Hacker News new | past | comments | ask | show | jobs | submit login
Upcoming changes to Cortana (microsoft.com)
73 points by waltherp 49 days ago | hide | past | favorite | 71 comments

Love current PR lingo. Just shove self-praise into every piece of communication.

> Cortana is continuing to redefine the nature of a digital assistant, ... accelerating productivity to help save you time and focus on the things that matter most

> The first change is to end support for all third-party Cortana skills

> Cortana in Surface Headphones will continue pivoting toward its mission to help customers with productivity throughout their day. We’ll be removing support for the previous version of Cortana in the first version of Surface Headphones

So much acceleration, wow.

My favorite:

> As we make this shift toward a transformational AI-powered assistant experience in Microsoft 365...

While I’m sure there’s a world of difference, I couldn’t help but chuckle at the thought of Clippy making a comeback. “transformational AI-powered assistant experience in Microsoft 365” certainly sounds like a PR-enhanced description of Clippy.

Hah, it sounds more like Clippy wrote this PR puff piece!

I can just imagine PR Clippy...

"It looks like you're trying to write spin, can I help you with that?"

>Also in 2021, Cortana in Surface Headphones will continue pivoting toward its mission to help customers with productivity throughout their day.

... By shutting it down? Maybe they found out people were only using it to slack off

We would like to reiterate that we have not sent Cortana to the glue factory!

> As we make this shift toward a transformational AI-powered assistant experience in Microsoft 365, we need to adjust our focus areas of innovation and development to give our customers assistance where they need it most. As a result, we are making changes to some U.S. consumer-centric features and functionalities with lower usage.

I can’t even parse that last line. “Making changes [to functionality] with lower usage.”

Does that mean that it is less used? Or does that mean they are lowering usage?

To clarify, they aren't shutting down Cortana completely - they're ending support for the standalone Cortana apps, which makes sense. What I find more interesting is how they're handling their smart speaker partnership:

> ...we’ve worked closely with Harman Kardon to create a Bluetooth-enabled device transition plan that we hope will help ease the impact of this change. Customers who receive a firmware update from Harman Kardon in early 2021 will still be able to continue listening...on their Invoke via Bluetooth.

If I'm reading this correctly, they're basically saying "very few people really use this as a 'smart' speaker, so we'll just turn it into a normal Bluetooth speaker and move on." I'm not sure whether that says more about Cortana or about the smart speaker market, but it seems like an...ok compromise? Certainly better than just bricking the thing.

I think very few use the voice feature, but those who do will hate losing it. I wouldn’t be surprised if it turned out that only a few percent of iPhone users uses Siri.

The hype surrounding digital voice assistants have died out. It’s still useful in a few situations, but not having one isn’t as huge a defeat as it would have been 5 years ago. Microsoft could drop Cortana completely and be fine.

I never bother with such devices, because they hardly ever support European Portuguese properly and I am not going to speak foreign languages at home just to make a device happy.

I'm a British English speaker, and about 15 years ago I spent an afternoon experimenting with the basic voice recognition on OSX. I found that I had to mimic a US or Australian accent to get it to work, and I wondered what my next-door neighbours were thinking when they heard me cycle repeatedly through "Give me the phone number for Firstname Lastname" with British, US and Australian accents.

Scot here, I have this problem with telephone voice menus. Actually, I've also had this problem in the US as well, particularly while on the telephone. Putting on a fake US accent seems to work, but it's so damn embarrassing and I guess could be construed as offensive by some people (if they thought I was taking the piss).

I had the same issue recently with a BMW voice activation system I was playing with. Mimicking an English accent worked for that one!

It has gotten better, I think; as a joke, I recently asked it "what is weather tomorrow" in the most horribly broken fake accent I could muster, and to my surprise, it correctly annotated the question and responded.

When I still had a Windows Phone, I couldn't use Cortana because my phone was set in an unsupported language and it wouldn't let me use Cortana in English - if I wanted to use it, I had to switch the entire interface to English.

I would have spoken to it in a foreign language if they had let me. Instead, I have no idea to this day what Cortana sounds like.

European Portuguese? Does that mean they support Brazilian Portuguese better that actual Portuguese?

Probably, if they support Portuguese at all.

That is a pet peeve that we have back home, because of market size, many US companies rather support Brazilian Portuguese than any of other variants.

European Portuguese gets left out due to market size, while African and Asian Portuguese get left out due to buying capabilites.

Thing is, having to speak to a device, in another language variant just feels weird, leaving out the little detail that one does need to learn the differences as well.

TIL about Luso-Asians and Luso-Indians https://en.m.wikipedia.org/wiki/Luso-Asians

> actual Portuguese

would you consider British English the only actual English?

