advocate widespread dissatisfaction with computing. research buzzwords include human–system interaction, knowledge representation, cybernetics/self-organisation, programming systems, ... phrases to poke people with include abolishing apps, objects are reference(s), PLs should be views/lenses, digital language is privatised.
@TodePond happy to, p5 really isn't needed here
@TodePond Hell yeahhhhhh
projecting time into the third dimension in @tldraw
Repo: https://github.com/OrionReed/tldraw-3d
Play with it: https://orionreed.github.io/tldraw-3d/
Bit of a tangent at the end there... Anyways, you can play with the 3d underlay here.
I want a computing where what matters is a kind of connectivity, not "does it go fast" but "can it get there"
When a series of transformations or relations becomes widely used, it can develop into a more direct, 'optimal' path.
(7/n)
I think our computing environments should make available all the things which are (abstractly) valid constructions — pluralism, not a rote notion of efficiency.
In mathematics you can solve an "impossible" geometry problem by translating to topology where the answer may be trivial, then back again. It's not highways, it's a well-connected network of possible transformations.
Computers are poor at this, but they don't have to be.
(6/n)
The interesting thing is not "let's do everything in 3D" but instead noting that we can coordinate 3D space with a 2D environment. 3D doesn't mean FPS camera controls.
Perspective can communicate *some* information more effectively, so why not use it when that's the case?
(5/n)
Sidenote: There's some interesting work on "feedforward" mechanisms by Joseph Malloch, Wendy Mackay, and others in this paper https://inria.hal.science/hal-01614267/document
(4/n)
You can embed time in the 3rd dimension, too:
1. you could show local shape history by projecting its past transformations
2. you could show possible futures by projecting the results of actions, like "this align function will move the shapes in this way."
(3/n)
The examples here are simple, but interesting:
1. Edges which curve in 3D means you can locally approximate their length, no need to zoom out, you can see if they're near or far.
2. By projecting shapes into the distance, you can see what's nearby just from the background.
(2/n)
2D canvases have a whole dimension spare, so why not make it semantically rich? Here's a 3D "underlay" in @tldraw — 2D control plane, 3D feedback.
Depth + movement makes a whole class of information legible.
https://github.com/OrionReed/tldraw-3d
a thread 🧵(1/n)
Imagine interacting with your connected notes with a more engaging canvas view… that is programmable! “Visual-Semantic Morphism” is just one thing that came up in this cool presentation from @orion
A quick PoC for regions/spaces in @tldraw where each can have arbitrary rules/constraints/behaviour for the shapes inside.
Here's 2½ examples:
1. snapping to grid / discrete coordinates
2. graph layout
½. binding to graph and snapping systems at the same time
Thanks everyone for joining #ToolsForThoughtRocks after a long time. Great energy!
@orion gave a great presentation, we have a recording, it'll be up later this week, along with chat transcription.
Thanks to the after party crew for deep discussion!
Let me know if you come up with fun ideas for constraints, the horizontal/vertical constraints are pretty trivial and there's a lot of power in WebCola.
Hit the (⁉️) button for some instructions.
*Known Issues:*
- I managed to break the non-overlap constraint, not sure what happened there...
- The edge-length calculation makes some arrangements unstable, better math needed.
Please let me know if you discover bugs!
I updated my @tldraw graph layout repo with better handling for rotation, arbitrary shapes, alignment constraints, and it's now much smaller and simpler too.
https://github.com/OrionReed/tldraw-graph-layout
You can also get weird with it, as most of the code is parent-relative, you can make scenes in frames and then rotate the world itself, while other shapes collide with the world...
Updated my @tldraw physics repo to handle arbitrary shapes and let you move them around, delete them, control multiple characters, etc… It's also much less code and easier to maintain and extend. It uses a tiny library which I made for tldraw “collections”, reduced the complexity a ton and now the whole physics system is just one file.
I’m quietly trying to make the case for this kind of collection primitive to be added to tldraw itself…
I'm going to be in the (virtual) audience for this one. See you there?
https://toolsforthought.social/@boris/112192422178822491