The Asset Keyboard

The keyboard you likely have in front of you (named "Qwerty", after its top row), was first available as a typewriter sold in 1874. A common myth is that this keyboard was designed to slow down typists to prevent the keys from jamming; but from the accounts I've seen, it was only designed to prevent the keys from jamming: no more, no less. The original typewriter design simply wasn't very good. Rather than improve it, they rearranged the keys from the original alphabetical order to one that kept common pairs (or "digraphs") of letters away from one another. Or at least, so it is said: I'm a bit puzzled that common digraphs like TH, IN, ER and ED are fairly close to one another.

It is clear, however, that it was not designed for fast typing. At the time, everybody typed with two fingers, and even with two fingers, key jamming was a problem.

Oddly enough, it took 60 years until Qwerty's speed-savvy competitor appeared, Dvorak. Sadly for Mr. Dvorak, his keyboard never really caught on. But there is reason to believe it actually isn't much better than Qwerty, and so, in the computer age, there have been a few attempts to improve upon it.

This page notes problems with the Qwerty and Dvorak keyboards and introduces a new layout called Asset. Asset is designed

to be easier to learn than Qwerty for beginners
to be easier to learn than Dvorak for those who already know Qwerty
to yield a higher typing speed than Qwerty for a typical computer user

Introducing the Asset keyboard:

Having been fed up with Qwerty, I decided in 2004 to learn Dvorak. I was surprised how difficult it was to learn, and noticed a curious phenomenon: I would often type letters on Dvorak with the correct finger, but the wrong hand. I came to the conclusion that the hand to use was distinct in my brain from the finger. This led me to hypothesize that there were three separate things I had to learn in order to use a new layout:

The hand to use
The finger to use
The finger motion to use

If this was correct, then a key assigned to the same hand as for Qwerty would be easier to learn, and a key assigned to the same finger would be easier still. One of my major beefs with Qwerty was that the keys were on inappropriate rows, so that the hand must do impressive gymnastics to touch-type on it. With these ideas in mind, I designed the original Asset keyboard on May 17, 2004. It gathered dust for awhile, and on November 10, 2006, I tweaked it to the layout shown above. It's called Asset because

The left home keys are ASET, and
This layout leverages users' existing asset, their knowledge of Qwerty, to improve learning speed.

Analysis

Principles of layout design

It is widely agreed that the following principles should guide layout design for touch-typing:

It is faster to type consecutive letters should be typed with different fingers. For example, "ED" is a common word ending, so "E" and "D" should be assigned to different fingers. It is a widely-held belief that hand alternation is the most important factor; for this reason, Dvorak's layout had all vowels on the left side, and all common consonants on the right. This relies on the tendency of consonants and vowels to alternate (though it's hardly a consistent pattern in English). Anyway, some designers believe that sequences of letters on the same hand can be typed at least as fast as sequences that alternate between hands. I do not know of evidence either way, but we can all agree that it is slow to type two different, consecutive letters with the same finger.
The most common letters should be on home row (the middle row), because fingers will require less movement and movement takes time.
Within home row, the home keys (ASDF / JKL; on Qwerty), except perhaps the pinky keys (A / ;), are easiest and quickest to press.
The bottom row is most difficult to reach and should be populated with uncommon characters. Also, the pinky (little finger) is the least dextrous, especially when pressing keys outside home row.

While the principles are agreed upon, their relative importance is not. I haven't yet searched for papers that weigh these matters.

I would add these additional considerations:

As I said, it's easier to switch away from Qwerty if the new layout resembles Qwerty.
The B and Y positions are difficult to reach, so take longer to reach.
The index fingers are assigned 5 keys each, so the potential for digraph clashes increases for these keys.
Backspace is a commonly-pressed key. Many have suggested turning Caps Lock into a backspace. Then the question arises of where to put the Caps Lock: shall it replace the old backspace? or replace a rarely used key such as the Right Windows key or Right Alt key? or require the user to hold shift while holding the former Caps Lock key? This is an worthwhile issue to address, but in Asset I have decided to rearrange characters only.
Keys that take longer to reach also strain the hands more; so it seems plausible that a better layout will lead to fewer repetitive stress injuries.

I've divided the alphabet and other common characters into classes according to similar frequencies, i.e. letters within a class occur with similar frequencies. This is based on a sample of over 1000 books (mostly fiction). That page also contains counts of how many words can be formed with a given set of letters--for example, about 30 words can be formed with the Qwerty home keys, ASDFJKL. These counts are based on a word list called 2of12.txt which is found in a package 12dicts-4.0 at Kevin's Word List Page. Note that this list contains abbreviations, some acronyms, and all word variants. By the way, here's a page with other interesting statistics, though it seems to be based on small sample sizes.

