Ask Siri" has less friction and appears easier to identify with than giving an order to a corporation is, "OK Google."
Maybe just, "Hey Google" would make a subtle difference.
And, in typical Apple fashion, "Ask Siri" is all you need to know to attempt it.
Further, discovery has error. If that can be best guess, combined with some fun, people will play.
"OK Google" lacks these human friendly connotations.
A name would be a follow, hard for Google, but... If it were me, I would flat out own that, get the name, and run ADS. "We hear you... meet [name not common, familiar sounding, friendly, easy to say.]"
There has to be either an old world name with good connotations, or a variant that is catchy, fun, gender fluid.
Lol, use goog. Mostly kidding, but a tiny bit serious.
Better, much better is out there.
Or, let the user name it. Why not?
"OK Google, can I call you inigo?"
"Sure, I will answer to inigo. And OK Google when I'm not properly introduced, OK?"
Got downvotes, but it seems to me Google has the superior product. True for me and how I use it anyway.
Add in a little personification, and it's nothing but win, IMHO.
So far, I've seen people like the bit of theatre related to Siri. Whether we do or not isn't necessarily the point.
Supposedly these devices have low-power chips that can only recognize certain pre-configured names (like "OK, Google"), and then wake up the main CPU to do the rest.
I always found the phrase ok Google simply awkward to pronounce
Unexpectedly, this can sometimes be a positive too. "Hey siri, turn my lights <pink/orange/purple/blue/red/white>" has on multiple occasions been interpreted as "Hey siri, turn my lights", which then results in the lights being turned off. This is hugely entertaining when you're demonstrating your new Philips Hue bulbs over FaceTime, and the person on the other end now just sees your camera go black instead.
But siri always interprets "Fifty" as "Fifteen". Pretty much every single time, it always makes the same mistake, no matter how clearly or slowly I try to pronounce it. I have to say "Hey Siri, set a timer for five zero minutes" to get it to pick it up accurately.
I suspect the problem with "50" is that whatever ML underpins it is heavily tuned/biased towards "15" because a lot of people probably set 15 minute timers all the time. 50 minute timers are much less common, I suspect?
That is assuming they were trusted to work on it instead of report to status meetings, or meetings to their dotted line manager, etc. Apple was built around Steve. It’s going to take a while.
It's how reliable it is when it does it. I use both iOS and Android, and Google Assistant is so much more reliable that I find I use it. With Siri I find I end up having to intervene so often it isn't worth it.
I do have an Australian accent, so maybe that is a factor.
http://h4labs.org/siri-next-monday-isnt-christmas/
Another one is when Siri tries to update an existing reminder when I say “create reminder”
I have Apple Music and I’ll often use my little Echo Dot when I want to play a song just because I can simply talk to Alexa from across the room and it will hear me.
Go try the Google assistant and get back to us that Siri is equal.
It's bonkers to me that this is still a limitation on a phone that lets me track a dozen-plus world clocks, as though it's a more common use case.
I've yet to meet anyone who got siri to do anything, let alone useful, on the 5th try. I can mumble microcontroller part numbers to google, or other odd non-standard words/phrases and 90% of the time its correct on the first try; half the time i'm astonished that it even was able to figure out what i said
One thing it can't do is 'remind me to do X when I leave work'. Can siri?
If I usually leave work around 5-6 and I start packing up my bag, that could be a signal.
For a lot of people it would be effective.
For instance, if I don't know what time aim getting off of work and I wanted a reminder to call my wife when I got to the car, it could do that completely accurately based on phone connecting to the car. It could only guess about doing something before I left work.
But I don't know if there's enough incentive for apps to add an assistant interface. How easy is it to add commands on your app that it will accept via assistant? Are there common APIs or structures for this to shift the incentives, allow it to proliferate app by app, rather than waiting on the assistant developers to manually do each one?
