Hacker News new | past | comments | ask | show | jobs | submit login
AirType (airtype.io)
231 points by pajju on June 6, 2014 | hide | past | web | favorite | 107 comments

If it works well, the demand is there, but we all know that, don't we?

The question is whether it works well in real use, and no landing page can show that. I can likewise put up an email subscription page for a $20k flying car.

Making this product work 50% of the time takes some serious work, but making it work the 99.5% of the time which would make it qualify as a keyboard replacement is something else entirely.

I wouldn't waste potential customers time with a landing page like that, and their video is downright insulting. Spend your time and resources making something that works instead.

Anyway, with all the buzz around hardware startups, I would expect more landing pages like this to crop up. People simply don't realize how difficult making good tangibles is, but they will find out soon enough.

Agree completely, just wanted to point out that 99.5% accuracy would mean 5 typos within the length of your comment. For any significant amount of typing it would need to be way more accurate. But then again, all this thing really is is a super-charged autocorrect, so they could probably fix those errors.

Or spawn a whole new class of damn you auto-corrects.

People who type a lot may be interested in moving away from the keyboard-and-screen model for ergonomic reasons. "This is how you don't fuck up your body while working" is a lot stronger selling point than "Wouldn't not carrying a keyboard with your tablet be nice?"

People who type a lot are likely to be working with some level of jargon (variable names for programmers, macros and cell codes for data analysts, etc.) rather than the pure prose autocorrect is designed for.

AFAIK, there aren't any good demos of autocorrecting keyboards for technical domains. That said, TextMate's autocomplete works pretty well most of the time. If you could seed the autospell dictionary with a combination of language-specific keywords and the variables that have already been defined in your project (ala TextMate), that could be pretty powerful for technical people trying to be productive with alternate input methods.

I am not sure of that. This might be fueled by a lot of negativity, but many people I know care 0 about ergonomic reasons, even if it fucks up their body while working. Chair too low, not configured, table not at the correct height, although easily changeable... I could go on.

Also, I am not sure how this thing makes anything better: you still have to do hand movements correctly and most people will need some kind of surface to orient their hands on. Same problem, without a keyboard.

> "This is how you don't fuck up your body while working"

As an aside, “Using Python to code by voice”[0] is quite an inspiring presentation on the topic.

[0] http://www.youtube.com/watch?v=8SkdfdXWYaI#t=542

That talk is why I switched to a standing desk and changed my keyboards.

> If it works well, the demand is there, but we all know that, don't we?

We do? I don't think I'd replace my keyboard with this even if it worked flawlessly. I like to have tactile feedback when I type.

You don't think you would, but there were a whole lot of people saying that about the keyboards on their Blackberries 5 years ago as well.

I've worked with a few of the mentors, as well as Capital Factory and Longhorn Startup, and it definitely seemed obvious to me that UT has some fantastically talented hardware engineers around there. Lynx Labs is one of the cooler hardwares I've seen in person, and it was built by students there.

Great, they can add this testimonial to their vacuous landing page. And I still won't believe anything till I see it.

Hmm, Dave Rosenthal suggested such a keyboard when the first accelerometers came out, basically put one on each finger tip and 'train' it by your typing on your regular keyboard, 'use' it by making some gesture and then typing. His concept though counted on 8 3DOF accelerometers per hand (5 on the fingers, one on either side of the palm and one on the wrist. The sketch showed something like fingerless biking gloves that had blobs and wires attached to your fingernails.

I'm not sure if it goes past the proposal stage however, the folks who were researching those sorts of issues were at UC Berkeley in the late 80's so that would be a good place for a literature search.

If this can be built and be as reliable as a keyboard (and I make mistakes/miss keys on my keyboard too) it could be really really useful. At some point, something like this and something like Glass will be a 'terminal' to the internet. The Myo folks have some interesting ideas in this space too, not clear if they get the fidelity they need through the upper arm though.

You've mentioned Rosenthal before, about the Myo: https://news.ycombinator.com/item?id=5280700

Was that conversation and/or those sketches ever published? I didn't find anything in a brief Google Scholar search. They'd be interesting to see.

