Text is fine, in fact most of our knowledge transfer is done in text, but real-time collaboration is different. It needs both more density of information, and more degrees of freedom. Pointing, gestures, facial expressions, images and much more go into communicating effectively. Much of the communication with AI still lives in the “text wall” space. I remember getting AI to help with coding by pasting and copying between the chat and editor. On hindsight it’s obvious it isn’t the best way to interact. I think the same is true for many other applications. It’s curious that (mostly) only applications that operate with text moved into editing a file. I like the idea of a multiplayer canvas. Editing persistent objects, [CRDTs](https://en.wikipedia.org/wiki/Conflict-free_replicated_data_type), immutable linked versions, etc. [Creating is about making decisions](https://diogosnows.substack.com/p/good-software-is-a-collection-of) (ideally good ones), I want to work with decisions as my object, I want to see related decisions together on a canvas, interact with them. I want my [[Organic software|software to grow organically]]. I want to be able to explore an idea and come back where I was. A wall of text, a sequence requires committing to a direction, even if you can fork your conversation, those branches can’t meet again. **Ideas are a directed graphs, not trees.**