Class		Approximate prominance of each character in English documents, respectively
0. Space character		18.7% (may be skewed upward by certain documents that have large runs of spaces)
1. Most common characters	E T	9.6% and 7.0% (total 16.6%)
2	A O I N H S R	6.2% to 4.4% (total 36.5%)
3	D L U	3.5% to 2.3% (total 9.0%)
4	M C W G F Y	1.9% to 1.6% (total 10.48%)
5	P , . B	1.3% to 1.2% (total 5.0%)
6.	K V " ' -	0.7% to 0.3% (total 2.8%)
7. Least common characters	? X J ; ! Q Z :	0.2% and below (total 0.7%)

See also: Letter frequencies of French and German (based on this page)

French	E.ASITN.RUOL.D.XCMP.VQYGFB.HZJ.KW
Spanish	EA.OS.RNID.LCTU.MP.BGYVQHF.ZJ.XWK
German	E.NIRS.ATDUH.LGOC.M.BFWK.ZVP.JYXQ

Asset's design is English-centric, but I did consider the above frequency information in my design.

Problems with the Qwerty keyboard:

Of the nine most common letters in English (ETAOINHSR, in that order), only three are on home row (ASH), and only two, A and S, are home keys. Five (ETOIR) are on the upper row, and one (N) is on the lower row, making it especially difficult to press.
A disproportionate number of very common characters are found on the left side of the keyboard, giving the left hand a bigger workload. The common characters on the left (ETASRDCWGF) are 42.5% of characters in English, while those on the right (OINHLUMYP) make up only 32.6%.
It bears no resemblance to the alphabet, making it difficult for new users to learn.
IMO, the most difficult keys to hit on the keyboard are B, Z, X, C, V, Y, slash, period, comma, P, and Q. Therefore, all of these should be assigned to rare characters. Unfortunately, Y and C (total 3.4%) are fairly common and B, P, period, and comma are more than 1% each.
The common digrams ED and DE (total 1.35%), among others, are pressed by the same finger, which can introduce significant delay in touch typing. In fact, when I'm typing fast, I deal with this specific problem by using my index finger to hit D and my middle finger to hit E.
Qwerty is very boring to learn because very few meaningful words can be formed with the keys on home row. Thus typing tutors typically have students typing nonsense for the first several lessons.

Problems with the Dvorak keyboard:

It is unpopular.
It bears no resemblance to Qwerty, so it is difficult and frustrating for experienced Qwerty users to learn.
I (5.2%) is not one of the home keys, and R (4.4%), is not on home row.
L (3.2%), on the top row, is hit with the pinky.
M and W (total 4.8%) are on the bottom row, while F (1.6%) is in a difficult position. In fairness, the same criteria weigh against Asset's placement of Y and M.
comma and period (, .) (1.2% each) are placed at the same place as W and E on on the Qwerty keyboard. It would be better to place more common characters in these positions.

Advantages of the Asset Keyboard

Asset resembles Qwerty. As illustrated above,

12 letters are in the same place as they are on the Qwerty keyoard: Q W A S X Y U H V B P M. All punctuation and numbers are in the same place except the ; : key.

11 letters moved, but are pressed with the same finger: C F G E T K L N I O J.

2 more letters moved, but are pressed with the same hand: Z D.

1 letter (R) and one punctuation key (; :) switched hands. Qwerty is imbalanced in that the left hand must press more characters than the right. By putting R on the right hand side, this imbalance is largely corrected

All ten of the most common English letters are on home row.
Eight of the nine most common letters are on the home keys, except H, which was placed to match Qwerty.
Over 1000 English words can be formed with the home keys alone, and about 3000 English words can be formed with the home row keys, including such basic words as
a an are as at did do does has had he her here there in into is it its no neither nor not on or she so to too then than that their these

Thus, while initially learning this keyboard, many meaningful sentences can be constructed with only the home keys and home row. Additionally, 60 words can be formed from ASET alone and 10 from NIOR alone. Unlike with Qwerty, Asset touch typing tutors do not require users to type dull sets of nonsense words.
Because 12 letters are in the same place, a Qwerty-to-Asset transition course can be developed which starts with those 12 letters plus E. Over 300 words can be formed from these ten letters. Then, if we add the other home keys (ETNIOR), almost 9000 words can be formed! Adding the letter D raises this number to almost 14000.
- Most of the letters that have moved are still pressed by the same hand, which means users do not need to change which half of their brain sends the signal for each key. For the most part, users need only to learn new movements, and it should help that many keys are pressed with the same fingers.

Problems with Asset:

It's new, and you can't buy one.
It bears no resemblance to the alphabet, so new users cannot leverage their knowledge of the alphabet to learn it.
I left M and Y (total 3.5%) in place to increase Qwerty similarity. Small issues like the placement of "A" (6.2%) on the pinky are also done for Qwerty similarity.
As with Qwerty and Dvorak, it is a poor choice for one-finger typing on a Palm, a PocketPC or a Tablet PC. I have addressed this problem on the Palm with specific layouts for MilliKeys.
It is not designed for non-English languages (though it is not especially bad for other languages). I am not multilingual but here are some thoughts:
- In French, H is rare and X is relatively common, so it may be good to swap those.
- In German, L is more common than O so it would be wise to swap them, which improves Qwertz similarity. However, the difference may be too small to justify. Y is very rare so it may be preferable to swap it with Z. German also needs a ß key.
Semicolon is on the left side. This offends my brother's sensitivities, at least--he figures all punctuation should be on the right. No doubt it's impossible to please everyone, which may be part of the reason new keyboards don't catch on.
I have not yet formally analyzed its hand-alternation or finger-alternation efficiencies, nor used a theoretical model to estimate maximum typing speed.
Round brackets are generally more common than square brackets. This leads to the suggestion of swapping the round and square brackets, so that the user need not hold Shift to get round ones. Asset does not include this because some methods of keyboard reprogramming cannot move "(" and ")" without also moving "9" and "0".