After that we need to go back to Aza Raskin's project, uh, Ubiquity? Anyone could build little scripts, verbs, that helped build a natural language input from the ground up. That's how we get to stuff like, "Hey Google, take a screenshot of news.ycombinator.com and email it to Joe."
I think once somebody finds a good API for that, it will be a watershed moment.
Amazon Alexa has apps called "skills", but they expose their own voice interface which is always inconsistent.
Maybe app actions are only available on Pixel devices.
I want every app to start thinking about voice ux, but that's probably down to the individual apps, and you're right that some do. I can play some things on Stitcher, say, though I can't remember if I have to use podcast names specifically or if I can use my playlist names.
Some Google apps don't seem to support it though.
Ok Google, log my weight on Google fit as X? Search results page.
Ok Google, play (my playlist name) on Google Play: plays unrelated radio station inspired by my playlist name, with sometimes comical results.
I'm sure it will improve with time.
I really wish the manufacturers would only have a few features, like weather or timer or location (home, work, school) that were truly bulletproof in execution. It feels like they are trying to do too much most of the time when I just want them to not suck consistently so I can actually use them reliably.
Going through those:
- Cooking & laundry timers - Yes. Works well
- Reminders (remind me to do X when I get home) - I find Apple geofenced alerts to be unreliable.
- Directions (Take me to the nearest X, take me home, etc) - Pretty good, search isn't as good as Google, but good enough these days.
- Checking hours of business (what time does X close today) - Never tried, doubt it would work.
- Initiating phone calls (Call X) - Yes - Works well
- Checking weather (is it going to snow today? How cold is it right now?) - Yes, works well
- Conversions (what's X ounces to ml) - Never tried. Just tried now - works OK
- Quick stock checks (what's X stock price) - As above works OK, says whether there has been a fall or rise.
- Playing music around the house (play X on the kitchen speakers)- works very well.
I've actually been super impressed with Siri. You can have it call you an uber, get the leading scorer from your favorite hockey team, contextually ask followup questions about a person you were searching, etc. I haven't used Siri in a long time (besides timers, weather, etc) and it has come so far since I did.
I found at least one measurement report: As of July 2017 Verto Analytics showed Siri on top[1] with 41.4M unique users per month (but falling), and Alexa with only 2.6M (but rising fast).
[1] https://techcrunch.com/2017/07/11/siri-usage-and-engagement-...
When I got mine, I was trapped in some Siri -> Lock Screen -> Siri combination. "Siri go away" and "Siri show me the home screen" didn't help either. I just wanted to get on the home screen but since there is no Back-Button...well it's disabled now.
It isn't that the google assistant is more featured packed, more so that it just works much better. Especially with slightly more complex queries.
And the quantity of articles complaining about siri would seem to back up the idea that siri is getting used a lot more - Google Assistant is certainly not perfect, it's got some glaring flaws [1]. If people were actually using it, they'd be complaining. The lack of press coverage about Google Assistant's weak points is not a good thing.
[1]https://www.androidpolice.com/2018/03/14/10-stupidly-simple-...
- Android definitely reads our voice better. Alexa is a close second. iOS has to be helped in loud or non-ideal situations.
- iPhone maps/GPS are way worse too.
Often times I'll just hand my wife my phone in the car because Apple maps is taking forever or, giving INSANE directions that take us through tons of backroads to save .12839083 minutes on our drive when the highway is a clear direction that any human (or good AI) would have recommended.
As for the Alexa, we use it every day and it works great. "Alexa set alarm for 5." "AM or PM?" "AM." Boom. Done.
And I'm NO Android zealot. I have tons of problems with it. The app store seems to be way less damaging to your phone for iOS. Like, the first result for any phone problems for Android is "delete some apps." Ugh, the entire point of having a smart phone is apps. So if "removing all your apps" is the solution to the only advantage of having a smart phone, then your entire phone architecture is insane. That'd be like saying "Hey, my car keeps stalling." "Well, it only stalls when you drive, so don't drive it then." "Okay."
