Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

[P]
10 big fingers

By adrien in MLP
Mon Jun 04, 2001 at 05:06:01 PM EST
Tags: Hardware (all tags)
Hardware

10 big fingers.
26 letters (plus numbers and punctuation).
Devices that fit in the palm of your hand.

See the problem?

Ten Big Fingers is an article at Point and Grunt about some of the current offerings for text input on small devices.


Sponsors

Voxel dot net
o Managed Hosting
o VoxCAST Content Delivery
o Raw Infrastructure

Login

Related Links
o Ten Big Fingers
o Point and Grunt
o Also by adrien


Display: Sort:
10 big fingers | 44 comments (37 topical, 7 editorial, 0 hidden)
monocular inputs (none / 0) (#1)
by alprazolam on Wed May 30, 2001 at 04:55:33 PM EST

binocular inputs that scan the eye for movement and positioning. when your eye rests for a certain period it picks the letter you're looking at.

borg factor (4.00 / 1) (#3)
by adrien on Wed May 30, 2001 at 04:58:17 PM EST

Even the Borg would think I look like a dork wearing something like that :-)

BTW, where can I get one?


-adrien
[ Parent ]
borg chicks (none / 0) (#26)
by djabji on Thu May 31, 2001 at 05:19:58 AM EST

Hey, there is no shame in wanting to pick up borg chicks.

I hear that the 'asimilation' process can be quite fun if you find a borg with large 'implants'.

[ Parent ]
40,000 letters (4.33 / 3) (#4)
by delmoi on Wed May 30, 2001 at 04:59:51 PM EST

But the japanese seem to be doing just fine with their I-mode phones.
--
"'argumentation' is not a word, idiot." -- thelizman
got any references? (none / 0) (#10)
by adrien on Wed May 30, 2001 at 05:16:56 PM EST

do you have any information on how they deal with this problem?

I am very curious.


-adrien
[ Parent ]
Well I don't know about phones specifically (none / 0) (#12)
by brion on Wed May 30, 2001 at 05:48:52 PM EST

but on computers generally, you input Japanese text phonetically in Latin letters, which are automatically converted into kana (Japanese syllablic characters), which the computer's input method tries to convert to kanji (the Chinese-derived ideographic characters) a sentence at a time. If it picks a wrong character (homophones with different meanings, hence different characters, are common), you scroll through a little box to find the right one. Fun, really. :)



Chu vi parolas Vikipedion?
[ Parent ]
Unless you have Japanese computer (none / 0) (#14)
by Skippy on Wed May 30, 2001 at 06:41:13 PM EST

They sometimes has hirigana on the keyboard (the Japanese phonetic "alphabet") which can save you typing two Latin characters by typing one hirigana. For example you would type a key with the hirigana which stands for ku instead of the latin characters k and u. There are issues with that method of input too as there are, IIRC, something like 56 hirigana plus pronunciation marks (which do things like unvoice mora). Its been a while since I studied Japanese. I would imagine that the hirigana input method might even be faster once you got used to it.

# I am now finished talking out my ass about things that I am not qualified to discuss. #
[ Parent ]
Yeah, that would save a small step (none / 0) (#19)
by brion on Wed May 30, 2001 at 09:26:33 PM EST

I think you could get away with about 49 for kana (46 for basic kana, 1 for voice mark, 1 for small vowels, 1 to switch hiragana/katakana) plus punctuation, control keys, etc. I've never actually seen one so I don't know how they're really arranged. (Real Nihonjin are welcome to explain how they really work!)

But in any case 49+keys are even harder to fit on a cell phone, so I wouldn't imagine a kana keyboard as the ideal handheld input method...



Chu vi parolas Vikipedion?
[ Parent ]
Japanese on cell phones... (4.33 / 3) (#20)
by driptray on Wed May 30, 2001 at 10:40:00 PM EST

...is quicker and easier than English. This because, as mentioned above, typing in hiragana requires half the keystrokes of English. Even though my Japanese is really poor, I often type things in Japanese on my phone rather than in English cos its so much quicker.