I sent you email. Checked with Dave and he does not recall that anyone pursued it, but that isn't necessarily the definitive word.

I didn't watch the video, just the screenshots. I would seem that this device uses EEG like sensors to measure muscle activity and map it back to the activated fingers. Nearly the same as the thalmic device.

Regarding the "EEG like" comment, measuring the electrical activity of the muscles is termed "electromyography" or EMG

Thanks I had forgotten the term. http://openeeg.sourceforge.net/doc/links-biopsy.html#schemat...

My first exposure to EMG was in a very low cost eye tracker.

I think this is going in the wrong direction. Tactile feedback is so crucial for keyboards and this just completely removes it. In my opinion, technologies like the morphing touchscreen keyboard[1] are the future.

[1] http://tactustechnology.com/

Neither. Truth is we are a generation that grew up using keyboards. The one coming after us will consider normal to use an onscreen keyboard or voice recognition. This is a problem that will solve itself.

No. Humans depend on feedback loops to coordinate their fine motor movements. Otherwise absolute positioning accuracy is bad. Even an on-screen keyboard provides such feedback. Typing in the air doesn't.

So type on the table?

Seriously, wouldn't that be the same amount of feedback as a touchscreen keyboard? Flat, and you get the wrong letter when you miss?

No, a table gives no feedback whatsoever in the manner that he's referring to at least. When you hit a key on a keyboard the key sinks into the board, you have a positive 'strike'. If you miss the key slightly then it feels different and you'll know before the wrong letter appears - this is what is meant by a feedback loop.

That's off-topic. We're talking about touchscreen keyboards, as is very clear in my post, and we won't pretend their keys sink.

With an on-screen keyboard, you get visual feedback seeing your fingers land relative to the keys, even if you don't give tactile feedback. With this, all you see is whether you got the letter right.

I don't understand where you and I disagree. That "no" seems confusing.

> This is a problem that will solve itself.

> Even an on-screen keyboard provides such feedback. Typing in the air doesn't.

Probably there.

Voice is a separate issue, but this has the same problem as a touchscreen. We are very sensitive to touch. It's quite possible that with the proliferation of touchscreens, children may find increasingly less tactile feedback in the world around them (due to various other reasons - less play time, sterilized playgrounds, etc) and we could end up with a generation of adults lacking the same tactile ability we have today.

So I don't think the problem solves itself, it just compounds.

Much like our loss of movement ability, it seems to become less relevant, until you notice that we've lost an essential part of what makes us human.

Its about time computers started fitting humans and not the other way round.

> Its about time computers started fitting humans and not the other way round.

This is a really interesting argument and has got me thinking. I have to say that I agree with your sentiment. The problem is that I think we've entered into a kind of feedback loop; a compounding problem as you described it.

Technology and intelligence have allowed us to rapidly accelerate our fitness while simultaneously and subtly forcing us to become increasingly dependent upon them for continued advancement. In short, tech informs the ways in which we progress and, in the interest of further progress, our new (less human?) norms inform the progression of tech.

I think there are examples of this all over the place: infants expressing confusion at a screen that is not touch sensitive, memes and texting idioms seeping into spoken conversation, or people asking their device to call someone and intentionally mispronouncing the name the same way the device does so that it "understands."

But I also disagree that tech can start fitting humans in the sense you describe, because I simply don't think that's possible anymore.

There's a post on the front page right now discussing self-experimentation with black widow bites, a comment from which [0] strikes me as relevant to the argument I'm about to make. To quote, "[humans]...are pretty weak creatures all things considered." A lot of cutting edge tech today is focused on AR and mobility. That's because, as the quoted comment notes, humans are weak; a weakness which also extends to our senses. Touch seems so fundamentally human, yet a computerized sensor can tell us so much more than a fingertip about a surface. It sounds terrible when I write it, but there it is: information is powerful (addicting?).

It's impossible for us to conceive the kinds of situations humanity will face in the future, but after a certain point the limitations of the human body will have to be addressed. Right now that means computers.

So, I suppose what I'm saying is that, imho, computers actually are fitting "humans," it's just that they are fitting what we have, will and must continue to become rather than what we once were. And that's probably a good bad thing (rather than bad bad) in the long run.