I carry an iPhone on my person but I use Android in my car. I don't ever find myself using Siri, but I use Google Assistant a lot. I tried using CarPlay a few times but the voice recognition just wasn't there. I distinctly remember trying to get it to navigate to a restaurant called Da Nang Quan and it just was stumped. Google can recognize these kinds of things. I've also noticed it has no problem recognizing the spoken commands of my young children which generally speaking Siri can't figure out at all.
This confirms my experience so far with Siri.
I tried again just now, this is what I got: Me: "What time does Ike's sandwich close?" Siri heard: "What time does Eichs sandwich close" Siri responded: "Hmmm... I don't see any hours for Rick's Sandwich Shop on Elkins St, Boston"
Things like this is why I've all but stop using Siri.
(I use it for reminders. That seems to work. Looking up anything is useless.)
If the sum of 20 years of voice technology is a glorified stopwatch, you got to wonder...
Obviously that's "no accent."
I've lived here in the US for 25 years and I notice the phrase every single time. :-)
More importantly, I was distinguishing because while Siri should be able to to understand various accents, it's struggled with that. When it launched, there were only two English variants (US and UK), so if you were a native Indian or Australian English speaker (for example) your experience would suffer. Now, Siri has support for 9 different English variants so they've clearly come a long way, but I remember it wasn't always the case.
We also have a 4k Chromecast and a Google home.
Just buy, plug in and log in. She already used Google photos.
I hear now there's an unofficial bridge (runs as a NodeJS service) called Homebridge which might get Harmony working with Siri, and Lutron has newer lights not using Z-Wave that are Homekit compatible, but when I was shopping for all this stuff several years ago Siri /HomeKit support was abysmal.
It also helps that Echo's microphones and speech recognition is much better at far distances, with background noise, than "Hey Siri" (and assuming I don't have my phone in my pocket). At this point, I pretty much only use Siri when I'm in the car, and that's the environment where it seems to perform the worst (long delays to start listening, frequent misunderstandings, request failures).
Alexa and now Google Home can integrate themselves into your whole life. Alexa is the primary interface for the home automation at my home and my girlfriend's. Her 9 y/o kids can use it. Our 67 y/o mothers can use it. We've got a baker's dozen of Alexa-enabled speakers between us, covering all of our living spaces and half the bathrooms.
They get used constantly.
Many of our use cases are "trivial" but having a voice assistant everywhere instead of tied to phones / tablets greatly changes their utility.
The "character", theatre, framing, if you will, with Siri is superior.
Few people in my circles use the Amazon device at present.
Somehow, people like Siri. Siri is an agent in their minds and that is how they treat Siri.
The "OK Google" is one too, but not as well personified.
If I were to hazard a gut guess, based on an anecdotal sample, personification, as well as relevance to pop culture are big draws and can make up for technical gaps.
Personally, it's no contest. I use Google voice constantly. Works well enough for me to continue to invest in the experience.
This is a common confusion among iPhone users. I suspect the reason is that if all you're used to is Siri (and it's myriad disappointments), it's hard to conceive of the the quantum leaps that is Google Assistant in product polish and usefulness.
First and foremost is the near flawless voice recognition. I, with my India accent even, can just say things to my Android device (and now, my new google home) it just ... gets it. It's spooky.
Then, after it recognizes the question/request correctly, it is very likely to take the correct action with a high enough probability that it's now almost a surprise to hear "Sorry I don't now how to do that yet" when it does happen.
Outside of the active queries, there are things it does automatically that are at the level of what a particularly attentive human assistant might do for you.
It'll check your calendar and remind you that you should start driving for your 4pm appointment because the traffic is unusually heavy on 101 today.
Or that your flight tomorrow morning has been cancelled. It knows about your flight because of the confirmation email in your gmail account.
Remember, you have to take no action for this sort of reminders. Just have to opt in when assistant is setting itself up on your new phone.
There are myriad other such examples. It has saved my bacon numerous times. We're truly living in the future.