Not only Brazil has 20 times the population of Portugal, but there are considerations to be made about how languages evolve in the periphery, which might not apply to Portuguese, but certainly apply to Canadian French vs French French.

No and many English variants suffer the same problems of US English first.

Try having a foreign wife!

I mean, it's not to make the device happy. It's to make the device do what you want, without you having to get up.

Respect though.

That only re-inforces the cycle of not providing Portuguese support.

Instead of paying for half working stuff and then complaying like many do here, I rather not give any money at all.

The only thing I regularly use Siri for is timers, where it’s genuinely quicker to say “15 minute timer” than it is to open the app.

Same here. And Alexa (Amazon Cube) is really relegated to muting my TV, lowering/raising the volume, and turning TVs off. If I install some outdoor cameras I might use it to check the front door, back yard or (if I had a child) the baby room. As a Windows user I never found a use for Cortana.

Alexa is the only one that turned out remotely useful for me.

I'd like to use the assistant on Android exactly for this, and probably a few other automation tasks. I love me some fiddling with semi-automated workflows. If only I wasn't averse to putting my voiceprint on Google's servers.

In case you’re not aware you can use the Shortcuts app to automate Clock for a timer and put it on the home screen as an icon. Works for Safari and Open Links if you want a home web bookmark, too.

I use Siri to ask it to use Waze for directions to locations I want to drive to. Works incredibly well.

Agreed, although I’ve found force touch on the Control Center icon a nice shortcut as well.

Nice, I've learned something today...

However, when busy, it still feels faster to say "set timer for x minutes" with just a long press on the home button.

It'd be useful in more if they were noticeably better today than when they were introduced. I wonder how much is down to intrinsic shortcomings with voice assistance and how much is just no one's made a good one.

IMO it's intrinsic shortcomings. These started coming out once the language processing got good enough to be reliable. And that part has continued advancing. But after you have that parsed instruction, what do you do with it? That's the part that's basically "AI hard" and is the real limiting factor. The "skills" approach routes around the problem by having exact keywords passed off to particular specialized handlers. But that means it can only ever do what it's been explicitly taught to do, and with explicit keywords.

> f I'm reading this correctly, they're basically saying "very few people really use this as a 'smart' speaker, so we'll just turn it into a normal Bluetooth speaker and move on." I'm not sure whether that says more about Cortana or about the smart speaker market, but it seems like an...ok compromise? Certainly better than just bricking the thing.

I tend to agree: I think switching it to a simple Bluetooth speaker is a reasonable way forward under the circumstances.

As a related aside this, along with a raft of other support withdrawals recently (the most recent that comes to mind is those smart glasses), I do feel somewhat vindicated in my policy to keep things as dumb as possible: speakers are just speakers, lightbulbs are just lightbulbs, my kettle is just a kettle, and - if I do need any smarts - I attach a general purpose computing device to provide them. E.g, it's just incredibly easy to connect my laptop or phone to my hi-fi via a DAC or audio interface to stream music over Spotify.

The point of this is that it means most of my household devices and appliances remain brick-proof, and I shouldn't have to keep replacing them every 3 - 5 years just because manufacturers stop supporting them.

> I do feel somewhat vindicated in my policy to keep things as dumb as possible

I have a similar but marginally more lax policy: I'm fine with smart devices as long as they don't have to phone home to operate and they have on-device controls.

For obviously Internet of Shit-worthy appliances (Cory Doctorow's smart toaster or whatever) I'll always choose the dumb version, and I make sure that I only use home automation stuff that's as widely interoperable as possible, but otherwise I don't particularly fuss about it as long as it meets all of my other requirements.

I think modularity is really key here. As long as you can replace or update each component individually without impacting the rest of it, you should be good.

> I think modularity is really key here. As long as you can replace or update each component individually without impacting the rest of it, you should be good.

That's exactly it: the ability to easily swap out components, and then high levels of interoperability. I think I perhaps didn't express myself so well because I'm broadly in agreement with you. It's the phone home aspect that often makes smart devices a problem.

> Certainly better than just bricking the thing.

I think we've seen a lot of backlash where companies treat their (disposable) products as disposable, and this is a good thing to see.

Also, companies like Harman Kardon have lifetime relationships with their customers.

> they're ending support for the standalone Cortana apps, which makes sense

This is actually the hardest part for me to understand. They want me to drag a Win10 laptop everywhere to use the assistant instead of just using my phone?

You mean your phone that has its own built-in, unswappable assistant? Since Windows Phone died[0], I can't blame them for getting out of a business where they are immediately outleveraged by the competition.

[0] I say this with lament, because it was IMO a legitimately good phone OS. My Lumia 920 is still my backup phone, 8 years later, because the ones I got after that died before replacement. But the death was certainly not a surprise. Once they started adding more features to Android apps than existed in the Windows Phone versions, the writing was on the wall about where the money was going.

