The Analog Keyboard Project: Text Input for Small Devices | Hacker News Reader

The Analog Keyboard Project: Text Input for Small Devices

research.microsoft.com

47 pointsmafuyu26 comments

Minuum seems more interesting than this: http://minuum.com/model-your-users-algorithms-behind-the-min...

Missing it completely. Don't use the existing alphabet and characters. Create a new one that is as simplified as possible. Sure, we'll have a steep learning curve, but once you learn it...

I agree. Graffiti writing on the Palm series of computers should be seen as the inspiration on this. It was an alphabet based on the Latin alphabet, albeit with a few tweaks so that every letter could be achieved with a single stroke. Honestly it didn't even have that big of a learning curve, especially since a lot of Palm cases had the Graffiti writing letters on them and the quick reference was always a tap away.

In my experience, it had a very high degree of accuracy. Once I knew the alphabet, I could write on Palm devices as fast as I could on pen and paper. I wonder why there aren't many implementations like Graffiti for stylus-based text input on a lot of computers these days? They're always striving to figure out these complex multi-stroke characters, which often tend to be their downfall on accuracy.

Graffiti is in fact already ported to Android: https://play.google.com/store/apps/details?id=com.access_com...

This was a wonderful input method -- there is no need to reinvent this particular wheel.

or maybe a shorthand system can be adapted for mobile http://en.wikipedia.org/wiki/Shorthand

I think trying to input text on a wearable or other small device using a touchscreen is a fundamentally flawed concept. There are ways to do certain things well, but this just seems like it would become boringly slow terribly fast.

With wearables such as watches, we need to think of different ways to input text and interact with content on the screen and with other people using similar devices. The future of text input is not swiping at letters on a screen because it's a cumbersome hack that misses the point of wearables altogether.

Wearables are for fundamentally different types of communication than phones are. Obviously I'll get flamed for this but I think Apple is the only company thus far to have realized that. People are not going to want to be manually texting others from their watch. It's unnatural and impractical. Watches are going to be for informal communication, like quickly sending pictures, emoticons, or heuristically generated predictive text snippets.

This is cool and all but they go into this project with the assumption that manual text input on small devices is important, which to me at least, it's pretty clearly not.

Manual text input per se may not be the most important interaction with wearables. But getting this to work may be an important first step toward building, for example, a gesture-based method to interact in general with wearable devices that are resource-constraint (eg, small screen, limited voice/camera capabilities, etc).

Seems like you have some thoughts on other methods to input text when you say "we need to think of different ways to input text".

This problem was solved a couple hundred years ago. Single button input can be rapidly and efficiently done using morse code. With two buttons, one can be dit and the other dah.

One extra nice thing about morse is it isn't necessary to look at what you're inputting.

This is not the same problem. The problem here is to come up with an input method that anyone can use, because anyone can have a watch. With morse only some people had access to equipment and those people spent time training. Everyone else just used handwriting to give the morse engineer a message to translate. Basically a large percentage of the work that computers are used for is to automate the encoding and decoding of messages - automate the morse engineers job.

Note that with computers, people don't have to decode morse. It would only be for input.

I find it hard to believe that teens, for example, can't very quickly learn to input morse. After all, they learned that wretched scheme used to input text on a phone pad.

I wear watches on my left wrist, but I'm also left handed!

Good luck deciphering my right-handed writing. Although as Steve would say, I'm wearing it wrong.

I'm still hoping for great voice recognition and gestures.

Ultrasonic gestures?

http://venturebeat.com/2014/10/06/coming-next-year-ultrasoni...

Looks like the old Palm input, back in the 90s.

yup. wonder who has those patents now?

Aren't these patents only granted for about 17 years? If so, we're close to their end-of-life!

Xerox apparently had the patent for the original Palm Graffiti... https://en.wikipedia.org/wiki/Graffiti_(Palm_OS)#Lawsuit

I KNEW my 1990s muscle memory for Graffiti for Newton OS would come in handy again someday!

Btw, is there any FOSS software for predictive text input on desktop linux? (Being on a desktop level, available to all apps.)

I'll stick to my Casio's trusty numpad.

That form factor seems like the right size to re-implement T9 :)

Palm's Graffiti?

They use Google in the demo from Microsoft Research, that's a little embarrassing :)

Embarrassing? Nah, pragmatic!

That's a lesser disloyalty, considering the whole thing is on an android device

I don't think its embarrassing. Their are often competitor products seen in their latest products advertising. For example their latest bluetooth keyboard ad is full of iPads. Impossible to think of Apple showing Lumia phones in their ads.

It's not that embarrassing if they get it to work well, write several patents on it, and then get Google, Samsung, et al to pay royalties for licensing them. On the other hand, MS could build their own wearable devices and use the technology.

Reply on news.ycombinator.com