To explain how it works, the "2" key, when in English mode, cycles through A,B,C,a,b,c, and when in Japanese mode, cycles through ka,ki,ku,ke,ko. So a word like "sumo" is four characters in English, but only two in Japanese ("su" and "mo").

I work with a bunch of Japanese translators, and I'm amazed to see that when typing Japanese on a computer keyboard they use the English alphabet rather than hiragana. It's twice the keystrokes! But that way they don't have to learn two touch typing patterns, one each for Japanese and English.


--
We brought the disasters. The alcohol. We committed the murders. - Paul Keating
[ Parent ]
Ah-ha... Thanks! (none / 0) (#21)
by brion on Thu May 31, 2001 at 12:15:48 AM EST

Come to think of it, that makes a lot of sense - at least kana can be put into some kind of reasonable order per key!

Chu vi parolas Vikipedion?
[ Parent ]
aha... (none / 0) (#24)
by adrien on Thu May 31, 2001 at 04:40:07 AM EST

is there a localized T9 for that?

In any case, I would not be surprised if the same principles could be applied -- langage is language, it's redudant, it repeats itself a lot, and there are rules for forming words (and dictionaries), so it can de intelligently processed to reduce keystrokes.

Interestingly, my Nok with T9 will show some choices for words even if I type if garbage -- pronounceable words with no meaning in english: it is doing some low-level language processing looking for "right looking words" (I don't think it is cheating by looking at dictionaries for the other languages: they fit english, german, french, italian, spanish, dutch, portugese, danish, swedish, norwegian, and finnish in my itsy bitsy 3210, but CEILSUZIMLEC doesn't appear to be any of those languages, but it is pronounceable!)


-adrien
[ Parent ]
Long words (none / 0) (#29)
by ambrosen on Thu May 31, 2001 at 08:40:16 AM EST

After you've typed about 6 or 7 characters, the T9 doesn't put up ?s for new characters it doesn't recognise, but instead it puts out the most likely letter based on (I guess) the previous one (or the bigram probability, to be terminologically accurate). That's my guess. I wrote a T9 interpreter in half an hour once, and I was thinking about doing a hook to put it into emacs, so I could type one handed (while using the mouse with the other hand, of course.), but seeing as the keypad's on the right of the keyboard, it only leaves the left hand free, so it's not much use. That and I couldn't be bothered with emacs lisp.

Ambrose

--
Procrastination does not make you cool. Being cool makes you procrastinate. DesiredUsername.
[ Parent ]

Feeble attempt at humor (none / 0) (#13)
by WinPimp2K on Wed May 30, 2001 at 05:56:43 PM EST

Yeah, but they don't have such big fingers!

[ Parent ]
Predictive text input on mobile phones. (none / 0) (#15)
by deefer on Wed May 30, 2001 at 06:47:46 PM EST

On my Nokia 8210, when you're typing messages, it tries a brute force dictionary attack on the keys you've hit so far. So typing 4663 could give you good, home, gone, hood, hoof, etc. There's a button to bump it along to the next entry in the dictionary. Hard to get used to, but it's OK when you get your head around it.

Now only if I could teach it some swear words! :)


Kill the baddies.
Get the girl.
And save the entire planet.

T9 (5.00 / 1) (#16)
by adrien on Wed May 30, 2001 at 07:19:39 PM EST

The software is called T9, and it's mentioned in the article.

I think its kick-ass.

Less atoms, more bits - smart interfaces for cramped spaces.

Mine does learn words (funky old Nok3210). It learned the name of my town (vevey) and some others.

What I really want is a hardware T9 keyboard that doubles as a lid for my visor. There is enough space for 9 really nice keys, a "bump to the next possible word" button, return, delete, shift, number-mode, and a couple others. T9 seems to me the most efficient, easy to learn, fast and cool text input system out there. Too bad it is currently restricted to the itsy-bitsy mobile phone keypad ghetto.

And don't steal my idea.


