It's not super complex. I ended up just modifying the locations of the layer toggle keys. In the default Miryoku layout, in order to switch the keys to a different layer on the right hand you need to hold a button on the left hand. I found this to be annoying since some actions like entering and using a navigation layer can be done on 1 hand.
I just got a toucan. The touch disc on it is great. Having a pointing device in fingers reach makes it appealing enough for me to consider upgrading from a corne.
What started as a joke a few years ago has actually turned into really good signal. I've found that the engineers who care enough to invest in keyboards like this spend a lot of time investing in their tooling and are extremely productive.
Some people like to over-optimise everything. Window manager, vim config, unix tool choice, split keyboard, DVORAK layout, mechanical keyboards, coffee brewing, Obsidian note-taking/Zettelkasten, mice (the rabbit hole for mice goes as deep as keyboards)
This is often more about enjoying the process of optimising than wanting to be productive overall. Some may spend a lot of time reading Hacker News to "keep up with new tools" and clipping their productivity bonsai tree at the deteriment of actually getting work done. They may be the type to spend weeks optimising a command that is run once a year. They may obsess over pointless details that don't matter.
My keyboard obsession definitely turned into a rabbit hole. Once I got into the kinesis split, I immediately realized the world of wanting to customize it. This led me to figuring out soldering/desoldering, and over the past year figuring out how to do a DIY build.
I'm pretty busy, but I've tried to find 2/3 hours a month to progress on it, and keyboards feel like the type of "investment in my craft" that is worth that kind of time for me.
> the engineers who care enough to invest in keyboards like this (1) spend a lot of time investing in their tooling and (2) are extremely productive
I think (1) is true. Whereas, (2) may be less so.
Or at least, "smart but unproductive" is also a class. :) (And I'm sure there are those who have had bad experiences working with such people).
I suppose using a keyboard like this is an expensive signal. As in.. it's fairly easy to buy a typical mechanical keyboard, but more difficult to get one of these small split keyboards. -- But I think this is just "interested in technical excellence", which is somewhat different than "highly productive".
;) As for these keyboards? The most pragmatic & superior tooling part isn't the "36-key keyboard" so much as "each thumb has 2-3 keys" each. That's what allows these keyboards to expressively bring the full functionality of the keyboard to within reach of the hands on home row.
I think 3 thumb keys are too much as the thumb is slow and awkward to move. You can easily get by with 2 and you can get by with just one for normal usage.
I put emphasis on splitting spacebar into thumb keys just because the large spacebar is so ubiquitous. -- There's practically no downside to splitting the large spacebar into 6x keys.
I think it's good if layout enthusiasts want to use as few as keys as each finds practical. -- I think a good keyboard will support layouts that bring the full functionality of the keyboards to within the hands on home row for two-handed typing (as well as support mouse + keyboard functionality).
But .. I think that the keyboard is physically better off with fewer keys is just fashion amongst enthusiasts.
You hit the nail on the head with the 2/3 thumb key bit. That is what was such a game changer for me with the kinesis. all the sudden you have real estate to take a layering approach that you just can't with normal keyboards.
Smart but unproductive is a class. We've all had experiences with those types of engineers. I think startups generally weed them out though. It's hard to survive at a startup without being productive. I probably should have put that as a disclaimer up front.
By "expensive signal" I mean more like "hard to fake the enthusiasm" rather than "spent a lot of money".
In terms of monetary cost, the DIY ones can be quite cheap. -- But it's still going to be more expensive & effort intensive compared to just getting a typical mechanical keyboard.
We decided to open-source Nuon after almost 4 years of operating as a closed-source company. As BYOC has become a more standard way to deploy software, we realized that critical infrastructure like we're building belonged to the community.
Our hope is that by building in the open, we can help create standards for BYOC deployments. Some of the hardest challenges we've found are around security, day-2 operations and supporting different customer environments.
I've been thinking more and more how modern coding agents are best when you can use them natively in your editor.
You as a developer have more context and in turn can give your agent more context. I think seeing developers not using editors anymore is going to turn out to be an antipattern.
I've been updating amp.nvim to let me drive it without ever leaving nvim and it's really good.
I will also add, part of the reason I'm most bullish on amp at this point is because their architecture allows interacting with amp from different contexts + threading is better than anything else I've seen so far.
I wrote some thoughts on Bring Your Own Bucket, the evolution of S3 and why I think that blob storage is the default storage layer for any software deployed into customer environments.
we wrote about how fx + fxtest has helped us do integration tests in a fairly large code base (200+ api endpoints, ~500k loc) with a small team