[0] https://news.ycombinator.com/item?id=7860997

Computers are fitting humans, thats why they are now a thin slate you can easily carry around.

And there is a feedback when you touch a glass surface, it's just it doesn't give you a non-visual clue that you have pressed a letter and that you have grown accustomed to.

I don't think that's a dealbreaker for the generation that is growing with tablets. I certainly don't see my teeneage cousins or nephews/nieces worried at all about using keyboards. If they have them around they might use them but they type fast on the screens as well. Most of them no longer have a laptop or desktop anyway. I've seen them writing whole essays on the phone while lying on the bed.

Talking about lost feedback, think about what a huge step back was for our grandfathers generation to abandon handwriting and switch to pressing keys on a typewriter. I'm pretty sure we could have argue back then about how little feedback you get from a key vs. the flow of the hand and the friction of the paper when you write. And somehow we adjusted so well that most of us are probably incapable of writing a handwritten letter nowadays. We are not "less human" because of this.

We will survive.

And note I'm not trying to say that this is the new status quo and we should all accept it and move along. I'm sure we will see a lot of different approaches to solve the issue in the future. What I'm saying is that the pool of users these solutions speak to is diminishing. We see this ideas and think that they'll appeal to everybody who uses a tablet. Most likely they'll be only used by a niche market. Nothing wrong with that but definitely something to take into account when trying to launch a company around them.

What about productivity? A professional programmer, blogger, etc. needs to create huge amounts of content.

Trouch-screen devices are good for content consumption but absolutely unfit compared to computer for content creation.

People are creating high quality keyboards[1] because there is an actual market :-)

[1] http://codekeyboards.com/

That's where machine learning plays a role. It learns what movement corresponds to a particular letter. If that movement changes over time, it learns that too.

I really want a portable keyboard that I can type on quickly while walking around, but I don't think these guys have anything like a product. This looks like an empty web page to measure interest.

That's what I'm hunting for as well; have been for the past few days. Something like BlueTwiddler ( http://hewner.com/programming/ ) is pretty much perfect for me, but it was never commercialized. I'm thinking about picking up http://theperegrine.com/ , adding Bluetooth support, and turning it into essentially a chording keyboard for wearable computers. I need something that will let me type on Glass, the Epson Moverio, and my phone with ease. I'm hoping in the next year that I'll be able to ditch non-wearable computers, from an interface perspective (I'll still have them for heavy lifting, of course).

The Peregrine looks to be the most promising platform for this kind of input. The only limitation I've seen is you need the finger dexterity of a guitar player to quickly and accurately type that way. Then again, the only other solution I've seen is the four button spiffchorder[0] and a trackball in the other hand. I don't like those only because you can't do anything else with your hands while you're using them.

[0] http://chorder.cs.vassar.edu/spiffchorder/forside

This is basically a long-term dream of mine.

I know these guys (same university). I remember them creating the prototype for this about a year ago at a hackathon, so I think they have a product.

I'd think if they had a product they'd actually demonstrate it in the video, instead of showing some fakery and then the inert product.

From what it looks like, the founders didn't submit it to HN. For all we know, this page was thrown together as somewhere they can direct people who ask them what they are working on, not somewhere they are actively trying to drum up interest with (if they had nothing beyond what the page shows, my guess is they would have gone with a Kickstarter, rather than just an email capture).

A few months ago, I saw a (seemingly) working model that used wires on the fingertips and rotational potentiometers to detect finger flexion.

I wouldn't be surprised if the same algorithms they developed can be adapted to myoelectric sensors (if they haven't done so already).

Check out the Corderoy project from Bengler: http://bengler.no/chorder

It's still just prototypes, but sounds like someone is working on just what you ask about.

I have a hard time trusting a product when it's promo video shows someone mashing random keys pretending to type.

I think he's hitting the right keys though

they weren't. I noticed it right away and it annoyed me.

I wish the demo video had actually been a demo. They typed nonsense on the real keyboard and then flailed similarly with the AirType.

Honestly. For a moment I thought this was another [1] viral in-universe ad for that HBO show Silicon Valley. Then I realized they are serious...