-adrien
[ Parent ]
While we're at it (3.00 / 1) (#17)
by xriso on Wed May 30, 2001 at 07:44:25 PM EST

Why not a super-chorded system: binary. Just have 10 buttons and you can have a nifty palette of 512 key combos (you need one reserved for "send", I suppose)

Or, go for the good old neural interface. Once the tech exists, of course.

Or (here's a good one), use a more bulky device for input, but have a universal protocol so everything can use it. Then you only need one.

Or, with the previous idea, add a feet keyboard for some extra buttons.

Or, do it Hawking-style. (The weelchair is a bit bulky to carry around, though)

Or, stop trying to do complex tasks on tiny devices - wait till you get to your desktop, or at least a laptop.
--
*** Quits: xriso:#kuro5hin (Forever)

typing in binary (5.00 / 1) (#31)
by adrien on Thu May 31, 2001 at 12:22:59 PM EST

"Why not a super-chorded system: binary. Just have 10 buttons and you can have a nifty palette of 512 key combos"

As mentioned in the article, The winnebikeo uses four sensors/buttons on each handlegrip to type "in binary" - that would be 7 bits of ascii code and a 'send' key. 8 fingers, 8 bits, nice match, huh? That way the thumbs can still be used for changing gears, whatnot.

Why not learn ascii / binary and type with that? I like the idea and it is kindof appealing (in a wannabe ubergeek kind of way), but I can think of a few reasons why not:
ascii is made for machines. there is some logic to it, but for a text input interface for fingers and human minds, it's not the best. A well thought out system oriented around language patterns and the human mind would be much easier, efficient, more accurate and probably faster. Of course, if you really wanted to you could type like hell on binary. If you really wanted to, you would be an astronaut, too.
"More" can be done with "less" -- look at the bat keybaord -- 7 keys, one for each finger and 3 for the thumb. It might be cool, but a conventional keaboard with 512 keys would be a bit impractical.
For one handed typing (like on a cellphone, yeah, ahem) a binary system would give you only 5 bits, giving you just 16 characters plus a send key. doesn't quite cut it.

There are a few nice things about typing in binary, I admit -- lowercase letters are 32 decimel numbers higher than their respective capitals, wich means one key would act as a "shift" key even in binary - nifty. Check out Daniel Kohanski's The Philosophical Programmer, I do remember him talking about this kind of stuff (but coming to similar conclusions as I do).

neural interface? it's coming -- but I still put it on the list with antigravity, AI, and nanotech -- nifty SciFi, but I'm not gonna hold my breath. But, I may be wrong.

One big input device and lots of little daemon devices taking care of small tasks without me having to directly work with them? Bluetooth. Great. How big of a keyboard and display do you want to lug around? Given the proper interfaces, your 'control center' could easily fit in one hand, with a display, and 80wpm. Why lug around a keyboard if you don't have to?

No, I'm not gonna walk around the city with HUD and neural interface strapped to my head.


-adrien
[ Parent ]
Chording Keyboards and Handspring Visor Edge? (none / 0) (#18)
by cp on Wed May 30, 2001 at 07:52:49 PM EST

I'm one of those idiots who got a handspring visor edge, which has a different number of pins at its base and is therefore incompatible with other cradles, and I've struggled to find a chording keyboard for it (much less any keyboard at all). Of the chording keyboards out there, the twiddler2 looks like the only one that supports palmos devices, and it claims not to support any handspring models, much less the visor edge. Has anyone had any luck in this area?

would be nice (none / 0) (#25)
by adrien on Thu May 31, 2001 at 05:17:37 AM EST

But I haven't seen anything.

the oragami keybaord (thinkoutside.com) is available for the Visor, though


-adrien
[ Parent ]
not good enough (none / 0) (#27)
by cp on Thu May 31, 2001 at 05:28:18 AM EST

Like I said, the Visor Edge isn't compatible with other visor models, because of the different pin arrangement. There are lots of portable keyboards for the other visor models (from targus and others), but none so far for the visor edge.