The design change

(Original design)

On November 10, 2006 I moved several keys to increase Qwerty resemblance. I originally developed Asset in 2004, but so far as I know, no one used it; thus, making changes would be harmless. The new design sets Asset apart from Colemak, which was designed somewhat later than Asset. Colemak boasts Qwerty similarity, but it only seems to be a minor consideration. The new Asset keeps 3 more keys in their Qwerty positions and a further 2 keys on the same hand. With respect to speed, I anticipate one disadvantage, because I returned the "Y" key to its Qwerty position. In my opinion, this is a difficult position to reach and, as a relatively common English letter, it pains me to put it there. However, the increased Qwerty similarity seems to make up for the loss. Also, there's less loss in other languages, where "Y" is rare. So while the changes make it the champion of Qwerty similarity, it still provides substantial potential speed improvements.

At this point I still haven't verified how easy it is to learn Asset. I plan to use myself as a guinea pig in the near future.

Patent?

I don't believe patents should be granted for easy or simple inventions such as this one, nor the tens of thousands of others that are patented every year. Asset is not patented, so it may be used freely by anyone.

Additional thoughts

The shape of the keyboard
It seems to me that the bottom row of the standard keyboard is misaligned. When your fingers are on home row and you move them down, they end up between the keys below, which makes it unnecessarily difficult to press keys on the bottom row, especially (in Qwerty) Z, X, C, comma, period and slash. I believe these keys should be moved left by about 1/4 the width of a key. This will improve alignment for touch typing while not excessively annoying experienced typists.

I also find that while pressing those six keys, I need to use my fingernails to apply pressure. This is especially true on my small laptop, where there is not enough room to stretch out my hands enough to avoid using my fingernails. I'm sure the ergonomics people have looked at this, but for some reason an improved shape for the keys, that will allow the base of the finger to apply the pressure, has not generally been introduced.

I hear that there were several keyboards in the 70's and 80's that had the Control and Caps Lock keys swapped from where they are now. At this point I am so accustomed to the status quo that I would resent going back. However, Caps Lock is easier to press from home row, so it is a shame it was wasted on the Caps Lock function which I almost never use. Meanwhile, I might use the Control key hundreds of times per day. Of course, Backspace might be a more beneficial function to put in place of Caps Lock.

I wish the keyboard designers would quick screwing with the extended key region. On the right side you can see the standard extended key block. (Insert, Delete, arrow keys, etc.) This style has been around for at least 20 years, but for some reason many new keyboard designs have a rearranged extended key block. Often they also make it less ergonomic by shrinking the keys. Look, it was fine the way it was. Besides, if you're going to change it, get some industry consensus first--I think I've seen three different layouts besides the standard one. And another thing: there have always been spaces between F4 and F5, and between F8 and F9. Some keyboards now remove that space. Why? It ain't broke, and that don't fix it.

Hey laptop designers! Quit putting the FN key where the Control key is supposed to be! The FN key is one of the least important keys on the keyboard so it should get out of the way. Hey widescreen laptop designers! A wide laptop should have a wide keyboard! There's room for a full size extended key block in some of these gigantic machines, so go ahead and put it in. Better yet, multiplex a full size extended key block with a numeric keypad. And while you're at it, how about a subwoofer? Yeah...
Caps lock as backspace?

The backspace key is pushed often and should be readily accessible. Many have suggested replacing Caps Lock with Backspace. If this is done, Caps Lock needs a new home. It could replace the old Backspace, or it could be activated with a key combination such as holding Shift and then pressing the old Caps Lock key. A rarely-used key such as the right Windows or Ctrl key would also be ripe for replacement. Asset does not include this change because I consider it a personal preference.
Punctuation etc.
It's unfortunate that the shift key is required for round brackets but not square ones. It should be the other way around. It's also unfortunate that keyboard designers can't agree on where to put the backslash.

It seems odd that / ? were put on one key instead of say, ! and ?. It also seems strange that ? requires shift while period and slash do not.

It seems odd that minus does not require Shift but plus does. I think + and = should be swapped.

The Windows key you see on new keyboards is rather presumptuous, isn't it? It presumes that everyone uses only Windows on their PC and solidifies Windows as the status quo. However, I think the key itself is a good idea. The Windows key is normally reserved for system-level commands, i.e. commands that are independent of the active program. This is useful. I would just like to see the icon changed to something generic. Also, there are so many modifier keys now (9: two Shifts, two Controls, two Alts, two Windows, and one Context Menu), that two Windows keys seems excessive.
Extended keys: skilled document editing requires heavy use of the "extended" and modifier keys such as the arrow keys, Delete, Home, End, and Ctrl. I wonder if these could be beneficially rearranged, but I have not really thought about it.