That was excactly what I was expecting to hear, thanks. :)

The keyboard design itsef was "inspired" by Maltron I have to admit.

The Shift keys should be left in their normal positions, moving them would be very counterintuitive.

I actually learned to use a layout where the shift was pressed with the left thumb in a few hours. After that the standard seemed counterintuitive. (I moved the right hand keys to the numpad and the left hand ones to the right so that Alt Gr / right command worked as a thumb-pressed shift). The most comfortable choice was quite surprisingly placing a few alphabetical keys under the thumbs, but that again would require a whole new layout and a design…

Currently there's only one Shift/Ctrl key, meaning it's very difficult to type some things with one hand.

It's difficult with standard keyboards as well. When I asked some people I know about it (some very established one hand typists among them), none of them ever used the right Ctrl/command for shortcuts. It seems easier to just bring your other hand to use for ctrl+p for example — we're more used to moving the right hand on and off the mouse than the left hand across the keyboard. And besides, no-one would care to use shift if they are typing something with just one hand, except maybe to access some special shortcuts like ctrl+shift+z where the right shift isn't needed anyway.

Home and End are very useful keys and should be conserved. Also you need to take account that there are many hotkeys that use these keys, e.g. you can press Shift+Ctrl+End to select everything until the end of the document.

It's shift+cmd+down on macs, which certainly makes a lot more sense. Shift+up would select text on the row above and cmd+up does the same as a Home key, so combine them and that's what you get. The arrows are also a bit easier to reach than Home/End. What my is point is that even though keys like Home, delete or print screen might be used for shortcuts today, we would do just as well without them. The same goes for Alt Gr basically.

But if either of the other two people who read this thread use Home/End frequenly, let me know where. This will determine if they will be seen useful enough and included in future keyboards, which will all obviously be modeled after Thumb-o-Board, or not.

Mac users will miss the eject key.

Actually you have a point here. I was about to whine how it's not up to the hardware-maker to decide which controls should be included in keyboards and which in the hardware itself, but then again almost everyone needs a shutdown, eject and the volume control buttons so they could actually become part of the new "standard" set of keys. Then hardware itself could be entirely free of controls. Monitor brightness and power controls would be extremely convenient in keyboards as well.

Although on normal keyboards the thumb is underutilized, you're probably overworking the thumb which can strain the thumb and the wrist. The thumb can quite easily move horizontally, but not so much vertically. The Backspace/Left/Alt won't be comfortable to type.

Of course the keys under the thumbs can also be pressed with other fingers like you'd do with the arrows and others on normal keyboards.

Pressing ANY of the special keys on standard keyboards causes strain for the wrists. In Thumb-o-Board mostly just Esc, Tab, Page Up/Down and the arrows, along with the punctuation cause it. And the thumbs are still underworked, IMO more stress should be put on them than even on the index fingers. They just tend to get strained on normal keyboards because the space is in such an unnatural position.

I found the Alt/Btckspace position by printing the layout out and just experimenting what felt the best for me (as I mentioned in the submission thread), but it should quite certainly still be moved a bit now that you've mentioned it as well.

Ok, this got a bit out of hand, but I'll try to wrap things up (even though I have worked on the layout about ten times more)

Very difficult to learn, moves many more keys

That's true, but it isn't *that* much more difficult to learn. In my layout 6 keys are left untouched, 5 are pressed with the same fingers, 1 is mirrored and 3 are pressed with their mirrored qwerty counterpart fingers. In Colemak the same numbers are 10, 2, 1 and 0. Learning mirrored layouts is actually ridiculously easy — I did this many times during the design process. Colemak has also many keys close to their original positions, which should reduce hunting during the first few days, but then again a question should be asked that how much should be sacrificed in learnability compared to the actual gain in ergonomics/speed over qwerty.

I also think that part of the learning process is about associating speech with motions of the hand. As I have placed most vowels — and common syllable nuclei — under the middle fingers, the sonorants n, r, l and m under the index fingers and the obstruents t, d and g under the first two fingers, people begin to associate fingers with syllable parts/sounds, which should make it at least slightly easier to learn. Reaching fast typing speeds should also be easier on this layout, which is the thing that ultimately counts.

ToB has 200% more same-finger typing than Colemak (worse than Dvorak)

There's so much more to keyboard design than minimizing same-finger typing, but it is obviously the area where I get my ass kicked by Colemak. Having three out of the common sonorants under the index fingers required some sacrifices… But if the average typer should reach 60 WPM (300 characters) and the average loss in same-finger presses would be 0.5s, it would add only about (0.033/2)*300*0.5 = 2.25 seconds more time per minute when typing at the maximum speed compared to Colemak. So I think you have overstressed the importance of this factor in your design.