[1] http://www.piedpiper.com/

Amazing ad by HBO for sure. The only other similar ad I remember was for The Office. http://www.dundermifflin.com/

Alas, this is a fool's errand.

The keyboard is a broken interface, but this isn't an improvement.

Suppose I am the user identified in the video. I have a tablet—maybe not a Microsoft Surface, but a tablet nonetheless—and I currently use a slimline keyboard to type.

You hand me these. First thoughts: what are they? What do I do with them? Great products are those that someone can be presented with for the first time, and in a matter of seconds figure out what it does, how to use it, etc. Think kids reacting to technology [1]

So for argument's sake I read the manual, or you give me a walkthrough. Great, how do I start using them? Oh, they have to 'learn' my writing style? Is my first impression going to be a string of largely inaccurate input?

How am I going to 'teach' this system? Through some proprietary software, which I may have problems downloading, and which may not even work on my device? How are people going to demonstrate these in stores; how am I going to proselytize to my coworkers? "Just try these out; wait, no. Yeah, don't worry about that, they have to learn your typing style. Yep, they'll do that for a while... no, but they're really great. Honestly... honestly."

This is already sounding like a lot of effort, but you've assured me there's gold at the end of this rainbow. Surely the cost-benefit analysis will check out.

So I've got them working. Are they accurate enough? Let's assume they are. This isn't too bad actually. I'm going to hop on Twitter and tell people all about this. Hashtag AirTypeRocks! Hashtag... hashtag... where's my hash key again? I can't seem to find it on this wooden bench.

Okay, that's a downstream problem. For now, I'm going to pack these up and head home. But where do I put them? They're an awkward shape. They look pretty easy to break.

What happens when I lose one of them? Is the entire system rendered useless? They look expensive to replace.

Are these yet more devices I'm going to have to charge? I'm already fed up of charging my laptop, phone, tablet, etc.

Are they going to protect the screen of my tablet device in transit? Looks like the opposite. I think I'll hang on to the keyboard-cum-screen protector for now...

[1] http://www.youtube.com/watch?v=PF7EpEnglgk

Looks to me like this type of device is the beginning of disrupting typing. Once you have a way to detect finger movement with something small, lightweight, unobtrusive (I'm making assumptions about AirType based on my own desires) and programmable, we can create the interfaces we want.

Or is it ? At least this one seems to be fighting the last battle.

It targets people who can blind touch, which may not be a growing population in 10 or 20 years. Computer will be more ubiquitous than now, but there will be a whole generation that would be raised on touch devices first and be very proficient with on screen keyboards or benefit from direct writing recognition (these two are both more discoverable than this technology), or more minorly speach recognition.

For a generation not trained on physical keyboards first, I'm not sure this kind of alternative input method will be better or faster that what they were already using as primary input device, and they'll want something wildly better if they have to switch to something else. Like a full size physical keyboard, if they're going to type for a long extent of time.

edit: spelling

Sadly, without referencing any keyboards, I not only know it's shift+3, but also that on iOS, it's hidden behind the 5. (Why yes, I've a Das Ultimate, how'd you guess?)

I doubt the ability to get a functional product to market.

The "Team" section lists only 2 actual doers and multiple product people. Never a good sign in a startup.

If I'm following their logic correctly: Touchscreen keyboards are inferior to real keyboards, so some tablet users buy add-on keyboards for their tablets. These are also inferior to real keyboards. Therefore, AirType is better, because it's different.

Seriously though, I don't see this catching on. Keyboard enthusiasts obsess over the tactical feel of their mechanical keyboards for a reason.

So it appears that you have to keep your fingers elevated off of the surface, and then strike down similar to a touch screen?

The best part about a mechanical keyboard is that you can rest your hand on the keys. I've actually never tried a MS Surface touch cover, but something like that without actual segmented keys might be best. Sort of like a touch cover and Fleksy hybrid.

The demo video shows a Surface tablet that has zero visual response to the hardware keyboard and their demo product. Who thought this was a good idea?

I wonder how accurate it will be.. For general typing, there is probably some built-in autocorrect, but, if you wanted to code, I wonder how that would work.. Exciting anyway.

