I liked how this has been commented on WinRumors: http://www.winrumors.com/microsoft-research-chief-talks-siri-vs-windows-phone-tellme/#comment-370900438

Seriously, how come Apple always gets credit for Microsoft’s (and others’) innovation?

IBM, anyone?

“Seriously, how come Apple always gets credit for Microsoft’s (and others’) innovation?”

Because they present the innovation to the user in an easy to use way? Have you taken a look at Windows Mobile 6.5 before the iphone came?

@Kurkos
Yeah, revolutionary reinvention and other crap.

I’ve actually had several WinMo devices. I don’t think there was anything terribly wrong with them (for most part Android just follows WinMo model, even source code for ~95% of WinMo has been distributed as part of Platform Builder).

Yes, Apple made some EVOLUTIONARY changes with the iPhone just when technology become available.

Technologically Apple has almost always played catch up. But somehow, when they’ve actually caught up, they manage to convince people that they Revolutionary Innovated everything.

As for Siri, I just don’t get it (it’s OK, I’m not an average customer and don’t understand WP7’s focus on socializing for instance), but maybe someone can explain me what Siri can do that TellMe can’t?

Can you really communicate with your phone (even though I find it weird, still real AI – like in Microsoft’s “vision” videos – would be actually useful)? No, Siri is too limited for that.

Can it launch applications, send SMS, make weather lookups and so on? Sure. Wait!

Bottomline, you are not really unconstrained, but constraints are unknown – you either find your own route to do your routine tasks or play trial-and-error each time you speak to “your personal assistant”.

from one of the comments on Linsuxoid’s link there:

http://www.youtube.com/watch?v=FVkH9Hgvda4

I think the time for adverts to start taking the piss out of the mac-fans is well overdue! XD

“[...] but maybe someone can explain me what Siri can do that TellMe can’t?”

Siri is not keyword based. TellMe is.
Then I can name you such: – It cannot schedule meetings – It cannot setup reminders – It cannot write e-mail – It cannot find you directions – It cannot setup an alarm – It cannot set a timer
( – It cannot find stocks ) – only partially true, if you’d say “Find X stock” it will use bing to find what you’re looking for but from what I know Siri does this differently.

[ I don’t know how good Siri does these though, I’ve never seen it myself – only used TellMe on my WP7 and compared it to the description of Siri ]

This talk about voice enabled assistants is like the talks about speaker amplifiers or “picture enchance” filters on TVs. Since there are no objective measurements for amplifiers and picture enchance filters, everyone considers theirs to be the best.

Similarly, since there are no objective criteria, everyone considers their platform’s assistant to be the best. “I like to give my phone precise commands instead of playing trial and error”, “if an assistant can’t understand natural language, it sucks” etc.

I don’t know, I don’t like talking to my phone, because I use it a lot in public places, but if I had to choose, i would go with Siri, because it looks like it has the most work behind it.

To be honest, barking instructions to your phone seems like a prime way to make yourself look like a total nobhead, regardless of how elegently it may have been implemented. But that’s just me…

@ChrisTX
Makes sense. I could only agree, that Siri can do more, but I still strongly disagree that it’s some sort of Revolution and bla-bla.

@Kurkos
I have even closer comparison: search engines. Bing positions itself as “decision engine” and MS complained about Google teaching people bad search habits (using keywords instead of “natural language”): http://www.huffingtonpost.com/2011/04/11/bing-director-stefan-weitz_n_847790.html
I’m using Bing exclusively for ~2.5 years (or how long is it on the market already?) and still prefer “WinMain” over “What is the prototype of WinMain function, please?”
It’s anecdotal and I realize that I don’t represent any significant portion of the market.

As for “Siri sucks” – it clearly doesn’t: it can do more then TellMe right now (once you find your own “structured query language” to make it work CONSISTENTLY). What sucks is Apple’s positioning of Siri as some sort of Revolutionary (as everybody knows, Apple and Revolution are synonyms) AI when it clearly is not capable enough to be classified as one.

You guys do realise that you’re responding to the fake Jerkface, don’t you?

Voted down because of a combination of that and the fact that it’s as obnoxiously written as you’d expect from its source.