I have used all three and can say from personal experience the Google Assistant is head and shoulders in front of the other two in UX.
I use it for all sorts of things that I would never try to get Siri to do.
It's a situation when I can't (well, shouldn't) use the phone's UI, so it's great to ask it to:
-find a place and check its hours of operation
-navigate to said place
-find a gas station along the way
-etc.
On an occasion, I've asked it to add things to a calendar.
As a lazy person, I prefer to do things with the minimal number of steps. Scheduling things in a calendar requires an inordinate amount of clicks. On the other hand, it's something that a CLI could be well-suited for. For such things, voice control is natural.
That requires machine learning, which requires data, which requires data collection, and the closest one to Google there isn't Apple or Amazon, it's Facebook.
Mashable: "Right now, it looks like Siri won't be blown up and a [sp] rebuilt". Yeah, well, that'd most likely be a huge mistake[1] depending on your definition of "blown up" and "rebuilt", and also Siri's already been rebuilt twice.[2]
The Information: "One of Apple’s original plans was to launch its speaker without Siri included, according to a source." Of course — there are always contingency plans. To me, this is a much more damning comment about the Siri team than Apple's "vision".
[1] https://www.joelonsoftware.com/2000/04/06/things-you-should-...
[2] Mesosphere blog post, definitely reviewed and approved by Apple: http://archive.is/YNMLh
There were always many components under review for reengineering; keep in mind that Siri was released before WebSockets was a thing, let alone RFC.
The snowball effect of Siri's failures originated from iOS Program Management pushing us to expand into as many other languages as soon as possible from the beginning, thus diverting resources from implementing or polishing original "Siri, Inc." features and various "nice to have" improvements to en-US usage.
Source: I worked on Siri from the moment it was acquired, and no longer work at Apple. Richard Williamson's recent comments are pathetic.
Regarding building a speaker without Siri, the origin story I read from Bloomberg was that the idea of a great-sounding speaker that would adapt to room acoustics was dreamed up by Mac audio engineers, before Echo was even announced, so a speaker with a smart assistant built-in probably wasn't even on their radar at that point. I didn't necessarily interpret it as a lack of faith in the Siri team to deliver.
I feel if he had stayed on, Siri would have been better.
Samsung seems to be trying pretty hard to get people to use it[3]. It's relatively new so I'm interested to see where it goes, especially since it has the lowest market share.
[1]:https://www.youtube.com/watch?v=MI07aeZqeco
[2]: https://venturebeat.com/2017/02/28/samsung-paid-around-215-m...
[3]: https://youtu.be/zlesUiYkp_4?t=55s
"Williamson denies the accusations that he slowed Siri development down and instead cast blame on Siri's creators.
'It was slow, when it worked at all,' Williamson said. 'The software was riddled with serious bugs. Those problems lie entirely with the original Siri team, certainly not me.'"
https://twitter.com/Dagk/status/974047743596404737
I mean I can't think of having seen anyone use them for a long time for anything but to either test them out for curiositys sake, or for demoing their new Alexa controlled drapes.
Hopefully this wasn't too off-topic. Maybe HN should have an ability to create a poll and attach them to a thread. ;-)
I use an Echo Dot on my nightstand to control my LifX lightbulb, adjusting its brightness between 1% and 100%, as my wife and I prefer extremely dim lights right before bed. When my daughter was younger, I also used the same Dot to set timers when we were trying to give her a few minutes to get herself back to sleep when she'd wake up crying.
I use a full-size Echo in the kitchen to set kitchen timers, to populate shopping lists, and to check the weather occasionally as we're going out the door. I also sometimes play music on it.
Recently, I've started using the Google Assistant on my phone to start music in the car (so as to avoid physically touching my phone while driving), and I've started to decide that it's easier to say, "Hey Google, navigate to 123 main st" instead of using the touchscreen, even when I could use the touch screen.
