I have a Georgi keyboard (mentioned in the article) which I started getting to grips with then put aside during a way too busy period at work. This has made me want to dig it out and try again.
I’m also thinking that two of those faunchpads put together on the long edges could make an even better ‘back of phone’ keyboard (or back of mini tablet) for a handy portable note taking device
Is this an interesting goal? I absolutely understand why it would be extremely valuable for most people to know their local sign language for communication with those who can't hear, but I fail to see how wider learning of a very special writing system (one that anyway can't be currently read from electronic devices) would impact the sight-impaired community.
Wider adoption could conceivably improve these interfaces through more users and commercial interest.
Parent didn't say that. Do you really think adding that ending is making your comment more convincing?
The discussion above was not about hand writing (embossing) braille. That is a separate consideration, and I have personally never sent a letter to a friend of any kind, so not being able to send a letter to a blind friend doesn't seem like such a problem to me personally. If I were in the habit of sending hand-written letters to friends and acquaintances, I do think it would be a nice effort to learn braille in case I ever need to send such a letter to a friend.
My second favourite mobile-device keyboard was that of the Ericsson T28  from about 20 years ago. At some point I accidentally discovered a feature I just found outlined at the bottom of page 28 of the T28 user manual , where the volume rocker switch on the left side would act as a text entry modifier key, so that pressing the number 2 button normally resulted in a B, holding the volume up and pressing 2 gave a C, and holding volume down and pressing 2 gave a D.
I remember getting so, so fast and so comfortable at texting with this method. By random chance, an acquaintance a few years ago in a co-working place had an old T28 that still worked and let me play with. Naturally the first thing I tried was the old input method - and it was like my muscle memory never left. I was so fast and so comfortable with it. She was blown away... after 20 years she still had no idea that that mode existed.
The SonyEricsson M600 series had something similar. The keys were rockers. Press the left side of a key and it was an Q. Press the right side and it was a W.
Interesting to use, and I wrote a lot of text on it while on buses and trains. But the keys were necessarily very hard and so your fingers would hurt after a while.
After years of shoulder, neck and back pain I discovered that the root cause was the standard keyboard design. It causes rounded shoulders. Perhaps my shoulders are just too wide. This pinched all sorts of muscles, tendons and nerves. Doctors and physical therapists just failed to identify the issue. Weight lifting didn't help all that much, even with a strong back I had to scrunch and round the shoulders to type.
I finally picked up a second keyboard, and used one for each hand. It was surprisingly easy to convert to. Just added some larger rubber nub things on home row.
It's been life changing. Years of pain was gone in days. No special design necessary.
Each keyboard is arranged at an outward angle, distanced so my shoulders are in a natural position.
I can't recommend this enough.
On Mac it required a system preference tweak (Karibiner app) so each keyboard can share shortcut keys.
I could buy an ErgoDox but it feels too much like a status symbol rather than a working tool.
You sick bastard. Seriously, though, it's most cool that you came up with such a simple solution. Hat's off, and glad you got that pain under control.
I started using a keyboard heavily around 40 years ago and knew from carpal tunnel problems as a pianist that I'd have to be careful. My kids love to joke that my posture looks like it's from an ergonomics manual, but no carpal tunnel since 1982!
(I'm saying my approach is better than yours or would have helped, just empathizing because I would be super miserable if programming caused physical pain.)
There's also tons of split keyboards available now.
The UHK can be mapped to a proper layout easily (it has proper widths for modifier keys on the right, so you just need to move the Fn somewhere else and you can get a standard set), but it's not cheap.
I keep thinking occasionally that some things that I do with the phone could be done with a small one-handed wireless remote/controller and audio output—so I could trigger them while doing something else, e.g. walking outside. For example, Anki already supports either a specific remote or bluetooth remotes in general, and has TTS. Also, there are games using audio as output, which might work in a similar manner—may be interesting to try, if not very engrossing.
Regarding your remote-controller idea, check out the Rivo: https://rivo.me/en/index.html
Have you considered a gamepad designed for phones? I use a GameSir T1s for my Anki reviews while out walking.
Only downside is the screen is lopsided in the controller due to the volume buttons, but this could be fixed with a more expensive gamepad which wraps the sides.
So I was thinking of using something in this vein:
Might come in handy in more occasions than just walking. E.g. home chores when I still have one hand free—currently such time is occupied with more passive podcasts and audiobooks.
A lot of med students use the 8bitdo Zero 2 with Anki one-handed, but I believe a standard bluetooth clicker would be more ergonomic as it'll be designed for one-handed use.
For audio cards, you typically want a controller with 6 keys: 4 answer buttons, replay audio, and undo.
Native key remapping will be available in AnkiDroid 2.16. AnkiMobile (iOS) supports it, and there's various programs/addons for the desktop version.
There's a fair number of reddit threads with further opinions which should be helpful
BTW, perhaps this could be of interest in regard to this topic: I previously was able to use Tasker to remap headphones' control buttons to keyboard keys, and thus trigger AnkiDroid's buttons. However, this turned out to be pretty awkward for me—but it's a generic mechanism for remapping, so may come in handy for someone who has other not-natively-supported inputs. (Though it will likely require additional paid addons for Tasker, namely AutoInput.)
I didn't have the time to get Bluetooth headphones working this release cycle, but it'd be great to do so if it doesn't require too many permissions.
Bluetooth headphones go through a different API than physical buttons/controllers/keyboards, and Google's changed the APIs a few times. At first glance it seemed like a minefield (either managing a media session, or managing Bluetooth devices, in both cases which we'd be competing for Bluetooth access with proper media players which have a legitimate want for the buttons).
Anyway, I digress. Thank you for the pointers, and I'll have another look when there's less maintenance work to carry out.
Looking through some of her other videos I found this demonstration of camera-to-speech in the iPhone pretty awesome–I had no idea that this was a feature: https://www.youtube.com/watch?v=8CAafjodkyE
Personally I only used recognition with photos a couple times, to identify some things. Now, that right there is a power user of the feature.
How does the phone even process the images that quickly? I was under the impression that generic models to recognize a wide variety of things require beefy processing and plenty of memory or disk. Or, are latencies on mobile networks that low in the US or wherever she is? And, do people really use mobile internet all day long—especially transferring dozens of photos?
P.S. While we're on the topic of magnifiers: MacOS has the feature where the onscreen magnifier can be shown temporarily with the keys ctrl-alt, and follows the mouse. I have rather moderately poor vision (so far), but I'm using this quite often to gawk at smaller things on the screen, instead of bending myself closer to the monitor or trying to zoom the webpages. This especially works wonders with hi-dpi screens, where zoomed-in areas just have the old-dpi resolution—so I really can see small details in images, as if having separate images of those parts. With landscape photos, the effect is great.
Especially when PS Vita has a rear touch pad, on which extra functions are mapped in some games—and I can't tell which areas have the mappings, resulting in me either pressing the areas accidentally with my grownup hands, or missing them when they're needed. Or when Apple adds the touchbar instead of keyboard buttons, with none of that magic.
Sadly it seems you can’t turn individual accessibility features on in iOS
I love (in theory at least) the concept of being able to work from literally anywhere, and the concept of some sort of smart display (glasses, lenses, etc) but having a physical keyboard to type out code is always a major barrier.
Something like this starts to edge towards that vision which is cool.
I wonder as well if there's a way to have some sort of bracelet (but probably something less comfortable) that can track the way your hands move, so you'd be able to just "type" normally with your hands (eventually with less actual movement and more intent I suppose) to make use of these wearable displays. No idea how feasible it is, but it feels like it is!
One day I'll be able to work whilst walking around in the forest... One day!
To be frank, this sounds thoroughly dystopian to me
Outside of that time, particularly on weekdays, I'm generally pretty busy with house and child care. That means that my time for just walking in the forest generally has to be crammed into the weekends. Usually on no more than one day per, since, on the weekends, even if I've got nothing else keeping me busy, I've still got to convince a couple other people who would much rather just play in the yard to come with me before I can go. And then, even if we have done that, I still can't really daydream much, since I've got to pay attention to and manage them.
Not so much the work time, though. A lot of it is spent sitting and thinking and reading and taking notes. The only thing about those tasks that's tying me to a place is the awkwardness of doing them while ambulatory, and that's really just an equipment problem. With the right tooling, that could happen anywhere. I wouldn't even need to sit or stand still to do it. Which means that I could perhaps get much more time walking in the forest than is otherwise possible. I could perhaps even make it a daily thing rather than a once-a-week-if-I'm-lucky thing.
I'm in a very similar situation, life is busy. Working at a computer doesn't feel like it should constrain me to a boring office, I could be in nature enjoying the fresh air, thinking about a problem and test out a solution right there.
When my dog was around, he'd have had the time of his life if my office was truly mobile!
There's wisdom to the idea of not mixing business and pleasure, and maintaining work/life boundaries. But I don't think that the principle is meant to imply that a joyless working environment is the secret to a happy life.
I'm glad to see this iteration, as it looks like it is getting closer and closer to reality. He is definitely on the cutting edge of these things in terms of usability.
It works because in Korean, characters are made up of base characters. So you can essentially build a character out of two other characters, and then a syllable out of three of those characters. Then words out of a match of syllables.
Edit: It's 10 keys sorry. Example:
- Steve Mann: https://en.m.wikipedia.org/wiki/Steve_Mann_(inventor)
The funny thing with those little handsets was, with a bit practice you could actually get very quick at typing. Quicker even than on QWERTY keyboards on touch screen phones (at least you had tactile buttons).
Typing on those phones required some practice to get muscle memory but it was accessible to barely anyone that had to use it. Even people who use computer keyboard all their life but are still watching they keycaps succeeded to type SMSs with the phone in their pockets.
I’m sad that we are getting nothing that stands between the old closed feature phones and the mini computers we call smartphones.
I still miss being able to compose messages in my pocket...
But if your goal was to be above the average speed, the probability, even low, of false prediction by the algorithm would make you lose « a lot » of time while with no T9 your error span is scoped to one character.
Plus, for the people who typed without seeing the screen, you couldn’t take the risk to accidentally send the wrong word that would totally change the meaning of your message while an error on one character could be understood by your recipient.
But for the average user who just wanted acceptable speed, reliability and used its phone while watching the screen, T9 was the best for them.
You see the same thing with games. Not so great game designers just copy whats on PC for mobile, and add load of virtual buttons all over the screen. Phones are no good for that, really you want to limit yourself to 3 buttons at any one time at the maximum.
If this could be learned fairly easily, it could be the solution I've been looking for. Of course it won't replace on screen keyboards for everyone, we're too far gone now, but at least there'll be something better availible
Its a rocker-key design: hit the left side, Q, right side, W. The U-shaped key made it super easy to hit the right letter, I could type amazingly quickly on this janky phone from 2006.
Once the capacitive-touch revolution happened, the closest keyboard that really took advantage of the small-screens was the TouchPal T+ keyboard.
It was similar, in that one soft key was Q and W, but you pressed it and swiped left or right for each key, and up and left for capital versions. It also held special characters with a swipe down. This also worked really well, in my opinion, especially on the small screened devices we used to use.
This is all a long winded way of saying, I agree, I think straight QWERTY is still odd, even today!
Worked a lot like a good old phone numpad, just faster. Crashed a lot.
This happens with technology all the time. When the printing press was invented the first books were designed to look like their handwritten counterparts. Similarly, when electronic maps happened they were just like paper maps with fixed scales ("zoom levels"), north up, top down etc. It takes a while for people to realise the potential of newer technology for some reason.
Keyboards are a funny one, though. People don't learn to type any more. In fact, people don't really learn to do anything with computers. Millions of people interact with computers every day but have never really learnt to use them effectively, and this goes for both hardware and software. If you want to be a lorry driver, you need a driver's licence. But people get employed all the time to work on computers and nobody ever asks "can you type?" or "can you produce typeset documents to basic quality standards?" Walk into any office in the world and you'll see hunt-and-peck typing and untrained, misuse of common software like MS Word.
The problem is interfaces like a keyboard and MS Word do allow you to do something without any training at all. There would be huge resistance to having a "better" keyboard that required users to learn how to use it. People have been conditioned to expect no training.
A big driver in the trend towards no training is that the powerful players in tech (ad companies) have an interest in getting as wide a userbase as possible. The manufacturer of a locomotive or MRI machine does not care about this and it's fine and expected to need training to operate those machines. I wouldn't expect keyboards, especially those on a phone, to move to anything that requires training any time soon.
Google Maps (any many map systems even today) used pre-rendered tiled images, which means you could only zoom at the levels which were rendered and rotating the grid was hard (especially on old hardware)
The classic pre-smartphone phone keyboard, where you write in pinyin and the corresponding Chinese characters get predicted.