I feel a bit of a tool, really, because there’s the kernel of a good FUD in there. Oh well.

“Seriously, how come Apple always gets credit for Microsoft’s (and others’) innovation?”

Really, keyword crap is now equal to meaning based verbal UI?

With this reasoning, Windows Explorer is a rip-off of UNIX shell – after all, both use text to display files to the user, so they are exactly the same, right?

———————————

Based on what I saw, I’d say Siri is the most impressive Natural UI in active use.

@Doctor

Well, Fake Jerkface is funny (though he’s had better times).

————————————

Admin should have renamed his account, though.
(Hello Admin, are you there?)

@RC:

Funny? Really?

Well, I don’t propose to pursue this any further, but he’s yet to get a laugh out of me. The only posts under his profile (which cover the current month so far) are unedifying, to say the least.

TBH I find Adam far funnier. When he reins himself in and starts “arguing,” he’s just close enough to reality to be worth dealing with.

Fake Jerkface, I’m afraid, is just a boring little troll.

Anyway, don’t let me get in the way of the Siri stuff. Like I say, there’s some FUD-worthy stuff here.

Just a thought, but is this going to tie in to the fabulous iTV? You know, the one people like Dvorak claim is going to use the iPhone as a remote, complete with gestures and all?

If I stood in the middle of the living room (and was married, obviously) and waved my iPhone at the iTV whilst making pinching gestures at the iPhone and yelling “I’m pinching you, Siri, you worthless git!” then I think that would be just cause for a divorce.

Or commitment, whichever comes first.

“Funny? Really?”

Well, he did post a couple of rather funny impressions of Jerkface over at LHB (assuming it’s the same guy, or rather, sockpuppet).

But I have to agree; mostly, he’s just boring (it’s quite unfortunate, really).

———————————————

I think MS already did what you describe (gestures and voice-comm)... with Kinect.

People (I mean Sony and Nintendo) claimed it was silly and crazy, just like people (I mean MS and Goog) claim Siri is silly and crazy.

Both times the “silly” and “crazy” tech sold like hotcakes.

I think it’s a matter of degree.

I can see issuing voice controls (in fact, Siri would probably be quite good in a TV context — limited machine learning required). I can see using an iPhone as a gesture-based remote.

What worries me is the sheer lunacy of trying to combine them. And you just know that somebody is going to do that.

Having said that, the average family has enough trouble with “who gets to hold the remote.” Can you imagine what it will be like with everybody shouting, not at Mummy and Daddy, but at a sodding smart-phone?

I am Spartacus! No, I am Spartacus!

This way lies madness. And it’d all be over which pathetic reality TV program you wanted to watch over your Saturday TV dinner…

@DrLoser
Yes, I do realize it (and he’s been out of conversation from the very beginning). As for downvoting – you’ve been proudly upvoting Adam sockpuppets’ idiocy many times. At least this one has SOME point in it.

@RCSE
Yes, I prefer “keyword crap” to semi-intelligent trial-and-error. With keywords I know what to expect (actual recognition quality is a topic for separate discussion). Siri on the other side is as annoying (confusing, disturbing, ... I just cannot find the right word) as any other uncanny valley is. Understanding limitations (and PROPER syntax) from the very beginning is less annoying then hitting all limitations one by one during your usual workflow.

@Linsuxoid:

I wouldn’t use the word “proudly.” And I wouldn’t use the qualifier “many times” (I believe it was twice). And from memory it wasn’t a sock-puppet, it was actually Adam. I can disagree with him all I like, I can even despise him, but he does occasionally make sense … within his own limited concept of “sense.”

I’m with TMR in principle. I’d rather not use censorship (or in this case crowd-sourced derision) unless absolutely necessary.

And I’m not saying there’s anything wrong with this post, other than that it is obnoxious and it is written by a troll masquerading as somebody else.

I’m just saying that both of those, taken together, are quite enough to make me down-vote it. And I also pointed out that this makes me feel like a tool.

Can we just leave it at that?