While my wife was bedridden after her C-Section surgery, she loved being able to use Echo's ability to call other Echo's in the house (or outside the home) when she needed something. Kind of like an intercom system. The problem with the iPhone is that it was usually either on the charger, or battery was dead, or it was left on vibrate. The Echo was always 100% reliable for her, which is really critical when you can't even get out of bed.
The only thing I use Siri for is on the desktop for starting apps - I've hooked it up to a keyb shortcut. I find its a bit faster than fumbling around spotlight or the dock. Alfred would possibly be even faster.
Google assistant has the best knowledge graph for generic questions, and much better machine learning models for voice recognition and language parsing. I primarily use it for route finding, finding out opening times, currency conversions, sometimes calling people.
Alexa for almost only music, and timers, since it's hooked up to my kitchen speakers. It's terrible for general knowledge questions.
For a smart speaker, yes.
Wow. It those were his original words, I don't doubt a bit that this guy should be responsible for Siri being left behind.
> "Somewhere between the janitor and the CEO, reasons stop mattering," says Jobs, adding, that Rubicon is "crossed when you become a VP."
> In other words, you have no excuse for failure. You are now responsible for any mistakes that happen, and it doesn't matter what you say.
http://www.businessinsider.com/steve-jobs-on-the-difference-...
A VP’s job is to find those people and give them the correct incentives. The VP has the rights to hire new people and restructure teams. The VP should use those leverages.
At the VP level, how a particular codebase or team is doing simply doesn’t matter, as the VP has all the leverages for organizational shifts. Blaming a codebase or team only makes sense if the whole company cannot produce quality code, and it’s impossible to hire new employees that can. I am pretty sure that this is not the case for Apple.
1. The system uses a database that is not duplicated.
2. The system has no monitoring.
3. The web frontend is written in PHP.
What do you do?
To fix 1, duplicate the freaking database.
To fix 2, add some freaking monitoring.
3 is presumably the hardest to fix. But remember that you get really high pay and you have a lot of engineers. What did the engineers at Facebook do? They fixed PHP.
Do you blame the previous team that made this horrible system? No. No. No. You own the system now. There are valid engineering approaches to fix. So fix it. It can take a while, but eventually you can fix it. In the worst case, you will have to rewrite. But that's still a feasible thing to do.
Now substitute technical lead with VP.
1. You cannot simply put your standard programmer(or any number of them) with undergrad (even with superb knowledge) of combinatorial algoritms in the former position of someone with a graduate degree that focused on algorithms engineering who didn't give a damn about even the smallest aspects of software engineering like version control or documentation. You have even less ability to do this on an embedded device for numerically sensitive algorithms. I've been in this position (except with a far broader background in mathematics and algorithms engineering) at least a couple times with other engineers with graduate degrees in the requisite areas and (I'm not exaggerating) it is truly a fucking nightmare.
2. Just because you are a VP doesn't mean your managers are willing to approve the amount money that goes into a number of algorithms engineers fixing or rewriting a product that just barely functions well enough. I am incredibly close to a couple people that are VPs at companies with ~10,000 people and you would be surprised with how little power they can have on hiring sometimes. Companies are not bottomless pits of money and the people at the top usually treat it as such.
Surprising that a person at his level wouldn't be mature enough and self-aware enough to realize that.
Apple Maps is remarkably good today, doing nearly all of the important things well. Here in Australia, it's on par with Google Maps for driving (slightly worse for novel route planning, slightly better for guidance) and when travelling to major world cities I've found Apple Maps is frequently better than Google Maps for navigating on foot and public transport.
I find it interesting that Google doesn't care enough to add the most recent several metro lines in (for example) Beijing or Shanghai, considering that many people try to use Google Maps in China when visiting and usually seem to end up switching to Apple Maps as a result of the complete trash the Google Maps experience is there. I wonder if they switch back when they go home.