[ Parent ]
Missed one (none / 0) (#22)
by Wondertoad on Thu May 31, 2001 at 02:06:21 AM EST

The article misses the "half keyboard" discussed on that other site in this thread. Which mentioned this article.

nope, it's there: (none / 0) (#23)
by adrien on Thu May 31, 2001 at 04:24:22 AM EST

"If you've got more than one finger and don't want to change to a more optimized keyboard layout, you can just cut your keyboard in half."


-adrien
[ Parent ]
Dave's Dasher (4.50 / 2) (#28)
by pw201 on Thu May 31, 2001 at 05:28:43 AM EST

My friend Dave has a thing called Dasher. It's a graphical method of text input backed up by a language model which learns as it goes along. You can use it on notepads: there's a Pocket PC demo you can download.

It works by letting you fly towards letters inside boxes. The sizes of the boxes are proportional to the probability of that letter occurring next. Inside each box there are smaller boxes for the next set of letters, and so on. Hard to explain: the easiest thing is to go to the page and watch the animation.

nifty (none / 0) (#30)
by ODiV on Thu May 31, 2001 at 11:21:03 AM EST

If I knew anything about programming I'd try to port it to palmOS. Maybe I should look into it anyway. I've been feeling pretty un-techie lately.


--
[ odiv.net ]
[ Parent ]
This sounded neat... (none / 0) (#34)
by Kaki Nix Sain on Fri Jun 01, 2001 at 07:49:47 PM EST

...So I tried it. It took me a few minutes to spell out my name. Not very helpful.

"Hey teacher, could you slow down, this damn thing won't give me an 'r'."

Btw, is "sounded" a word?

Niche at best.



[ Parent ]

Once it's learnt your name (none / 0) (#35)
by pw201 on Sat Jun 02, 2001 at 07:15:15 PM EST

it should be faster: the point is that the language model adjusts the size of the boxes to make it easier to type things you commonly type.



[ Parent ]

I still think... (none / 0) (#36)
by Kaki Nix Sain on Sat Jun 02, 2001 at 08:02:18 PM EST

... it isn't a good idea to limit what letters one has to pick from for (sometimes) long periods of time. I mean, this thing imposes idle time while one waits for it to decide that you might actually like to pick a 'w'. The speed increase of when it gets things right needs to be much larger before the penalty when it gets things wrong can be offset.

Also, it forces one to think in letters. I don't know about you, but I left letter thinking behind a long time ago with keyboard entry. My brain was able to abstract away from the letters because they are always in the same places. Not so with this scheme.

On the other hand. It makes a nice random sentence generator. Just follow the big boxes. :)

Oh, and I just realized that this is a 1-D input model. Come on, I've got more channels of output to use than 1. People with 1 channel of output, I'm sure this could be a good solution for them.



[ Parent ]

Aha! (none / 0) (#37)
by pw201 on Mon Jun 04, 2001 at 11:48:46 AM EST

... it isn't a good idea to limit what letters one has to pick from for (sometimes) long periods of time.

It doesn't. If I remember rightly, the more improbable letters are still there, they're just in small boxes. If you head towards the place where they should be, they'll become visible. I agree that if you didn't know this it'd be rather painful to type with.

Also, it forces one to think in letters. I don't know about you, but I left letter thinking behind a long time ago with keyboard entry. My brain was able to abstract away from the letters because they are always in the same places. Not so with this scheme.

That's true. I'll point Dave at this discussion and see what he wants to say about it.

[ Parent ]

How small? (none / 0) (#41)
by Kaki Nix Sain on Fri Jun 08, 2001 at 10:04:17 AM EST

If I remember rightly, the more improbable letters are still there, they're just in small boxes. If you head towards the place where they should be, they'll become visible.

They seemed to be so small that I couldn't see them or choose them (even after I figured out that they were alphabetically ordered) for a while... i.e. until something other than me decided that it would give me the option.



[ Parent ]

Hello (none / 0) (#38)
by davidward30 on Mon Jun 04, 2001 at 12:11:22 PM EST

Ideally, the language model would be trained on a large database.

