Keith Devens .com

Most efficient keyboard layout found by genetic algorithm!?

Via Slashdot (the first good link found via the new, first useful version of my aggregator), someone did research to discover an efficient keyboard arrangement using a genetic algorithm:

In the United States, the most common keyboard layout for computer keyboards remains the one designed by Christopher Sholes for the original Remington "Type Writer" in 1876. This layout is commonly called QWERTY after the order of the first few letters on its top row.

Other layouts exist. I have been using the Dvorak keyboard layout for about a year now. I like it a lot for my daily work, which involves a lot of typing. I used to feel a numbness of the backs of my hands after a long day with QWERTY, but I don't with Dvorak. And quantified measurements bear out its efficiency relative to QWERTY.

But Dvorak designed his layout in the 1930's without the aid of computers. It contains a couple annoying features that lead to common errors in my typing. Could a modern genetic algorithm and a huge input sample discover a better arrangement? I had to give it a try. The results surprised me!

This is right up my alley. I've used the Dvorak layout for years, and I never want to go back to QWERTY. Now I can be even more iconoclastic by using an even rarer layout.

Or not. After reading the article it turns out he wound up with something surprisingly like Dvorak! And it was only marginally better according to his program than Dvorak is, although QWERTY was about twice as bad as the average between the new computer generated layout and Dvorak.

As I was reading and I looked at the first keyboard he listed, thinking that it was his final version, I thought it was crap. The 'e' on the left pinky?? I thought, "He didn't take finger strength into account", but then I noticed that wasn't his final version, and in his current version the 'e' is exactly where it is in Dvorak.

Here are his current scores:

' , . p y  f g c r l   Dvorak layout
a o e u i  d h t n s   32129548
; q j k x  b m w v z

q w e r t  y u i o p   Sholes' layout, with quote replacing /
a s d f g  h j k l ;   59514344
z x c v b  n m , . '

k , u y p  w l m f c   Best evolved layout
o a e i d  r n t h s   28281895
q . ' ; z  x v g b j

Also, very interestingly, IBM had already done research in this area. The poster links to some research papers as well as the ATOMIK layout for keyboards on handheld devices. It's interesting, because this has different requirements than normal-size keyboard layouts. Dvorak prizes switching back and forth between hands, whereas on handheld devices you typically use a stylus, which is like using one finger to type.

Finally, a very interesting post from a piano player challenging some of the assumptions of both the Dvorak layout and this new computer generated layout.

Permalink | 6 Comments | id 2343

12:47 (utc-4)

Comments

All of the above is fascinating reading, including the alternative ideas by pianists and the rebuttals to these. For yet another view on keyboard design, readers are invited to examine the XPeRT Keyboard, found at www.xpertkeyboard.com, which strives to take some of the attributes of the Dvorak keyboard (namely high opposite hand keystrokes) and combine these with minimal change from Qwerty, to make transition easy for experienced keyboard users. It adds a 2nd key for the letter E, the most common letter in English at 13%.

Here is an interesting aspect that Dvorak and the genetic algorithm constraints seem to have ignored: The small finger and ring finger share a tendon. In addition to being short, the small finger does not operate independently. Placing a high frequency letter there (ETAIORSN) is awkward. Dvorak has A+S at small finger locations and N,R+O at ring fingers. For those of us who were not born to play a piano, this is tough.

For folks who can learn to touch type and those who have mastered typing at > 60 wpm, a layout like the XPeRT keyboard does not have much to offer. However, statistics show that fewer than half of touch typists actually type over 40 wpm. The vast majority of typists are of the hunt & peck persuasion, and may be limping along at 15-30 wpm. This is what Qwerty has done for us. Afterall, it was designed in 1878, to slow down typing and prevent mechanical keys from jamming on typewriters.

So, if there are folks out there who do not have the patience to learn a superior, but entirely new layout like Dvorak, the XPeRT Keyboard offers an easy and practical way to type faster, without effort or training.

Best regards, from the caterpillar.

This "optimal layout" does look very good. I have lately been fascinated with "beating" Dvorak, but I don't have a computer program, I'm just going off intuition, and looking at other good layouts such as Dvorak and Colemak, and finding their strengths and weaknesses. Some things I don't understand about this layout are why the period is down there. It's about as common as the median letters (m, c, u) but from my experience, that spot is one of the hardest to access. I liked the Dvorak spot better, because it was easy but not too easy. Also, why NTHS on home row instead of HTNS? It's NT and TH are both very common (especially TH, the most common digraph) and it's easier to go pinky-ring-middle-index than the opposite. That formation does allow for an easy SH, but other than that, it seems to be slightly worse. Along the same lines, EA is the most common 2 vowel digraph, and A,O, and I are all very close in letter frequency. It would work better to switch the I and A, so EA is easier to type.

Don't let my criticism lead you wrong, I'm criticizing the bad things, but there are too many good things to talk about them all.

There is also the Colemak design - but I think its crap. And the XPeRT is even more stupiderer.

What is the formal name of this improved Dvorak layout? You provide no real sources - the ones you did are not valid websites (anymore) - but you never even mentioned the name once, thereby completely preventing any research, which makes me doubt anything you say.

I read the bit the piano player wrote... and I have a few assumptions to add to his list.

We have made the assumption that the keyboard should be laid out in a rectangular form... yes, we have ergonomic layouts now, vertical layouts, split keys, etc... but its still pretty rectangular. I did what the piano player recommended - I placed my hand on the table top in a restful position - and I realized that my fingers form a circular arch. Why cant keyboards be round?

I also recognized that all keyboards are flat. Yes, we have vertical layouts, etc... but all keys lie in the same 2-D surface relative to one another. What about a 3-D layout? I know it may not sound reasonable, but I recently saw a keyboard which was a projected image on the table top, motion sensors detected finger movement... I realized then that a keyboard doesnt have to be a physical thing. That being the case, why cant motion detection allow for typing in three-dimensional space?

Furthermore, I appreciate the concept of multiple keys of the same letter. I also appreciate the concept of chording, eg two-key combinations produce a third unique letter.

Hey, more ideas...

All keyboards that I have seen thus far, QWERTY, Dvorak, etc... all cater to the population at large. Dvorak has two unique keyboards designed strictly for people with one hand. But the idea came to me that we could design layouts for left-hand- and right-hand- dominant individuals.

And for that matter....

Why cant each individual human being have their own individually optimized layout? Why does a layout change have to require a computer reboot and/or a registry change? I had to download and install Colemak, separately. Stupid! USB devices are pretty popular. I foresee each individual sliding their own USB device into the computer and wham! They have their own optimized layout.

Saturday, July 6, 2002

Most efficient keyboard layout found by genetic algorithm!?

Comments

Recent comments