[ExI] [Open Manufacturing] Re: [Robotgroup] Alternative human keyboard interfaces

Bryan Bishop kanzure at gmail.com
Mon Jun 8 01:34:33 UTC 2009

On Sun, Jun 7, 2009 at 8:22 PM, Paul D.
Fernhout<pdfernhout at kurtz-fernhout.com> wrote:
> Bryan Bishop wrote:
>> Really the design process should be flipped around. The task is to
>> type quickly and to type a relatively equal or relatively larger set
>> of symbols out, without breaking the laws of motion of the human body,
>> and so on and so forth.
> I didn't notice it in the list, but have you looked at the idea of "data
> gloves"? Basically, anything that records the relative motion of your finger
> joints can be used for typing input, even without finger tips contacting a
> surface.
>   http://www.google.com/search?hl=en&q=data+gloves
> You could also get haptic feedback in some and use them for 3D manipulation.

In the back of my mind I was considering data gloves at one point,
yeah. Haptic force feedback would help the problem that I was thinking
of with them: namely, it's easier to do typing if you have some
tactile sense of the topology of the interface. It seems to be how the
motor cortex works, or something. So, force feedback might help. I
wonder how to map it to 3D space though for different keys and motions
.. that's a hard 5D image to visualize.

> But if you don't need to be mobile, stenographer's Stenotype keyboards might
> be the fastest. And might take several months to years to train on.
>   http://en.wikipedia.org/wiki/Stenotype
> "A stenotype or shorthand machine is a specialized chorded keyboard or
> typewriter used by stenographers for shorthand use. A trained court reporter
> or closed captioner must write speeds of approximately 225 words per minute
> at very high accuracy in order to pass the Registered Professional Reporter
> test[1]. Many users of this machine can even reach 300 words per minute and
> per the website of the California Official Court Reporters Association the
> official record for American English is 375 wpm."

Uh, what? I don't see how this satisfies the other requirement of
special characters being easily typed. IIRC, stenography machines are
mostly about phonetics more than they are about specific ASCII or
unicode keycodes.

> But voice input can easily get up to 150wpm, but with 1% to 5% errors.

What about special keycodes? I guess if I go learn the international
phonetics language it might be easier to map different grunts to
different keycodes, but that sounds kind of annoying.

> http://www.voicetypist.com/The%20Question%20of%20Speed%20and%20Accuracy.htm
> "For instance, Nuance says that Dragon NaturallySpeaking (DNS) can keep up
> with a user speaking 160 words a minute, with an accuracy of 99%. In a
> noise-free setting with a practiced user, that's undoubtedly the case. But
> it's almost necessary to say, "So what?""

- Bryan
1 512 203 0507

More information about the extropy-chat mailing list