One of our inspirations for modeling text operations in this way was atjson, a rich text format which stores formatting spans alongside a plain text sequence.
Never seen this atjson library but this is really cool-- would be interested to see what is possible if syncing this syntax with a tool like Cambria.
However, comments behave very differently from colored text — although a single character can’t be both red and blue, multiple comments can be associated with a single character in the text. We can render this in the editor by showing the two highlight regions overlapping:
Also really like how this solution still has an interface that makes syntatic sense given the original comments -- and could branch as more users add comments/data
Bob’s coloring on the word “jumped” could have been preserved without issue. Another option might be to blend the two colors together on the word “fox”, but then we would be creating a new color that was used by neither Alice nor Bob, and that doesn’t seem right either. In our opinion, the most reasonable behavior is: in the region where the two formatting ranges overlap, we arbitrarily choose either Alice’s color or Bob’s color.
I really like this solution -- it defaults to a solution that is random, uses logic in cases where that logic still makes sense, and still offers users the ability to merge branches or revert to older edits.
Makes it so that hopefully anyone can be editing the same piece of content online/offline/in real time together
The same way Roblox allows its users to build their own games and play more than 40 million games made by other players, Airtable can allow its users to build their own SaaS software. Last year, Airtable introduced apps that make the product extensible and opened up their API for external apps.
Doesn't necessarily make them the Roblox of SaaS, just of Spreadsheet-based management. Great to have a tool like Airtable but very difficult to make on-the-fly adjustments with tools like Airtable and Zapier that are built around long-standing background automations.
The RECALL Augmenting Memory architecture. It can, for example, help users restore context before their next conference or class. The student, walking to a lecture, could be primed with a summary of it through his smart glasses, surfacing relevant information. The description of the "Memory vault" in this architecture exhibits a high similarity to Vannevar Bush's Memex.
It's these deep learning breakthroughs that now make a lot of these memex and semantic web technologies accesssible. This is a note I also referenced in the SWTs whitepaper for Cortex. Great to see Moritz and RemNote picking up on this change as well.