@RCSE
“People (I mean Sony and Nintendo) claimed it was silly and crazy, just like people (I mean MS and Goog) claim Siri is silly and crazy”

I’m an early Kinect adopter and I claim it’s silly and crazy (even though it’s fun SOMETIMES). One of those “uncanny valley” things in Kinect is its small lag (I didn’t measure it, but I think it’s somewhere in hundreds of milliseconds). It’s VERY annoying (or awkward, confusing or whatever) if something doesn’t react immediately – it makes you to retry your last action almost subconsciously.

In fact I think all motion sensing is silly, but compared to Wiimote and Move, Kinect is actually best one.

@DrLoser
No disagreement here. Post is obnoxious in its form, but it actually lights up thing that I generally agree with. And I didn’t mean to be rude or judging – just wanted to point out, that not all trolls are always downvoted.

@Linsuxoid:

Another interesting feature of Siri is, indeed, the syntax. I find it hard to believe that it will cope with, um, “conversational” English across a broad range of users. In fact, I find it hard to believe that it will cope with the range of responses you get in a typical blog.

Cue some nitwit standing in the middle of the street yelling “You’re a fsking Grammar Nazi!” at his phone.

Actually this could be quite entertaining.

Well Linsuxoid, I’ve seen your exact argument being made by UNIX weenies while they were bashing Windows and praising their precious CLI shells.

Siri is useful in exactly the same way GUIs are – the fuzziness (or rather, lack of formality) is a feature.

(BTW, you never know what talking to another person will be like either, despite of all the fabled “Human Intelligence” that people have, so I really don’t see any potential problem with Siri; yes, some commands may not work, big deal, classic UI is right there).

——————————————

Also, I often find myself annoyed with the necessity to manually intervene with something on the PC, whilst I’m away – if Macs came with Siri, and it could be used to meaningfully control programs, I’d already be buying one.

Yes, and Wikipedia Wisdom on Uncanny Valley states:

Reducing conflict and uncertainty by matching appearance, behaviour and ability In terms of performance, if a robot looks too appliance-like, people will expect little from it; if it looks too human, people will expect too much from it. A highly humanlike appearance leads to an expectation that certain behaviors will be present, such as humanlike motion dynamics. This likely operates at a sub-conscious level and may have a biological basis. Neuroscientists have noted “when the brain’s expectations are not met, the brain.. generates a “prediction error”... As human-like artificial agents become more commonplace, perhaps our perceptual systems will be re-tuned to accommodate these new social partners. Or perhaps, we will decide it is not a good idea to make [robots] so clearly in our image after all.”
—

Which is exactly my feeling about semi-intelligent intelligence. When you make something “human-like” (“Use your natural language, Luke”) and then fail to recognize it, you make user’s brain to itch. It’s that annoying feeling that makes Android look cheap because of frame-drops and stuttering animations even though actual speed is about the same.

> Well Linsuxoid, I’ve seen your exact argument being made by UNIX weenies while they were bashing Windows and praising their precious CLI shells.

Broken analogy is broken. I never told, that voice recognition doesn’t have future (on the contrary, I’ve told that it will be pretty useful once it becomes capable enough). And GUI never tried to pretend to be human (which is the source of most “uncanny valley”). More to say, when GUI actually TRIES to pretend to be real-life object, but fails to support same interaction model, it’s a BAD GUI (not as uncanny as pretending to be human though).

“Use your natural language, Luke.” Heh heh.

Shame the Greek one is the most boring of the Gospels.

I’m thinking I may have missed a reference right there.

@DrLoser
It’s another Luke and another Gospel
http://www.youtube.com/watch?v=2LTsvOHq3xc

@Linsuxoid:

Ahem.

I wonder whether Siri can detect irony?

It’s well known that “Any sufficiently advanced parody is indistinguishable from a genuine kook.”

That’s why smiles were Revolutionary Invented. Oh, it wasn’t Apple – forget that revolutionary part.

@Doctor
“I find it hard to believe that it will cope with, um, “conversational” English across a broad range of users.”

Siri is thought to be driven by Dragon NaturallySpeaking (in the cloud, not on the phone) – its understanding of English language is among the very best, if the articles on Ars are to be believed.

