Pinned Tweet
Introducing Spear.
The future of the desktop, 17 years in the making.
75
58
1,128
119,287
I don't think LLMs are going to JIT-design interfaces for us in the foreseeable future because unless the entire concept of software and platforms becomes obsolete (not impossible, but structurally a big stretch from where we are), the variance within a given domain will never be big enough to justify foregoing fully-baked UX.
Much more likely is that some kind of Jev model navigates us to exactly the screen we need, turns on/off the right panels, etc.
You know these Star Trek LCARS interfaces that always looked like they were specifically designed to answer a plot-driven question visually and were never the same, always looked haphazardly constructed in that moment?
I think, yet again, Star Trek understands the user interfaces of the future better than we do in the present.
I think we are a very short time away from applications designing and assembling UI in the moment versus having it pre-designed. Contextual interfaces won't just consider what they present, but also who they're showing it to (on a highly (inter-)personal level) and when those people need which information in their workflow.
1
1
6
1,203
This is a much better riposte to reductive equivalencies between human minds and language models than any of my reply-guy contributions.
If AI doesn’t fill you with awe over just how amazing the human brain/body complex is, then you are really missing the forrest for the trees
Yes, modern AI is awesome, but humans are more awesome
1
2
456
One question about Spear I've been getting a lot is: how close is it to Niri (a window manager for Linux)?
It's an interesting point of comparison, because while they both share the same model of organizing windows along an axis, they are also philosophically somewhat different.
Niri is a tiling window manager that adopts the concept of a continuous virtual desktop. Windows are locked into a grid, with no overlap, and you scroll through that grid, snapping to window edges. I think it's pretty cool!
Spear is actually closer to the standard PARC model, in that it's still largely freeform (in part because macOS is) but within a sort of semi-rigid structure that is repeatedly loosened and tightened based on the user's needs.
There's probably some literature on this that would be interesting to read. The cycle of "everything starts in order -> your workspace gets messy -> you put things away" may well have a name I'm not aware of, but it's foundational to Spear.
wow!! i’m so excited for this, i’ve been wanting something like it for mac. been playing with niri on my linux box and love the experience, cant wait to try this
1
12
1,516
The signup form is fixed! If you tried signing up earlier and got an error, you don't need to try again – your signup was recorded; I just had to buy more credits before it was released into the list.
5
929
Six months ago, I was building a proof-of-concept for a very niche Mac utility when I had a moment of clarity: the technique I was using for this niche use case might actually work for something else.
Something I'd wanted to build for 17 years. 🧵
1
7
1,285
The doubts were, and still are, ever-present. Does it feel native enough? Despite the iteration and dogfooding, will I be the only one who actually likes this paradigm?
It's a different world from the days of 10/GUI. Will anyone even care this time around?
One way to find out.
1
2
138