Can Evolutionary Computing Improve Keyboard Efficiency?

[Bartholomew]

This is a thread for two ideas I came up with while discussing the parallel keyboard. The first is tactile feedback on each key, perhaps by ridges on the keys that point back to the center of the keyboard, so that the subconscious mind has a better sense of where each finger is at any time.

The second is the use of evolutionary computing methods to optimize the key placement. This is so obvious that I think it must have already been done--anyone know?

 

[chroot]

You don't need to use evolutionary computing necessarily, but it might be a fun exercise to write.

The most common letter and most common words in english are easy to determine -- if you ran a copy of Moby Dick through a program you'd have that information in a couple seconds. Then, optimize your keyboard to produce the most common letters and words with the least effort.

- Warren

 

[Integral]

This"optimized" keyboard already exits. You can access it in software on any WinBloz computer. This is the Dvorák configuration. The commonly accessed keys are placed in the center regions near your strongest fingers. What you will need is a keyboard layout so you can locate the keys then some practice. If you do not move your key caps it makes your key board unusable by anyone not aware of the change or the Dvorák keyboard.

I am sure the keyboard is optimized for English. I do not know how well it works with other languages.

Edit:

In win98 SE I went to control panel-Keyboard-language-properties.

The Dvorák layout was one of the options.

 

[Moonbear]

If you do not move your keycaps it makes your key board unsusable by anyone not aware of the change or the Dvorak keyboard.

Also a good April Fool's joke for someone not too computer saavy. Actually, any keyboard remapping will serve that purpose.

As for tactile cues, where I run into trouble with touch typing are the numbers/symbols at the top, especially when switching back and forth between full-sized keyboards and smaller keyboards on laptops. It's just enough of a stretch for my fingers from their home positions and they are keys I don't use as frequently as the letters that added tactile cues on those would be helpful, maybe on the 1, 5, 7 and 0 keys. The 5 and 7 will help keep my fingers somewhat centered, and the 1 and 0 would define the ends of the numbers. You wouldn't want to add too many tactile cues, or else you'd wind up not knowing one from the other and causing more confusion than help.

Have you ever tried one of those split keyboards that are supposed to be more ergonomic to use? They feel really weird at first, but once you get used to it, the shape of the keyboard helps with finger placement too, and you can't accidentally put your index fingers beyond the center point since it's already split for you. They aren't very portable though.

 

[ohwilleke]

The first is tactile feedback on each key, perhaps by ridges on the keys that point back to the center of the keyboard, so that the subconscious mind has a better sense of where each finger is at any time.

Most keyboards have this for a couple of keys. My Dell QuietKey (TM) keyboard has a raised F and J. If you use standard positioning for your keyboard you need only one raised key per hand.

The second is the use of evolutionary computing methods to optimize the key placement. This is so obvious that I think it must have already been done--anyone know?

As you should know, the current keyboard was actually designed to slow down typing (so the mechanical typewriters wouldn't jam). The problem is teaching your fingers placements other than QWERTY.

 

[ohwilleke]

The Dvorák layout was one of the options.

A layout and some history are here: http://www.mit.edu:8001/people/jcb/Dvorak/

The key point:

In the mid 1950s, U.S. Government's General Services Administration commissioned a study by Earle Strong to confirm Dvorak's results. Strong's study, which included proper controls and which was set up to allow direct comparison of qwerty and Dvorak data, found that after sufficient training, Dvorak typists were able to match their previous qwerty speeds, but not surpass them. Furthermore, additional qwerty training for qwerty typists resulted in a greater increase in speed than additional Dvorak training for Dvorak typists who typed at a similar rate. These results would suggest that Dvorak's claims of faster and more efficient typing are bogus, and switching layouts on the basis of speed and efficiency would not make sense.

 

[Bartholomew]

Ohwilleke, yes, I am aware of the raised f and j keys; that's actually where I got the idea. More tactile feedback would probably lead to speed improvements. It's not about just keeping your fingers on the right keys; it's about having a better spatial sense of exactly where your fingers are.

The dvorak keyboard is still a keyboard designed by humans. A few years ago I was reading about people who had an evolutionary program to optimize the placement of things in airports. The program averaged about a 30% gain in efficiency over the best efforts of the human airport designers before it. Keyboard layout is something that would be easy to program a computer to generate. Additionally, you could gather statistics on finger strength and capability and "finger independence" (thanks hitssquad) and plug that into the program, which could use the information more comprehensively than any human designer could ever do. Also, people have fingers of different lengths; the reach of my middle fingers is much greater than the reach of my pinky fingers, yet on an ordinary keyboard they are confined to the same few keys. Clearly some of the keys should be somewhat smaller, and some somewhat larger, and the keys should not be all crammed into one equal-height left-to-right strip.

All of these considerations involve great complexity, which nevertheless can be simulated easily; thus, it is ripe ground for an evolutionary computing program.