I don’t know if dragon tries to interpret the meaning as well, or if it turns the data to another service; but no matter who does it, over time, a significant information build up should enable Siri to become ever more capable of understanding what people mean and want.

—————————————-

@Linsuxoid
So Kinect is badly designed, eh, well, good to know; this doesn’t prove the idea is bad, it merely proves MS botched the implementation

You further bring Android as proof of how things don’t work – so if MS and Goog can’t make it work, Apple can’t either?

Again, your arguments mimic the attitudes of UNIX neckbeards who bashed MS and Windows – do you remember who won?

@RC:

I admire your naivete, but I’d have to suggest that you spend some time on the cutting edge of Machine Learning before you launch into such unsubstantiated assumptions.

I am of the opinion that an as-yet unconnected paradigm shift is required.

And in the mean time, I still expect to see loonies in the street loudly proclaiming to their phone that it is a Grammar Nazi.

I mean, if Linsuxoid cannot detect obvious irony, why should a phone do any better?

“So Kinect is badly designed, eh, well, good to know”

Pretty bad trick. I never told or even implied this. Fruit Ninja Kinect, for instance, has no observable lag at all. My conclusion is that it has to do with spare processing power: the less points you need to track and the simpler overall scene you have – the better you have Kinect software to do its work.

> You further bring Android as proof of how things don’t work – so if MS and Goog can’t make it work, Apple can’t either?

Um, what? I’ve provided several example of non delivering on promise interfaces. And Apple is among worst offenders here. At least the only thing Kinect seems to need is better processor, while Apple DESIGNS its UIs to be naturally-looking-unnaturally-feeling.

> Again, your arguments mimic the attitudes of UNIX neckbeards who bashed MS and Windows – do you remember who won?

Again, you use some broken analogies. UIs don’t pretend to be real objects anymore. It’s just too awkward.

So, even though some “collective UNIX neckbeard” lost, his argument was correct: it’s too confusing to suggest one interaction model by your look and actually deliver COMPLETELY different interaction.

“Broken analogy is broken. I never told, that voice recognition doesn’t have future…”

Oh I see, well let’s try this again shall we; you said:

“Yes, I prefer “keyword crap” to semi-intelligent trial-and-error. With keywords I know what to expect…”

You see, you’ve just made the CLI argument – “yes we can remember all those commands”, guess what, normal people do not want to remember any commands at all (even if they have the capacity to do so).

UIs that require knowledge of formal procedure are destined to fail (CLI was stillborn; those who weren’t blind knew this as soon as they saw any GUI).

For a UI to be usable, you must be either able to guess how to use it (yes, it has to be skeuomorphic to the max) or it already works exactly as you would expect it to – in the case of a verbal UI, that means ordinary conversation.

>“UIs don’t pretend to be real objects anymore.”

Yes they do, that is their whole point.

>“It’s just too awkward.”

No, it’s just natural.

>“So, even though some “collective UNIX neckbeard” lost, his argument was correct…”

If their argument was correct, UNIX CLI would have won, GUIs would have been rejected as inappropriate.

Now, how did it really happen, do you remember?

>“Pretty bad trick. I never told or even implied [that Kinect is badly designed].”

Do you listen to yourself(?), because previously you said:

>“One of those “uncanny valley” things in Kinect is its small lag (I didn’t measure it, but I think it’s somewhere in hundreds of milliseconds).”

That would make the “frame rate” of Kinect somewhere below 10fps; that’s a disaster to any user interface (for a video game UI it’s a bloody catastrophe).

And you are well aware of that, because you wrote next:

>“It’s VERY annoying (or awkward, confusing or whatever) if something doesn’t react immediately – it makes you to retry your last action almost subconsciously.”

So it is true, you didn’t imply, you said it outright – Kinect is badly designed (and I don’t care, if you can find a game, where the problem doesn’t manifest itself, especially if it’s Fruit Ninja).

And finally:

>“Um, what? I’ve provided several example of non delivering on promise interfaces. And Apple is among worst offenders here.”

Exactly as I said – you mimic the neckbeards to a comic degree (really, I must ask, are you typing this from a BASH terminal ;))

Right now, Apple is producing the best looking hardware and the best looking UIs.