They're adding it to the Teams and Outlook apps and pushing Cortana toward businesses.

While they're at it, why don't they include Windows 10 on that list?

I haven't met a single person that uses Cortana and those who tried are usually not coming back. For me Cortana has basically no useful features (maybe its because of my language settings), its just software bloat and in addition to that, its a privacy nightmare. The only association I have with Cortana is it screaming at me when I do a clean install of Windows 10, which is a thing of the past thanks to my switch to macOS.

A favorite pastime of mine was lining up all the new hire computers that had to be setup and seeing if I could mute it before the entire office turned around and told me to shut Cortana up.


Worst part is it keeps running in the background hogging memory and cpu even when completely disabled. I've tried gpedit, shutup10 and so many other tools to get rid of it but somehow it comes back every time :/

The start menu, that need to index your files in background, or do other stuff, is named "cortana".

I don't think you want to disable this "cortana".

Serious question for better programmers than me: Doesn’t it violate seperation of concerns that something like the start menu that has a specific function and has been around for decades get smashed into the Cortana code and renamed. Cortana is clearly not core to the functionality of the start menu and if it became so it surely wasn’t out of necessity right?

It reminds me of when Microsoft was dead set on conflating the file manager “explorer” and Internet Explorer.

Can someone justify this?

It's like monolith and micro services.

From the programming standpoint, a monolith can be splitted in components, but compiled/bundled/linked to a single executable. This doesn't violate the seperation of concern. It's bad taste/marketing/whatever.

I checked on my Windows and they did separate Cortana and the Search/Start menu recently.

Oh thanks I didn't know that. I thought it was the annoying assistant.

If your windows is up to date, there is no cortana anymore.

(Well, it gone on my PC and I did nothing in particular). It was the annoying assistant, and the start menu.

Cortana is now a standalone app that can be uninstalled and updated separately from Microsoft Store.

The general problem with all "AI-powered assistants" is that they're not AI-powered assistants. They are voice interfaces atop tiny libraries of utilities.

If I had an AI powered assistant, it's actual data would be on my system, and not out on the internet somewhere. It wouldn't need to send my speech out.

But further, it would actually be an assistant. It would be able to operate all the applications on my computer.

"Cortana, copy the last three Slack messages in the current channel to a new Box document. Call it Requirements and add the date." -> Uses two separate apps for which I have open sessions, can exercise my credentials (where I complete the MFA when needed).

"Cortana, search all my communications for things related to Bob. I need to trace all of his activities that I have record for. Include Sumologic logs for the applications I support." -> Searches chat logs, email, and web application logs out in my log collection.

"Cortana, find out from Mom what groceries she needs today, and place Instacart orders accordingly."

The "assistants" we have today can't combine actions. They are just isolated pools of menu selections. They can't replace hands-on-keyboard for the things I regularly need to do with my computers or devices. They need to become intention-processors, not just language parsers to activate a menu.

And not even good ones at that. They never take negative re-enforcement, you can't tell them what they just did was wrong or report a problem. You can't say "Good Bot" and let it know that the response it gave was what you expected and to continue to return results similar. Hell, if it was just 'stupid' you could keep programming new features into it and new ways to get to the same features with slightly different wording. But no. We don't even get that. Just this awful black box that "sometimes does what it's asked" and can barely reference previous incantations. Uhg.

Other than for setting timers and turning the lights on and off, all of this "voice assisted" stuff doesn't really work. So providers should either make it work or abandon it if they can't (and yes, they can't), leaving only the scenarios that do work operational. Kudos to Microsoft - they're admitting what was obvious to everyone for quite a while.

This is especially egregious with Siri, which Apple repeatedly tries to gaslight everyone about. Apple does demos every year, yet Siri continues to suck just as it always did. But Apple keeps pretending that it doesn't suck. This goes completely against the ethos of Apple not releasing stuff until it's "insanely great".

Yes, other than timers, reminders, delivery notifications, home automation, weather, music/podcast playback, Bluetooth speaker, unit conversion, and grocery items (e.g. OurGroceries), voice assistants are useless.

The big hurdle for voice UI is discoverability.

That's probably why it's most successful in a handful of verticals, like timers and music control. There are a finite list of likely commands, and people can probably guess them first time.

When you move much beyond that, you start getting a high miss rate because there's no good way to pull a list of valid commands or options.

Importantly, especially for home automation, "option guessability" is a big thing-- you might know the commands, but the relevant options are specific to each location, and you were likely not there when it was configured. How do you know that the light fixtures are named "Left sconce, right sconce, ceiling fan, and table lamp", and not "north, south, Bob, and Matilda?"

