geekhack Community > Ergonomics

Glove80: The End Result of 500+ A/B Testing Experiments

(1/2) > >>

OddField:
Here is our 80 key split contoured that is a bit different from all the other contoured in the market. I started this thread as a place to discuss the ergonomics aspects, and to share our long long journey to find the "perfect" ergonomics.







Back story:
Years ago I was at a fun software company full of incredibly talented geeks. Like most geeks, we overused our hands and nearly everyone had some sort of repetitive strain injury. Kinesis Advantage was the best ergonomic keyboard at the time and quite a few colleagues had one. Unfortunately my hands just donít fit an Advantage; my pinky is too short. Talking to others, I found out not all hand shapes fit an Advantage. Because of this, and because we realized many ergonomic aspects could be improved, we decided to make a better contoured keyboard that works better for more people.

It started as a passion project with a few friends. The problem however is that we were all perfectionists and we couldnít stop. Many hundreds of prototypes and A/B testing experiments later, we have finally settled on a design that works really well on a wide variety of hand shapes and hand sizes.

The ergonomics were great, but the keyboard was somewhat ugly (okay rather ugly). After all, it was a very flexible ergonomic test rig first and foremost.

So we faithfully kept the ergonomics, and made it pretty and compact. We re-designed it to be manufacturable. We managed to add a few extra tweaks too. Introducing Glove80.

Glove80 is highly customizable. It is wireless (BLE), but can also connect to a host via USB. There are no wires between the two sides. It runs the excellent open-source and highly configurable ZMK firmware.

We have a Survey at https://forms.gle/xjHFfoAt7AxC8K147; it would be great if you could help with a better understanding of what you would look for in a contoured keyboard. Thanks!

Key features:
- 80 key split contoured keyboard
- Reinvented ergonomics - result of 500+ prototypes and ergonomic A/B testing experiments
- Compact and low profile - the most compact contoured keyboard (pro-rata for key count)
- Wireless (no more wires) and USB-C support (only if you want)
- Capable of connecting up to 3 Bluetooth LE devices and 1 USB device simultaneously
- Premium components: POM keycaps (see https://geekhack.org/index.php?topic=114541), Kailh Choc key switches, Nordic nRF52840
- Tenting and flexible mounting options
- Detachable 3D sculpted palm rest - extremely comfortable and detachable for travelling
- Open source ZMK firmware supporting full programmability
- Supports any key layout
- Supports MacOS, Windows, Linux, Android and iOS devices

Interest check: https://forms.gle/xjHFfoAt7AxC8K147.
Reddit thread: https://www.reddit.com/r/ErgoMechKeyboards/comments/q2s8w2/glove80_the_end_result_of_hundreds_of_ergonomic/?sort=top
Discord Channel: https://discord.gg/cdywSRX9qF

OddField:
Why 80 keys and not the more fashionable 40/60/65%?

Layers is ultimately a compromise. With layers you are trading the number of keystrokes (I.e. increased muscle stress and lower wpm) for better ergonomics/comfort. On most boards it is simply not possible to get 60 never mind 80 keys to be comfortably reachable, so that kind of compromise makes sense. However we tried very hard to make nearly all 80 keys accessible and I would say on the whole we succeeded. Without moving my palm much if at all, I can access 76. I can't reach the top pinky row without moving my palm but I have a very short pinky. Testers with longer fingers can. Not to boast but I doubt there is another keyboard in the market that has managed this feat - even Kinesis Advantage puts the F keys in an obviously unreachable tiny-key row.
 
So in this context, I would argue not having to layer for common keys is a win. Of course ZMK supports multiple layers and we do use it ourselves for the less common things (such as for numpad, host selection actions and any number of other things).

Secondly although it is cool in keeb circles to go 60/65, one of the primary goals of this keyboard is to make life better for people with RSI or a high risk of RSI. 75% and TKL are well established and well-tested to be very functional, without the extensive use of layers. Most of these keyboard warriors with RSI/RSI risk just want the easiest transition path. It is already a barrier going to ortholinear. So no reason to add 60/65 to the list.

Thirdly some people go 60/65 because smaller/cute. I hope you would agree that we managed to succeed too in the aesthetics area despite having 80 keys. It is the most compact contoured keyboard key for key. We have tried to make Glove80 as minimal and clean as possible. No cables. Clean lines. Very thin case.

nevin:
looks very nice. and kudos for going bigger.

OddField:
"Reinvented ergonomics - result of 500+ prototypes and ergonomic A/B testing experiments" - Sounds extremely compelling, but is this true?

Good question. That's enough of a topic for a very long post. I would likely do that in the future, but here is a quick answer.

Is it true? Great question. The traditional process of creating a prototype is exceptionally long. Never mind a non-functional clay model that is all we have 20/30 years ago, which would take a while to build and with very poor fidelity. Even in modern days of 3d printing a full keyboard would take at least 3 days to print. So how did we do it?

We brought a couple of things to the table that I believe is the crux of the real difference of Glove80.

First is the philosophy. We come from the background of software engineering, where it is now standard practice to trust real data and not plain guess. A/B testing provides valuable comparative information Our muscle memory is short. We can't tell the difference between two ergonomics setups if we wait for a few hours. However our hands can tell the most minute difference if we quickly switch between two. Even 0.2mm or 1 degree plane change.

Second is the architecture. As I said before it takes days just to print a full model. Well that is not going to scale. A hundred models are going to take a year at least to print with a 3d printer, never mind the actual testing time, analysis and redesign. You are right to be sceptical.

So how did we do it?  A good software architecture allows a huge degree of flexibility to make necessary changes without affecting the rest. We take this insight and apply it to the way we build the (clarified: hardware) functional and non functional ergonomic test rigs. Instead of taking days to try a new variation, we could easily try 10 and way more in a day. The best part is we could keep the flow of thoughts and creativity going.

vvp:
So 500+ A/B tests includes software changes. It is not only about comparing different physical keyboard shapes.

Well, I considered the 500 number a marketing stunt anyway. My contoured split keyboard (K84CS) was developed in about 9 iterations as for as the keyboard shape goes. So 500+ obviously looked like pulled out of a marketing department.

Anyway, I like the push for contoured split keyboards with thumb clusters. You are going in the right direction. Though you are still far from the' superior shape of K84CS :)

Navigation

[0] Message Index

[#] Next page

Go to full version