I started optimizing for real-time distributed collaboration. After deep design work, I’m pivoting: JJ-style workflows fit LLM agents better. Agents work independently, then fetch/rebase/push. Lower complexity, clearer history, easier scaling.
Yuri Not Igor
@ydewit
- 7 Followers
- 69 Following
- 47 Posts
Real-time matters for humans co-editing docs.
For LLM agent workflows, robust async integration is usually the higher-leverage investment.
For LLM agent workflows, robust async integration is usually the higher-leverage investment.
The upper layers/universes are categorical realities where universe (n+1) is the sparse summary of the detailed concepts in (n).
What if … Intrinsic HNSW?
It is quite interesting that Hierarchical Navigable Small World (HNSW), a graph-based algorithm for high-dimensional nearest-neighbor search, seems isomorphic to the tower of universes in category theory …
@oisdk hyperfunctions - the assembly language (combinators) of control flow!
Anne Applebaum
A "category" is a space of possibilities! Mathematically, it contains every possible object and every possible morphism". We don't compute or generate a category, unless it is somehow finitely presented, but we simply explore the paths within it.
@egbertrijke @ivankobe This thesis is amazing! In a very superficial and limited way, last week I was taking CaTT and “parameterizing” the base type * to be able to talk about CaTT of CaTT… so reading about SCaTT_{w,1} as (ω,1)-categories of (ω,1)-categories in the thesis above made me happy!
What is the underlying reason why all cells of dimension ≥ 2 are invertible? Is that simply because weak cells above dimension 1 are not needed? Tame complexity?
Synthetic category theory ... a prescriptive substrate for computation in addition to a descriptive tool for formalization?