Accuracy aside. What do you you if you need to look at your keyboard to find an unfamiliar symbol. I can touch-type fairly fast but my error rate goes up dramatically when typing non-alphabet keys.

What would be neat is if you could bring up the tablet keyboard through a gesture, like clapping your palms together, select the unfamiliar symbol, gesture to remove the keyboard then continue.

Although I like the idea of something like this, I can't imagine myself using it for a long period of time.

Could this work as a piano keyboard? That would be way more useful for me since I usually have a physical keyboard (laptop) but I can't really carry around a midi one.

I actually think this is pretty brilliant, have the hand parts hook up to some ear buds and plug them into your ears, any place your at you can practice piano.

Piano hand movements would most likely also be easier for the software to reliably track, and some people might like the gimmickry of "air keyboards" for live performance

So the solution is to invent something worse than detachables and on-screen keyboards?

Not saying AirType is bad...yet. But I didn't see it working either.

There's nothing I can say about this that Bret Victor hasn't already said better:


TL;DR: Give me some tangibility in my interfaces! (You really should read that rant though)

This is always the first thing that pops into my mind when talking about touch interfaces.

There is so much and so deep harm in taking away the tactile feedback.

And then I go to shake somebodies hand/ wave to somebody / hold a sandwich and I have this thing stuck to my hand.

Edit: Might be different if it wasn't so hard, big (across my whole hand), and plasticy. Gloves and rings don't seem to cause much trouble during day to day activities.

Let's assume that this product works faultlessly.

What proportion of keyboard users are touch typists? I'm not.

Is it just me or the intro video wasn't really clear how it works. Am I going to just move my hands on nothing and expect this thing to write? Can someone clarify how this works?

I've always wanted to try a laser keyboard like this one http://www.celluon.com/shop_epic.php but not to the point of buying one...

Seems to me like it is a more user friendly approach no physical feedback and no visual sounds like AirType is a power user tool. And my Das keyboard lover friends who have blank layout would hate the lack of physical feedback.

I received one of these as a gift once. Played around with it for 2 minutes, then never used it again.

The main issue I had with it was that you had to 'hunt and peck' in order for it to register. Merely sliding your fingers around like on a standard keyboard would just register errant keystrokes.

Also, resting your palms had the same keyboard-mashing effect. It was cool to show off to others, but totally impractical for real use.

I've come close to buying one but I can't help but wonder how painful it's going to be on my fingers. My keyboard keys have nice padded, spring like qualities. My desk however is waiting to draw blood from my finger tips after a days coding...

I don't see myself ever replacing a real keyboard for this, at least not for long typing sessions (being tactile is the most appreciated aspect of a keyboard for me).

But there is absolutely a huge demand out there for this. Not to mention that other applications for it may appear (other than just replacing traditional QWERTY keyboards).

But in the end the question is the same: how good it is?

Anyone got a video showing it working (other than just the concept)?

What problem does this solve exactly? I want to think this is really cool, but I'm having trouble convincing myself why people would buy this.

Ideally it would make writing on a tablet as good as writing on a normal keyboard.

So for this product to succeed we're also arguing that tablets can replace laptops? I own both for totally different reasons, and I don't think that will ever change.

Or this is just a "nice to have" accessory for tablets?

My point is this isn't really something the masses will adopt, and is instead a very luxury item for wealthy nerds (myself included). Totally fine if that's the case, I just want to understand the target market.

The biggest problem for the average user, he can't see the keys. So while techies on this forum can pretty much type without looking at the keyboard, your average user still types with only two fingers, looking at the keyboard.

Also it has the same drawback as tablet keyboards, i.e. no positive tactile feedback when pressing a key.

But let's see, maybe this does replace keyboards altogether.

I don't know... This won't be as accurate as we want it. No matter what they use. I tried making a similar prototype using motion sensors, muscle sensors, neural network. But the thing is, a keyboard is just alot of possibilities for anything like that to predict. Plus, no feedback means our motion will be whole lot different than on normal keyboard.

I really hope they can handle chorded keystrokes if they want folks like programmers to use it.