Voice UI is also a terrible situation for error recovery. I can't figure out a voice-only UI that would handle a simple case like "which light do you want to turn on?" well. At best, it could read you a list of all the devices that might be relevant for the question, but that's so low information-density and will feel like a bad telephone IVR service.

The "Siri with a screen" design convention might have been sensible, as that would have provided a way to disambiguate queries quickly, with buttons to show the options, and at idle, a browsable list of supported commands.

Of course, at that point, you're making a more obvious machine with a (voice-based) command-line. That's shatters the illusion they're selling of this being your secretary/concierge, with arbitrary and broad capabilities, who happens to be a robot.

Sure, but what did the voice assistants ever do for us?

The other use case is text message dictation while driving - which is excellent.

These types of system can’t be released “only when perfect” unfortunately as they need many iterations with real world data.

If you own Microsoft Surface Headphones you can get a $25 gift card from them: https://www.bing.com/surfaceheadphoneredemption

Also a $50 gift card if you own the Harman Kardon Invoke speakers: https://www.bing.com/cortanainvokeredemption

Pretty crappy that the links to these are buried in the FAQ

So does that mean if you created a third party action word (skill) it's game over?

Cortana worked pretty well--better than Siri, though not as good as Alexa.

Will it still work on my Zune?

People made fun of the Zune but if you look at it objectively it was actually a good product for its time.

The timing and the marketing was perhaps not as good as the product itself.

Anytime I see someone mention the Zune it is always followed by what you just said. And then someone does what I am doing now.

We are all connected in the great Circle of Zune.

I agree though. Maybe except the first or maybe second release, these were kinda rushed and really just reskinned Toshiba devices.

Microsoft is surprisingly good at designing hardware and devices, and surprisingly bad at marketing them. Just like with Windows Phone, they were late to the party but acted like they already own the market. (I guess you just can't help yourself if you dominate the desktop OS market that much.) Both times they added unnecessary restrictions to the devices and set the pricing relatively high. You can do this if you're Apple and created the whole market and are generally known as a premium brand, but MS just isn't.

Their hardware sucks, just like their software. This is from a person who went from Nokia Symbian, to Microsoft's Nokia and finally the dead Windows Phone.

They somehow managed to take Nokia with their amazing hardware (look at Nokia N9, N900 and Maemo) and turn it into pure garbage (Microsoft Lumia series and their POS Windows Phone 10).

Completely different experience for me. Symbian was a confusing mess, so much that I stayed with a feature phone for another while. WP8 felt responsive, the UI consistent and surprisingly clean. Scrolling was smooth in 2013 when on Android it had frequent stuttering except maybe the flagship devices. The problem was that it was missing all the important apps and then some. They made sure to set the bar high for starting to develop apps for it and then wondered why nobody started porting to WP.

Nokia was in the process of switching from Symbian to Maemo, but Microsoft installed a Trojan of a Stephen Elop and forced the crap that is Windows Phone down Nokia's throat.

Oldie but a goodie


The Zune was flashy and sexy looking, but the text on the screen was so large that it conveyed a fraction of the information an iPod screen did in a much smaller size. All the backgrounds and all were also distracting.

It was also huge compared to contemporary iPods. From what I remember, one of the few nice features were the sharing capabilities, where you could Zune songs you had bought over to friends, but those only worked Zune to Zune, so to actually avail that feature it had to first become popular enough that you would frequently meet people who also owned a Zune.

The Zune wasn’t a bad device. It was just too little too late. And the iPhone introduced less than a year later (and the obvious iPod touch that everyone knew was coming) meant that it was all over for the Zune before it even had a chance to get started.

Windows Phone was hostile to both developer and OEM though. By comparison publishing apps on Play Store on Android only requires one time fee and Android was free and open source to OEM except with Play Store, which can be free if several Google Apps is preloaded.

That's what I broadly covered by bad marketing and "thinking you already own the market". Apple could pull it off to be dicks to app developers because they were first to market. You simply had no choice. Google understood they need to catch up to IOS at any cost, so made it as easy as possible for anyone to start mucking around with app development. Sideloading required one change in the settings menu and you could shove as many random apks to your device as possible. For WP you needed to sign up for an account to even just launch the IDE. The included emulator didn't work on Windows Home editions, so you elegantly pissed off the curious teen on their budget PC. Sideloading was pretty much only possible in this dev setup with iirc one app at a time. Microsoft was third to the party and seriously came up with this kind of bullshit. And they wonder why they failed.

One of the assistant's could stand out if they allow 3rd party devs to use it and allow an incognito mode. Otherwise it's just a way for them to collect data^H^H^H^H^H^H^H^H^H^H^H^Hbig brother us.

Good riddance!!

Applications are open for YC Winter 2021

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact