deskthority

facetsesame wrote: Half height convex mods (as on the Wang 724 and beam springs) are nice, incentivising use of the thumbs while being distinct from the space bar.

Not all IBM beam spring keyboards have half-height convex modes. Some keyboards have full-height mods. Some have both high and half-height keys on the bottom row.

The Wang keyboard's keys look almost as reverse space bar profile to me.

facetsesame wrote: If the toolmaker me was intent on convex throughout, I'd like to see what the numpad vertical enter would look like!

Something like this pre-chiclet Apple USB keyboard, I would recon... The "sphericalness" looks pretty mild on these..

Another option would be a hybrid, like this Shift key on a µTron keyboard (space bar row, cylindrical alphanumeric keys): On this and older Tron keyboards, the thumb keys are not so rounded but more flat with rounded edges.

Great point. The µTRON's caps are damn well delicious:



And well remembered about that Apple board. The very beast that finally made me notice what I was typing on simply by how atrocious it was. Wasn't 100.0% bad, with sneaky convex mods like those. Although that sharp polycarbonate edge in front of them… full of insects and spoo.

Findecanor wrote: I press both Alt keys with my thumbs. But hey, all concave would be better than all convex: The arrow keys are not used with the thumbs - the hand is usually moved down a bit to use them.

The arrow keys should absolutely not use the same profile as the rest of the keys on the same physical row on the keyboard, and when keyboards use those shapes for the arrows it’s incredibly stupid.

One reasonable choice in my opinion is for the up, right, and left keys to use the same profile as the home row, while the down key uses the same profile as the bottom letter row (ZXCV row).

jacobolus wrote:
One reasonable choice in my opinion is for the up, right, and left keys to use the same profile as the home row, while the down key uses the same profile as the bottom letter row (ZXCV row).

I completely agree. But each to their own.

Indeed. My own arrow key profile favourite being this:



I'd love a convex bottom row across the alpha block of the board, with concave arrow keys like that. I never thumb them, I always use them with the three fingers of my right hand. A cupped shape like that makes the arrows really stand out by feel, and they're comfortable to work on. My thumbs are for chording, which I don't need to do so much with only 4 arrow keys!

How about convex mods in row 6 to start with and utilizing row 4 which is fairly neutral for concave mods where desired? If the row 4 (concave) keys are flipped the front edge would present lower and the overall height would also keep it just below the space bar. Developing the additional widths in row 4 would probably a worthwhile pursuit anyway since there will likely be users of customs like the Ergo Dox that could make use of them as well.

I like that idea very much.

@rsbseb To use a metaphore: If some people propose black, and some white, please don't end up with grey. Stay with black or white.

@Eszett My proposal was meant to be a possible solution for providing black and white, um I mean concave and convex

Really the problem is that the standard keyboard layout sucks. Keyboards should be split and should use a column stagger, ideally with height adjusted to match the different lengths of different fingers, and there shouldn’t be keys two rows below the home row, three rows above the home row, or three columns to the side of the home pinky keys. Ideally, there should probably be about 3–6 keys for each thumb, 3–4 keys for each middle and ring finger, and 3–6 fingers for the index finger and pinky, and that’s it. 20 keys on each hand is sufficient. 25 keys on each hand is great. 30 keys on each hand (with 5–6 for the thumb) is still okay. One giant key for both thumbs and 10 keys for each finger is ludicrously bad design. Both ANSI and ISO have terribly placed right shift, backspace, escape, backslash, and about half of the number row. ISO left shift and return are also terrible.

So anyway, there’s inherently no good profile for keys like the Control key on a standard 101/104-key layout. These keys are in a terrible place to strike with either the corner of the palm or any finger. If planning to use a finger, the profile should be similar to the bottom letter row. Personally my advice is to just avoid keyboards that place keys in that position, or if using such a keyboard, remap caps lock to control and ignore the physical control keys (or maybe use them for additional rarely used layers). Likewise, a spacebar that extends to the left of the QWERTY C or to the right of the QWERTY M is a terrible design, and bottom row modifiers which are as far away from the center as QWERTY A or semicolon are quite unpleasant to reach with a thumb, and hard to reach with a finger.

@Jacobolus, while I agree that existing keyboard norms do suck, I feel like that is an outlier position. They are norms for a reason and short of changing the whole goal of this profile / keycap endeavor I think we are barking up the wrong tree to try and change those norms. Perhaps we can make a note to re-explore layouts when we know what rsbseb is capable of after caps for the existing norms have been vetted for. This is a very flexible project at all, but I think we need to help him get to an acceptable level of quality before we start moving the goalposts.

@jacobolus Can you give a sketch of what you consider an approximately reasonable layout? Maybe via http://www.keyboard-layout-editor.com ?

jacobolus wrote: Really the problem is that the standard keyboard layout sucks. Keyboards should be split and should use a column stagger, ideally with height adjusted to match the different lengths of different fingers, and there shouldn’t be keys two rows below the home row, three rows above the home row, or three columns to the side of the home pinky keys. Ideally, there should probably be about 3–6 keys for each thumb, 3–4 keys for each middle and ring finger, and 3–6 fingers for the index finger and pinky, and that’s it. 20 keys on each hand is sufficient. 25 keys on each hand is great. 30 keys on each hand (with 5–6 for the thumb) is still okay. One giant key for both thumbs and 10 keys for each finger is ludicrously bad design. Both ANSI and ISO have terribly placed right shift, backspace, escape, backslash, and about half of the number row. ISO left shift and return are also terrible.