ToB has 200% more same-hand row jumping than Colemak (worse than Dvorak)

Same-hand row jumping only matters in a fraction of the key progressions, it shouldn't be calculated with such a simple algorithm. Sometimes it may even be harmful *not* to do a row-jump. For example OJ is easier to type on Qwerty than OU is. Arguably even ON might be considered easier than OU, even though it jumps two rows compared to OU's zero. KN is also about as easy to write or even easier than KH.

Among the most uncomfortable row-jumps are Qwerty's KU and especially DR, its mirrored counterpart. In our layouts the first is replaced with AP (ToB) and EL (Colemak), the second with EY and SP. The first is (both ways around) about 70% more common in Colemak, the second is about 250% more common.

ToB has 5% more finger distance

If your finger distance is calculated with the applet by Jon A. Maxwell's, then it is far from accurate as well. It doesn't take into account that the default resting positions of the fingers are not directly on top of the keys and that all the fingers move when you reach for any key not in a "home position". Some of the first layouts I did had about 10-15% less finger travel than Colemak, still somehow miraculously, my fingers seem to move less on this layout. Anyway, I wouldn't trust the applet in this aspect.

ToB has very low hand-alternation, possibly even worse than QWERTY, meaning that there are many times you must write a long string of characters with the same hand, which might cause strain on the hand.

This is excacly what my combos/rolls are doing, and it's only positive. There are some common single-hand syllables that are uncomfortable to type like sing, gain/goin and words like foam/soap/pink/thinking/sinking/singing but generally it's really hard to find complete (long) words you could type with a single hand.

D and O should be on the home row, but aren't.

Having O and the other vowels under the ring fingers makes it so much more probable that they, uhh, involve in a roll with their neighbours. There are also languages like German where O is not that common. Placing D on the home row would be possible, but it would likely increase the amount of words/cyllables typed with the same hand. I might still place it there for the final version, because right now the stress on the left index finger is bigger than on the right because F under the right one on the home row…

ToB strains the pinky fingers, e.g. words like "I'm" and "I'll" are a same-finger on the pinky, which are very uncomfortable to type.
Punctuation keys are strewn about, and it doesn't seem too logical or aesthetic.

These are another downsides which I apparently just have to accept. There's no way to move the apostrophe to a better position without approaching the layout from a completely different, most likely a worse aspect.

I *could* swap Q and comma to make it a bit more aesthetic, but I'm already so in love with the comma's new position that it might be hard. And the reason the bottom row goes like ZXCVBKM:Q is due to a certain migration - W is moved to colon's place in most international versions.

ToB places the common D on the W position, which is very uncomfortable.

There's still only about 8% stress on that finger, the same as with Colemak or Qwerty. And the right hand's ring finger on Qwerty has as much as 12% stress.

I find it the most comfortable to keep my ring fingers resting on the top edge of Qwerty's S and L, mainly because the ring fingers are so much longer than the pinkies and that keeps the hands in a more relaxed position. Then it is no problem reaching for the top row's W and O — or D in my layout's case. Of course this is also an individual thing, but because Qwerty's O, which is even more common than D is placed on a similiar position, most people should have no problem with it.

ToB is a work progress (I don't think you even use it yourself). Colemak's design is considered stable.

You just had to put that in didn't you. But true, it is unstable. It doesn't even have a name yet. :D But I have always used the layouts I've created (this one for a few days now, but I've used other layouts designed with similiar principles for all of december…) It has allowed me to see how they work in practice and to constantly improve them.

Actually I think I've come up with a way to move D to the home row now…

Oh yeah and I started doing the layout by slightly modifying Colemak. Right now I really think it's a lot better than it. You should try it out as well. :P

Similarity to QWERTY is very important in the learning stage in order to make learning easier and faster. Once you master a new keyboard layout, you'll often start to forget how to type in QWERTY. People who have been typing in Dvorak for a long time and need to use QWERTY for one reason or another find themselves almost unable to type. If for example, you move the Shift keys, it would be much more difficult to adjust from ToB keyboard to a normal laptop keyboard, for example.

It's a QWERTY world after all. I used Dvorak for two years before I gave up on it. I had to use QWERTY some of the time, and switching between layouts was very error prone and difficult. I find it much easier to switch back and forth between Colemak and QWERTY because it's much easier to hunt keys when I'm typing on Colemak.

The letter O definitely should be on the home position. O is the 4th most common letter in English, and it's the 3rd in Spanish, the 3rd in Portuguese, the 4th in Polish, the 4th in Italian, etc. That said Colemak is designed for English. It's impossible to design a keyboard layout for multiple languages. It will never work well for all of them. That said, in Colemak I did take into account that H while somewhat common (8th-9th) in English is rare in almost all other languages, and that's one of the reasons why in Colemak it is in the home row, but it isn't in the home position.

