I don't think you're giving the future enough credit. You discredit voice control, but base your opinion solely on their present capabilities.
> the voice robot "press one for x, press two for y" is a very good subject for jokes
Interactive voice systems are among the most primitive voice control system possible. The simplest require you "talk" using generated tones. The more complicated accept a real voice but are hardly conversational; they present a rigid and invisible menu of options. At least they present options, some text-based games of lore required you to read the minds of their game designers.
But they won't always be this way. There's no reason to believe computers won't someday become as good at conversing as humans are. Surely you've noticed talking with a human over the phone is much faster than talking over IM, and eventually computers will also hit that point. Imagine it: you pick up your phone and call AT&T, the system immediately picks up and asks "Thanks for calling customer support, may I have your PIN please?", followed by "Hello gbog, and why are you calling today?". You respond with a simple "I'd like to pay my phone bill" and the conversation continues.
> With touchscreens in everyone pocket, I bet these usability nightmares will be replaced with apps or websites.
What happens when you don't have a touchscreen in your pocket? Touchscreens are magical and the current rage, but there's no reason to believe they are the last word in HCI. What about Google glass? Implants? Some other form-factor we haven't yet imagined?
Voice recognition will become faster than using apps, it's already happening! Telling Siri to "set an alarm for 9:30 tomorrow" is much faster than unlocking your phone, finding the right app, waiting for it to open, then messing with those damned spinners.
> So, in all, voice control, Siri and Google Now seem to be technological dead-ends
Siri is a vision of the future. For the first time (if you'll allow some narrative license) you can have something resembling a conversation with your computer. You talk to it like you would anybody else, and it tells you what you'd like to know. Speech is the fastest method we have of dumping information from our brains, it's incredibly natural, and it's nearly universal; I wouldn't be surprised if it eventually became our primary method of controlling computers.
I don't want to make this comment too long, but a similar argument applies to Page's vision of predicting your needs. You are focusing too much on the present and not enough on the futures potential.
If tomorrow you have some unusual reason to wake up at a different time, I agree that voice control is simpler, but this one is the kind of complex one-time configuration I was talking about above. You should not need to give the same order every evening, right? And even in this case, if your gf/bf just fell asleep next to you, would you not prefer some silent way to give this order?
I agree too that touchscreen is not the last word in HCI. But I still don't see a future of people talking to robots. More precisely, I think that robots will always be machines, and that the main interaction channel with machines will not be sound waves carrying human language. I'll try to explain why.
The most useful and powerful tools and machine we have today, from chopsticks to pen and paper to cars to computers, are all using hands and fingers. Human being are as much defined by their language capacity than by their crafting-by-hand capacity. When sound channel is used, for example for sherperding and hunting, it is because it can reach far away, not because it is particularly convenient, and it is not using normal human language anyway.
Most powerful tools are powerful because they have near infinite precision and expressivity. Think a Chinese brush, or a violin: human language is extremely rough compared to the range of actions allowed by these. Try to voice-control a pen to draw a single smiley face, I would be surprised if we would recognise the thing. In fact, I think next and next-next generations of electronic devices, be it phones or glasses or implants, will go closer to the tight coupling and immediate feedback we have with pens, brushes, pianos, and even hammers or game controllers.
Many (if not most) of the interesting interactions we can have with machines and tools are not worded and cannot be expressed in current human language. This is a bit out of sight these days, but not everything need to be words or text. Reading famous chefs recipes will never help you cook food if you do not have the "hand" (and taste) for it. The HN crowd is a bit biased, with so many bloggers and blog-readers, but many people do not consume as much text as we do here.
The many attemps to use human natural language to describe the behavior of systems have all failed hard, and we have to fallback on programming languages, which, from linguistic perspective, are abominable caricatures of human languages. This is telling. In order to control properly a computer, we need to talk in the computer's language. Not necessarily because it is not clever enough to understand human language, but because human language is not the right tool for the task. Human language is very efficient in communicating with other human beings, but not so with machines, and useful robots will be machines.
The future will them me right or wrong, I hope to grow old enough to see it.
But allow me to imagine the intelligent glasses UI. It can display anything on the screen. Very good, so it means that, except when alseep, the audio channel is not needed not receiving feedback from the glasses. Talking to the thing using human language is a possibility, but only one in many other possiblilities. Imagine there is a very well thought "UI" that is sensing the taps of your fingers on your forearm, or your leg, or whatever part of your body is easily accessed. Imagine you can give "input" to your glasses by drawing shapes on your blue jeans, or tapping little rythmic commands, "tap-tap-tap" is cancel, "slide-left" is next, "draw a round" is repeat, etc. I think this could be worth trying.
> the voice robot "press one for x, press two for y" is a very good subject for jokes
Interactive voice systems are among the most primitive voice control system possible. The simplest require you "talk" using generated tones. The more complicated accept a real voice but are hardly conversational; they present a rigid and invisible menu of options. At least they present options, some text-based games of lore required you to read the minds of their game designers.
But they won't always be this way. There's no reason to believe computers won't someday become as good at conversing as humans are. Surely you've noticed talking with a human over the phone is much faster than talking over IM, and eventually computers will also hit that point. Imagine it: you pick up your phone and call AT&T, the system immediately picks up and asks "Thanks for calling customer support, may I have your PIN please?", followed by "Hello gbog, and why are you calling today?". You respond with a simple "I'd like to pay my phone bill" and the conversation continues.
> With touchscreens in everyone pocket, I bet these usability nightmares will be replaced with apps or websites.
What happens when you don't have a touchscreen in your pocket? Touchscreens are magical and the current rage, but there's no reason to believe they are the last word in HCI. What about Google glass? Implants? Some other form-factor we haven't yet imagined?
Voice recognition will become faster than using apps, it's already happening! Telling Siri to "set an alarm for 9:30 tomorrow" is much faster than unlocking your phone, finding the right app, waiting for it to open, then messing with those damned spinners.
> So, in all, voice control, Siri and Google Now seem to be technological dead-ends
Siri is a vision of the future. For the first time (if you'll allow some narrative license) you can have something resembling a conversation with your computer. You talk to it like you would anybody else, and it tells you what you'd like to know. Speech is the fastest method we have of dumping information from our brains, it's incredibly natural, and it's nearly universal; I wouldn't be surprised if it eventually became our primary method of controlling computers.
I don't want to make this comment too long, but a similar argument applies to Page's vision of predicting your needs. You are focusing too much on the present and not enough on the futures potential.