What do you think about the Datahand?

Will there be catering? It'll take a while to describe.

jacobolus wrote: […] ISO left shift […] are also terrible.
[…]

Interestingly enough, millions people use ISO left shift without even noticing how terrible it is.

In the same vein, have you ever noticed there are doors you actually have to open with your left hand?
Now that is terrible!

Opening doors left-handed is much less horrible than ISO left shift.

Hear hear. In fact, I often as not use my left hand on doors. Don't know why. I'm a bit left handed. 100% so when using phones.

SL89 wrote: @Jacobolus, while I agree that existing keyboard norms do suck, I feel like that is an outlier position. They are norms for a reason and short of changing the whole goal of this profile / keycap endeavor I think we are barking up the wrong tree to try and change those norms.

Sorry, my comment was not intended to suggest goals for this project. I was just trying to explain why there sometimes no good choices, as with the profile for some of the bottom-row modifier keys.

Eszett wrote: @jacobolus Can you give a sketch of what you consider an approximately reasonable layout? Maybe via http://www.keyboard-layout-editor.com ?

Sorry to get further off topic answering here. For a keyboard where each half is flat, we’re limited to standard switch spacing and keycap sizes, and the keyboard follows the general stylistic theme of the Ergodox, this works pretty well:

This is a compromise: more keys than I think are strictly necessary, but at least all of the keys are in easy reach.

To me, this is a very exciting development, as I would like see more different kinds of step and hat keys become available.

Here is what I want to build:
Even keyboard-layout-editor.com complains to me about the 1.25U hat keys!

@jacobolus, I totally get what you mean, no worries. Also is that paddle board a working keyboard?

@jpatters, that is rather interesting, is it symmetrical and staggered?

jpatters wrote: To me, this is a very exciting development, as I would like see more different kinds of step and hat keys become available.

Here is what I want to build:

COLEMAK A.png

Even keyboard-layout-editor.com complains to me about the 1.25U hat keys!

My only suggestion might be to let the case take up the excess space on the staggered rows, and leave all the alpha keys 1U. The non-1U keys are going to feel subtly different simply due to balance and weight, not to mention that at some point you may wish to try another mapping on the same physical layout. A feel difference is acceptable when it's Tab or Backspace, not so much when it's just another letter like Q or O or especially A. Basically, just exchanged the stepped areas on those for areas where the case extends inward.

Incidentally, I am a Colemak supporter without actually being a Colemak regular user. Even its inventor doesn't think it's worth switching for someone who already is proficient with Dvorak, but I gave it a spin for a week just to see what it's all about. Although I was still dead slow (~20 wpm) at the end of that, I understood what it was about enough to recommend it over Dvorak for anyone wanting to move away from QWERTY. It accomplishes pretty much the same thing, but with fewer side effects and a less steep learning curve. It also lacks a couple of annoyances inherent to Dvorak, like the difficult placement of the letter L. (I solved that by a different method, going to a matrix layout and lengthening the spacebar vertically, allowing my hands to point inward more because of the shorter thumb reach required. This makes L closer, but also effectively lengthens the little finger because my hands are biased in its favor.)

It was a working keyboard temporarily, but without a bottom case some of the wires came unsoldered, and now only about half the keys work. It worked long enough to demonstrate that the layout is pretty effective though. A couple of similarish prototypes which I don’t have pictures of online are working.

I’ve gotten side-tracked from keyboard related projects to work on other things in the past 6 months. At some point I should really get back to finishing up a proper DIY ergo board or two for myself though.

SL89 wrote: that is rather interesting, is it symmetrical and staggered?

Thank you, the goal was to reduce the width as much as possible and have it still be a standard layout.

Mal-2 wrote:
My only suggestion might be to let the case take up the excess space on the staggered rows, and leave all the alpha keys 1U. The non-1U keys are going to feel subtly different simply due to balance and weight, not to mention that at some point you may wish to try another mapping on the same physical layout.

That is a good suggestion and a problem I hadn't considered. The first prototype would have had 1U keys for Q, A, and O anyway due to the non-availability of caps stepped like that, and looked more like this:
I like to design with hat keys because you can always just use a normal key in it's place, since the switch location will be the same:
Of course, there is a problem with this design, if you want to reconfigure for QWERTY, then you would have to move Enter down a row, and the switch locations would not be correct.

To get around that, here is a version with 0.75U added to each side of the middle section:
Left and right columns are reduced and moved in to compensate for the increased width.

I like the modifier placement better with this design.

With the bottom row modifiers specified as convex, I count 17 distinct key molds would be needed for this. If row 2 and row 4 were the same except flipped, there would be 15. If row 5 concave were the same as row 3, then we're down to 13. If we dispensed with convex keys except for the Space Bar, then we can get down to 11.

That's still a tall order, but any progress would be fantastic!

Looks awesome!

So many possibilities for hat keys:

We are getting off topic a whole bunch lol.

Jpatters is *all* about hat keys. Let him live the dream!

Tbh so am I, the only ones you ever really see are the ones on IBM boards.

SL89 wrote: We are getting off topic a whole bunch lol.

Stray away I'm learning a bunch

In that case, how much of a pain would hat keys be to use. Do they wobble a lot?