It's good enough that I use it primarily on my phone over Google Maps, in the UK
Google Maps now provides wholly personalized points of interest and attention affordances in its UI. It's also well on its way to 3D scanning and annotating POIs in every structure in the world, including very large pedestrian-only interiors. It also has realtime transit and traffic feeds, and more pedestrian trails and venue detail than anyone else. All of these are areas where its competition is not just behind, but falling further behind. But the personalization combined with superior POI search really illustrates how Google's competitors don't even have the platform to properly conceptualize the type of product Google is building, much less build it.
Also anecdotally I’m not sure if I can agree that Google’s POIs are vastly better, mostly just different but of a similar low quality (which is to say neither pleases me).
TomTom just don't have this issue, I've been using it for 5 years now and it's never done anything like that. Also TomTom Traffic is vastly better than Google Traffic, consistently beating it for time and route prediction(I've been driving with Android Auto and TomTom 5000 set up side by side for a while, and if TomTom suggests a detour it's usually a good one - take the one google suggests and you're getting stuck on some small street that google thinks you can drive through but you actually can't).
Both maps have similarly good cartography where it matters. In most countries, both maps are similarly accurate.
When you set hard deadlines for software release dates because you want to have tentpole features for IOS releases, and you mostly refuse to do widespread beta tests because then it won't be surprising, you're creating the maps debacle. Now I'm sure maps was harder than it looked, but I'd also bet lots of people on the inside knew that. They should have let the date slip.
We are hitting territory where there are now problems so hard that it will now take time to create the software. He never got that.
They couldn't let the date slip, because the deal with Google was set to expire and not having a built-in maps application at all was not an option.
The question you need to ask is, did a pushy management style make Apple Maps better than it would have been otherwise or worse. I think it's hard to argue that pushing the team hard didn't produce a better product than they would have had otherwise.
Honestly, I don’t know the full details, but I still hold that Forstall took the fall for the poor execution of Williamson in the Maps debacle.
Well that's not true because Eddy Cue fired Williamson [1]. Maps might have been the straw that broke the camel's back, but Forstall was fired also because he couldn't work well with Bob Mansfield (who came back out of retirement after he was gone [2]) and Jony Ive.
[1] https://www.cultofmac.com/203032/apple-svp-eddy-cue-fires-ri...
[2] http://allthingsd.com/20121101/behind-silicon-valleys-un-ret...
"@Jessicalessin This statement, wholly false, was made by the architect and head of the biggest launch disaster in Apple history, Apple Maps. In reality Siri worked great at launch but, like any new platform under unexpectedly massive load, required scaling adjustments and 24 hour workdays."
https://twitter.com/Dagk/status/974047743596404737
The broader context is that Apple doesn't seem to get: "It's the 'AI,' stupid!"
As an aside (OT), it's interesting the way Mashable essentially republished paywall-protected content. If there was new reporting there, I missed it. But that's the nature of modern media.
I speak fr-ca as opposed to fr-fr, and some conversations go like this:
In fr-ca, we talk about cell phones. In the contact app, the mobile phone is registered as cell phone, but in fr-fr, they use mobile.I also sometimes have this conversation:
I only have one list, on purpose. I have no way to tell Siri that I never want to create new lists.It appears to me as if some words have more importance than others, such as the word « list ».
Those are my biggest annoyances with Siri. I will continue to speak and use my iPhone in fr-ca, to promote my language and improve the software by providing data points about which languages are used.
Siri is still better than Alexa for most of the world....
I'm French, have a good level in English, all my OSes are set in English. I used to use Siri in English but stopped cause it was weird. The ability to select a different language for Siri than the system language is a blessing.
Just adding a small "That isn't what I asked" button or letting me say something like "Hey Siri, you got my last request wrong" would give Apple a mountain of usable data.
Problem is discoverability. And having a mountain of "this request was wrong", without knowing what the answer should have been, would require too much manual sifting.
When it works, it can be useful for certain tasks, but when it doesn't work, you lose trust in the system pretty quickly, especially if you're in a situation where you expect and need the assistant to work, like when driving.