Although I don't use it very often, I'd definitely miss the right control. Without it, hitting shortcuts will require moving from the home position and stretching. e.g. pressing Ctrl+Enter with one hand is impossible, pressing Ctrl+Tab would require moving from the home position. On Colemak I use RightCtrl+CapsLock which acts as Ctrl+Backspace (erase last word). Pressing Ctrl+P with one hand causes a lot of strain.

You should definitely avoid moving keys to their mirrored position. e.g. even after using Dvorak for two years I got confused sometimes between the T and the Y which are mirrored in Dvorak, while typing in Dvorak as well as typing in QWERTY.

Every time you move punctuation keys, you have to relearn 2 keys, not one. e.g. when you move the comma key you have to relearn both the comma and the LessThan key. Take into account that keys that aren't used often and punctuation keys take the longest to relearn because they get the least practice. I also think that keeping the keys on the same hand eases learning quite a bit.

Same finger is definitely a bad thing, especially with the weak fingers. e.g. try to press QZQZQZ and see how tiring it is. When typing in high speeds (>80WPM), same finger becomes one of the most important factors for speed, and combos/rolls become much less significant. Every time you have to hit a same-finger, it slows you down and interrupts your typing stream. I really don't think I'm overstating the importance of same-finger, it really is very important as it affects typing fluency, speed, accuracy, and strain. Words such as "wryly" require pressing 5 times with the same finger. Moreover, designing for low same-finger doesn't have to conflict with any other design principle. Colemak, Capewell, Klausler's, Maltron, Arensito all achieve very low same-finger. Arensito is also designed for maximum combos like ToB, so there's no excuse to design a modern keyboard layout with such high same-finger ratio.

Example of words that are uncomfortable to type due to low hand alternation: everywhere, excluded, rescheduled, structure, imagining, exceeded, insignificant, executed.

You perform the finger balance calculation as if 'X' position is as easy to press as the 'S' position. Again, you shouldn't put the letter 'D' on the 'W' position.

I'm not saying you should move the apostrophe, but don't put the I (I'm, I'll, I'd etc) or S (what's, it's, etc.) on the pinky finger.

I did try it a bit, and everything I tell you is because I noticed it while trying it out. The pinkies work too hard, the pinkies do too much same-finger, same-hand jumping combos such as "BU" cause strain, large distance same-finger combos such "BY" are difficult to type, I miss O on the home position, D is uncomfortable to type. Typing 'the' is nice, but I don't think it's worth all the other sacrifices you made in the layout.

Similarity to QWERTY is very important in the learning stage in order to make learning easier and faster. Once you master a new keyboard layout, you'll often start to forget how to type in QWERTY. People who have been typing in Dvorak for a long time and need to use QWERTY for one reason or another find themselves almost unable to type. If for example, you move the Shift keys, it would be much more difficult to adjust from ToB keyboard to a normal laptop keyboard, for example.

It's a QWERTY world after all. I used Dvorak for two years before I gave up on it. I had to use QWERTY some of the time, and switching between layouts was very error prone and difficult. I find it much easier to switch back and forth between Colemak and QWERTY because it's much easier to hunt keys when I'm typing on Colemak.

Perhaps part of what makes it more difficult to switch between Dvorak and Qwerty is that Dvorak is very asymmetric. Once your brain has adapted to the left hand doing all the vowels and that about the only thing can be done in rolls are diphtongs and consonant clusters, it's harder to go back to Qwerty. But I don't know.

For me, personally, the easiest changes to learn are 1. moving the keys up or down 2. mirroring them to the excact opposite posititon 3. swapping similiar letters, like P and B. The hardest are those that move keys from the stronger two fingers to the weaker ones and to make it even worse, flip hands while doing it. And by hardest I mean those that I find it disgusting to accommodate to my nervous system, not the ones that are take the longest to hunt or whatever.

But I'm not that familiar with the long term effects on maintaining “fluency” on multiple layouts. If you must move some keys, is it better to just try move them close to the original positions, or away from them, to avoid mix-ups? Then what again is close, and are there some kinds of changes, despite being easy to learn, that would melt down in your mind over time and often cause mix-ups? Like that Dvorak Y and T thing you mentioned.

If languages are to trust, learning another, similiar language would be easy as long as you're fluent with the first one. Also at first the hardest to memorize would be words, but in the long run what would take the most effort would be learning the structures, sentence formation and all the other more abstract aspects of thought that drive the upper, actual levels of speech. Interference is also so much more common in these deeper areas where it is often hard to draw excact lines between the two languages.