If the entropy of the entered text under the language model is close to its true entropy, then there is no problem making inprobable letters small. And therefore you would never/rarely have to sit and wait for your 'w'.

> Also, it forces one to think in letters.

Not necessarily. When you've practiced a bit and get up to 20wpm, you'll find that whole words come
flying out at you. This effect would be increased with more training text or a better language model.

Yours
David Ward.

[ Parent ]
Gestures and things (none / 0) (#39)
by pw201 on Mon Jun 04, 2001 at 12:47:39 PM EST

Not necessarily. When you've practiced a bit and get up to 20wpm, you'll find that whole words come flying out at you. This effect would be increased with more training text or a better language model.

Indeed. However, Kaki's objection is less about the letters than about his desired for a consistent gesture which forms a word: he argues that he thinks in these gestures rather than the letters which make up a word, in the same way as you don't really think of the individual notes when playing a keyboard (on a musical instrument, I mean). I think that I work in a similar way too, as I have to think much more about typing an arbitrary sequence of characters than I do about typing an English word.

But it's hard to see what he'd want, in that case. I guess a completely gesture based system where you played "chords" for words on some kind of pad would be the ultimate thing, but it'd be hard as it'd require you to learn the gestures. Isn't this what stenographers in courthouses do, or the people who write live subtitles for TV programmes? Probably too specialised for people to want to learn it unless they have an incentive.

As you've also pointed out, Dasher isn't meant as a replacement for a keyboard but is meant to be for situations where using a keyboard isn't practical. It seems to be that there's got to be a tradeoff between fixed gestures and keyboard "size" (if you can consider Dasher as in some sense a small keyboard in that you only need one finger!).

[ Parent ]

Good job. (none / 0) (#42)
by Kaki Nix Sain on Fri Jun 08, 2001 at 11:04:16 AM EST

You did very well answering for me before I could get back to this. Thanks.

As for my dream input ideas... they are numerous and probably only about half of them are practical or would work all that well at all (1/2 if I'm lucky). The cold hard truth is that a big keyboard, with more keys then fingers is a really good solution to the input question.

Getting more generalized in thinking, one wants to make full use of all of the degrees of freedom that the person can comfortably use up to the number of dimensions needed by the device to determine a unique input for the type of information at hand. Which is why the Dasher would very much work for some niches, I'd be crazy to dispute that. One that might be interesting to try would be a head-tracking or eye-tracking input. Since, in that case, the person is giving 2-D output (well, I guess 3-d or 4-d for the head tracking would be possible).

Getting back to my dreams, I like the idea of having 5 or 6 keys for one hand and 5 or 6 for the other to shift and alt and meta the keys for the more dominant hand. Give each hand a wheel and/or a ball, and I'm close to crying from joy. Since that is a lot to ask for, and might take a good bit of room, how about five or six keys for one hand, with three shifts (that is more than enough to give all the letters in my language) and a wheel for the other (and I can do without the wheel, though I love them so).

I thought of another thing since my original trial of Dasher. I went through a few steps. First, why is it cartisian in layout? Why not try polar? Then I realized that one could use a polar layout of letters with a trackball input. The most common letters could be immediately acessable by ranges of angles, while groups of less used letters could be nested down in a little subspace that one gets to by first selecting the arc for that group and then making a choice among the letters in that group. This scheme benifits from some rather obvious scalability. And if the letters aren't shifting positions or arc sizes, then one should be able to learn the gestures for the letters and symbols that are down in the subspaces. I've drawn up a few ideas for the layout of this scheme, but plenty of trials would have to be done. For one thing I don't know the accuracy with which someone can move a trackball. The ball might need to be made heavier or provide resistance some other way. But I think this could provide something like 46 input options with 3 ball spins or less.

I think that is sufficient to give an idea of why I said what I did about 1/2 my ideas... ;-)

Oh, and why don't high-schools teach what stenographers use instead of qwerty? Is there a reason or is it legacy?