But that is my opinion, what do most people think, I wonder…

Well, if Apple is to be believed, they hold more than 20% of the US market (new PCs sold, not the whole market) and given what I saw, I’m surprised it’s that low.

They claim to have majority market share of smartphones, since android users seem to be using their Durden infested bricks as dumb phones (frankly, from what I saw, it’s strange they use them at all, Android sucks its own balls).

And they have near complete dominance in the tablet PC market.

...

So, if you think Apple is wrong in their designs… well, most of the people strongly disagree, myself included.

>> “UIs don’t pretend to be real objects anymore.”
> Yes they do, that is their whole point.
No they don’t. There are many point in GUI (discoverability among main ones), but none of them is “pretend to be phisical object”. If’s not comforting – it’s confusing

TL;DR repeating your flawed analogies.

> Right now, Apple is producing the best looking hardware and the best looking UIs.
Nope it doesn’t. And never did. What Apple always excelled is making childish interfaces and taking others’ credit. See, I can do this too.

> So, if you think Apple is wrong in their designs… well, most of the people strongly disagree, myself included.
Oh yeah, how dare you are not to agree with 1.5 BILLION of Windows users? Please, don’t get “most of the people” on your side without their consent, OK?

So I read most of the comments, so excuse me if I repeat anything…

While TellMe may be keyword based, TellMe supports many more languages than Siri does (TellMe supports 26 [source: http://www.microsoft.com/en-us/Tellme/technology/default.aspx#tab=engines], Siri supports 3 [source: http://www.apple.com/iphone/features/siri-faq.html]).

Siri has had many complaints for the inability to properly interpret speech because of accent, which TellMe doesn’t have a problem with.

However, I disagree with the person who said they prefer keyword based commands over Siri’s more natural interpretation.

Why should I have to conform what I want done in a specific structure? That kind of takes the “fun” out of it, I just want to talk to it in any way I please. Seeing as Siri is cloud-based (just as Android’s is, and I believe TellMe is as well, to some extent), it evolves on a day-to-day basis and will get better and better at figuring out how to interpret a request.

So I think Microsoft has a better speech recognition engine, but Apple has a better interpreter (by that I mean figuring out what you want to do).

Siri works pretty well, while I haven’t used it, my friend said he played with it for quite awhile and it handled his requests as he intended.

Then there is “why does Apple get credit but Microsoft doesn’t?” Which is answered simply by: advertising.

If you don’t tell everyone you did it, why would they think YOU actually did it? If people saw ads better demonstrating WP7’s speech technology, and then they saw Apple’s Siri months after, people would recognize Microsoft had it first… Whether or not they are capable of the same thing, people need to remember consumers do not research things for themselves! That’s why people were gobbling up those “Macs don’t get viruses” BS.

@Ian
> However, I disagree with the person who said they prefer keyword based commands over Siri’s more natural interpretation.

Well, if even my “team-mates” don’t understand me, then it’s something with the way I’ve told it (or maybe you just didn’t read my explanations).

Again, I don’t use any of speech recognition services right now, but I still think REAL natural language processing will be great and useful and things. Right now it’s not on the level where it could be used with COMFORT (just as you speak to your maybe slightly retarded “personal assistant”).
And that’s exactly my problem with it. If it’s marketed to understand “natural language” it’s expected to, you know, understand it. Once it’ll fail to comprehend anything above 2-3 y.o. old kids’ phrases, you’ll resort to same keywords for your daily tasks, having almost the same bottomline but with unfulfilled expectations in the process.

Basically it’s not “keywords vs natural” it’s “predictable keywords vs unpredictable semi-natural”.

Technically, iPhone had Siri over a year ago, before Apple bought it.

God, my iPhone 4 feels so old and out of date!

“[...], but I still strongly disagree that it’s some sort of Revolution and bla-bla.”

It’s Apple. Their marketing is great.

“Seeing as Siri is cloud-based (just as Android’s is, and I believe TellMe is as well, to some extent)”

TellMe requires speech recognition over for web search (bug: if you disable it and it recognizes a “find” keyword a broken error message is shown here – HTC Titan) and text message dictation.

Paid M$ $hills.