Is there a point to make out of this all? Even though it would take time to learn a thing at first, it may only mean that it's this simple “upper level” detail which won't bother you much in future. The most difficult to get used to are changes at the deepest, possibly even invisible levels where even small alterations can cause complete structural collapses. But which are the smallest detail-like aspects in layout memorizing and which the deeper ones which we rather just leave to intuition, that I would like to know…

Punctuation keys are ridiculously easy to learn anyway, even though they do require a bit getting used to, and I can go back to the Qwerty way with no problems.

You perform the finger balance calculation as if 'X' position is as easy to press as the 'S' position. Again, you shouldn't put the letter 'D' on the 'W' position.

You're correct, the keys should be weighed differently in the finger balance percentages. I still fail to see the problem with the D position though, but it is probably just me then. I have put keys like M, F, P or G there in recent versions as a compromise.

I however find the the Qwerty U position, where you have L, to be even worse than Qwerty W. Pressing it almost always requires a long stretch of the index finger and forces the other fingers to be turned a bit outwards as well. After it pressing other keys with the same hand is quite uncomfortable. It also makes the index fingers roll around the center a lot more than would be necessary, resulting in an increasing actual finger distance. Placing uncommon characters on the top row would greatly reduce index finger stress and distance…This was one of the reasons I started modifying Colemak in the first place.

Same finger is definitely a bad thing, especially with the weak fingers. e.g. try to press QZQZQZ and see how tiring it is. When typing in high speeds (>80WPM), same finger becomes one of the most important factors for speed, and combos/rolls become much less significant. Every time you have to hit a same-finger, it slows you down and interrupts your typing stream. I really don't think I'm overstating the importance of same-finger, it really is very important as it affects typing fluency, speed, accuracy, and strain. Words such as "wryly" require pressing 5 times with the same finger.

There are certain bad combos which slow down typing almost as much as same-fingers. Try something like RAT vs RTD on Colemak. Those should also be eliminated, the roll direction should be kept towards the center, Qwerty positions should be respscted, combos and hand alteration should be optimized; scores under 2.0 are impossible without sacrificing anything.

Moreover, designing for low same-finger doesn't have to conflict with any other design principle. Colemak, Capewell, Klausler's, Maltron, Arensito all achieve very low same-finger. Arensito is also designed for maximum combos like ToB, so there's no excuse to design a modern keyboard layout with such high same-finger ratio.

-Colemak is an almost Dvorak-like layout which almost never keeps the fingers in a flowing motion and is far from symmetric
-Capewell (close) and Maltron are not Qwerty-like layouts, have a bad finger balance and as far as I know have more bad combos
-Arensito is designed for ergonomic keyboards with even rows (it places three keys on ring and middle fingers), which makes layout designing a lot easier, and it isn't Qwerty-based
- The fact that Klausler couldn't find a layout better than Dvorak already shows that we should completely disregard him

The pinkies work too hard, the pinkies do too much same-finger

If the apostrophe thing is let aside, the only difference in pinkie usage between our layouts, is that in ToB the comma is added on the left pinkie. The right one still has more stress, if shift presses are not counted anyway. Maybe you just found it hard to get used to the new location of the punctuation and felt it as strain at first… The way comma and dot are pressed should put more strain on the wrists more than on the fingers, so once you get used to it, it shouldn't stress the pinkies as much. I will change the punctuation back to normal though if others report strain as well.

same-hand jumping combos such as "BU" cause strain, large distance same-finger combos such "BY" are difficult to type

Since B is about as close to both hands, it is supposed to be altered between them depending on the following / previous keys. This is something I learned to do on Qwerty anyway. This should solve the issue of uncomfortable stretches and same-finger typing involving it. Moreover, Y can be pressed with the middle finger to avoid same-finger typing, much the same way you'd likely type GR on Qwerty.

I miss O on the home position, D is uncomfortable to type. Typing 'the' is nice, but I don't think it's worth all the other sacrifices you made in the layout.

I think I'll have to move the O down in future versions, which will probably make my layout more like Arensito or Capewell. Only better. :P I'll also stop making compromises for the other languages. I've got the alternatives pretty much thought out already, even though I have to read something about how the memory works before moving further… But even before that, I'll take a little break from keyboard designing, spend more time with my friends and fulfill all my other New Year's promises (here we actually make promises instead of wishes), finish Twilight Princess and the list goes on… But I'll be back anyway to kick all your other designer asses and actually release the layout someday.

Finally, I like the first version better, but here's my current one anyway: (less same-finger, much more bad combos)

,GUPWJKYH.
ASITDRNELO
ZXCVBFM:Q

Happy New Year to everyone and congratulations to Shai for winning. ^_^