[ Parent ]

polar dasher (none / 0) (#43)
by davidward30 on Mon Jun 11, 2001 at 05:28:02 AM EST

We already considered a polar version of Dasher to make more use of the 2nd dimension. However, it means that you would end up writing backwards when on the left hand side of the circle. Yuk.

'And if the letters aren't shifting positions or arc sizes'

But thats defeating the whole point of Dasher which to make use of the HUGE redundancy of English.

Then you just end up with something like quickwriting don't you:

http://www.mrl.nyu.edu/~perlin/demos/quikwriting.html

Which is non-probabalistic. Interesting, but nothing like Dasher.

You mention eyetracking - we have Dasher working with an eyetracker. Its not bad - better than using an on-screen keyboard in my opinion.

David.

[ Parent ]
more... (none / 0) (#44)
by Kaki Nix Sain on Sun Jun 17, 2001 at 02:27:18 AM EST

...gee, I hope you check your comments page for replies, else you will likely never see this...

We already considered a polar version of Dasher to make more use of the 2nd dimension. However, it means that you would end up writing backwards when on the left hand side of the circle. Yuk.

Actually I was thinking that you don't really need half of your directions just to tell the cursor to backspace. Something like 45 degrees would seem to be more than enough for that, leaving more room to spread the letters out on.

'And if the letters aren't shifting positions or arc sizes' But thats defeating the whole point of Dasher which to make use of the HUGE redundancy of English.

Yeah, I guess I'm just wanting the dasher input-guess to never get in the way of less probable inputs. So I'm dropping it altogether, i.e. going to far.

Its just that the coding redundancy of english is already exploited to some extent by the muscle memory I have for typing. But this is comparing apple and oranges, since dasher is for niches that don't allow for keyboards.

Perhaps dasher could look at where you are headed in the input space and update its guess about what you want. Maybe it does and I missed that part... I'll retry it.

btw, I tried the quickwriting demo and loved it, but I can't get it to work on my palm. It seems off kilter, behaves strangely, and I can't get to the configuration display. But thanks for the url; I'm still hoping I can get it going... I liked graffiti for a while, but the trade-off between learnability and top speed has since become a bad trade.



[ Parent ]

laptop and even normal keyboards (none / 0) (#32)
by tforce76 on Thu May 31, 2001 at 12:28:05 PM EST

i haven't even attempted to type on a PDA. i have enough trouble typing on a normal-sized keyboard. i have an ibm laptop for work, and the keyboard is way too small for my fingers. my fingers are literally too big for the keys, and i wind up hitting like 3 other letters i didn't mean to hit, so "the" becomes "thfre". i often have trouble with the m$ natural keyboard on my system at home as well.

i'd pay for a keyboard that's a little bit larger.



stylus, fitaly (none / 0) (#33)
by adrien on Thu May 31, 2001 at 01:40:41 PM EST

One word - FITALY. It's optimized for use with a stylus -- you might have big fingers, but I bet you stick isn't bigger than mine! (sorry)


-adrien
[ Parent ]
gloves (none / 0) (#40)
by speek on Tue Jun 05, 2001 at 02:33:39 PM EST

why not gloves in which you sign? That way, I don't have to carry around anything (except wearable gloves), I can sign as fast as my fingers will let me, and I don't have to invent some bizarre keyboard protocol - heck each person could "extend" the basic sign set with their own macros. I think with wireless technology, this could work very well.

Side benefits:

  • Everyone would learn to talk to deaf people :-)
  • less disease spreading (wearing gloves all the time)
  • The inherent humor of pissing off your co-workers by gesturing at their computer.

    --
    al queda is kicking themsleves for not knowing about the levees

  • 10 big fingers | 44 comments (37 topical, 7 editorial, 0 hidden)
    Display: Sort:

    kuro5hin.org

    [XML]
    All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
    See our legalese page for copyright policies. Please also read our Privacy Policy.
    Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
    Need some help? Email help@kuro5hin.org.
    My heart's the long stairs.

    Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!