Also, while tablets are an obvious starting position, perhaps consoles would be another good arena - having to lean over to type on that wireless keyboard in order to chat or enter detailed text seems more intrusive than just typing it out using AT.

I think there are some keys that I couldn't reliably hit without a quick glance at what I'm aiming for. It's a reliable phenomenon that expert typists can't tell you where many keys are, so I'm curious how I'd fare when I need to hit the ^ or & symbols and can't look down.

Add in some speech recognition and it would be an ideal system.

Cool and I would like to try it. But doubt I would sit down at a coffee shop and put my bionic hands on in front of everyone.

I would have to be better than typing on a keyboard.

Video was weak. Right when they show the product it cuts out.

Would be good if you were able to touch your finger to the surface and it becomes a trackpad as well.

This keyboard will have the same problem that iPad/tablet keyboards have - lack of a mechanical response, and without that, you'd find it just as painful to use AirType as you would a tablet. IMO, sure, but you wait for it ;)

Isn't this basically like the Magic Cube laser keyboard? http://www.celluon.com/products.php

I wonder if the same thing could be done with a Leap Motion controller? Seems to be a more versatile solution.

meta: its interesting that all of the team members are described with a short informal first name and a tagline while all the mentors are described with full names only.

it certainly shows the disparity between the generations and their different expectations in the business world.

I'll believe it when I try it.

I wonder if you could do something similar with a regular camera watching your fingers.

I have done that using Kinect. It's difficult to achieve a high level of accuracy, though.

Showing a fake demo, it's not helping their landing page.

I would love to see the real product.

I personally like the tactile feeling so much that I bought a relatively expensive mechanical cherry-switch keyboard. There is no way replacing that feeling on your fingers.

it is Senseboard from 2001 I think they are the same people reviving the old product which somehow failed at that time.

Maybe I can use this to code while driving!

Looks great. Need to see a real demo.

i would pay so much money to be able to walk around and code

Would you buy an app that allows you to program on your phone (that is sensitive to the limitations of the on-screen keyboard)?

Keep a voip chat open to a PA overseas!

Ben’s photo is 1.2MB in size.

They fixed it. You are welcome.

The terrific song in the video is Boe Zaah by Mac Demarco.

It's unfortunate to see it in a product video unattributed (or at all — though I know his music is licensed elsewhere).

Do you know that they didn't buy the rights?

No, good point. I'd edit my comment to just point out the artist, but cannot.

Wow, really exciting!

And I know I am (not) being paranoid, but imagine how excited NSAs of the world all will be! A new kind of keyboard, that has capabilities to understand and predict what the user will write? This just looks like something Facebook could also be interested into (if you think at their "I keep what you wrote and deleted before even publishing" attitude).

This kind of tool really needs to be open source and open hardware. Otherwise they better stay far away from me.

This kind of cynicism is getting tiring. Being cautious about your personal information is good, but how exactly is this any worse than a bluetooth keyboard? OS X and iOS already have predictive typing with regular keyboards. There is no need for FUD.

Predictive typing for single words is usually based on a locally stored dictionary, and for sentences it can be done with a model like prediction by partial matching. The device functionality is similar to a system used for speech recognition, in that better accuracy would only be gained by sending motion and text data to the cloud at the very least in the training stage and then aggregate and analyse data using a prediction model.

However, no-one seems to care that most voice recognition tools like Siri nowadays send and store data on servers, and nearly all those aren't OSS.

> However, no-one seems to care that most voice recognition tools like Siri nowadays send and store data on servers, and nearly all those aren't OSS.

I think one of the reasons Jasper [1] was so well received was that it was offline.

[1] http://jasperproject.github.io/

Thanks for pointing out Jasper: great!

This isn't cynicism. It is the new reality we live in. We now have to look at every piece of technology we buy and do a risk analysis to see how it could leak our data. It isn't just the NSA. It is also hackers who are after our data. Having a breach, especially by the latter, is a painful mistake that can ruin your credit and turn your life upside down for decades to come.

Why risk it, open one more door to a security violation, so you can type in the air. No thanks.

Registration is open for Startup School 2019. Classes start July 22nd.

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