Animated demonstration: https://media.giphy.com/media/gjOWCOlhd98wI4dYcd/giphy.gif
An anecdote: Early in Steam's controller support, they used a really nice alternative keyboard, where you picked one of 8 directions with the analog stick, and pressed a face key for the specific letter. It was fast, convenient, and built for controllers.
It didn't last.
Steam shortly moved back to a standard QWERTY keyboard layout with the joysticks moving virtual fingers. If I had to speculate, I'd say that they received too much negative feedback, because people didn't immediately recognize it as a keyboard, and didn't want to learn something new.
Even if it's slower, virtual QWERTY keyboards are (almost) universally familiar, and there's nothing new to learn.
I think a qwerty keyboard makes sense as a default since that is what people used to and it is hard to teach people new things. But there are alternatives for people wants to try them
With Latin's smaller alphabet, you can even have the best of both predictive text and precise input: maybe have the plain tap on 1 be a wildcard (a|b|c) for the predictive engine to infer
Add to that the sad story of the QWERTY layout: proven to be inefficient and yet still widespread.
Wanted to use https://software-lab.de/penti.html but have difficulty learning all the chords.
the irony of the typo in your comment is not lost :)
I make a lot of typos on regular keyboard as well.
Each letter is a two-button chord, but in practice, each letter is an arpeggio because the Twiddler only sends a key event when you release a button. The layout is designed so that when writing English, there is a very high probability that the second button you press for a letter will be the first button you press for the next letter, so you can keep that button held down, press the second button, then release the first button to send the next letter. Your first three fingers (it doesn't use the pinky row) will "walk" around the three rows of the Twiddler.
This is kind of making me want to pull out my Twiddler again, I really haven't used it since I gave up on trying to run Illustrator on a Surface tablet instead of a Mac laptop.
I believe this idea can have potential for breaking the world record because there is no finger-to-key travel. But maybe it is diminished by the having to press a key multiple times over to get to the target just like in old cellphones .
Alternatively maybe one could brute force the problem of optimal keyboard layout/etc using simulations. Like that robot (human) hand that solves rubik's cubes by learning through simulation (apparently, from OpenAI ).
I tried to create something similar using web audio and a midi piano a while ago but the reason this works is you are applying digital processing to analogue output, along with Ólafur being one of the most talented composers alive at the moment.
It's actually a clone of another keyboard, 8pen, which was abandoned. Here's a video of how it works:
Nice write-up OP! Excited to see where you go with Peggi :)
What's nice about having the full keyboard is that you can keep normal keybindings and just not use them when you're feeling lazy. I'll have to play with this more, it's a really good idea.
If you haven't come across Doug Engelbart yet, I highly recommend it. He invented the mouse (yes, that mouse), and was a staunch believer in chorded keyboards.
I'll admit, though, I don't really get it.
I just bought the hardware recommended in the fine article, let's see how that build goes!
This is one extreme. The sweet spot is in the middle, using tap/hold keys on a more conventional keyboard as supported by QMK firmware. It's brilliant to keep your fingers near home row. It's dogmatic to not have the other keys even be there, or to force yourself to use one hand if you have two.
A finger of one hand holds a key to select a mode, a finger on the other hand taps a key using that mode. This mode selection layout should be mirror images for the two hands. The Matias one-handed keyboard established beyond a doubt that our brains are wired to support this.
I use custom modes for punctuation, exotic "control-xylophone" editor commands, and controlling my computer. And yes, a cat walking across my keyboard will rearrange the orbits of the planets. However, the decades I spent thrashing would have been all worth it just to realize I shouldn't have to reach so far from home row for shift and modifier keys.
For example, you can put shift keys under your pinky fingers:
(defsrc a ;)
(deflayer main (tap-hold-next-release 300 a sft)
(tap-hold-next-release 300 ; sft))
and I think Android supports something similar.
I love typing on a hardware keyboard, I once owned an HTC Desire Z phone, which had a hardware keyboard and it was pretty great. Too bad nothing viable with a keyboard appeared since that time.
This guys on the right path I think, and big props for taking it as far as he has done.
Good post, but even better design thinking and execution. Thanks for sharing, and good luck on next steps!
(Unfortunately the ;+A didn't work on Safari for me on an M1 MacBook Air).
I wonder if there are benchmark for swipe versus normal keyboard...