failiaf

https://web.archive.org/every.to/source-code/my-ai-had-alrea...

oh! i mistook 'dm' to be 'dm mono', but this appears to be correct

(unrelated) what's the font used for the cursive in the article? the heading is ibm plex serif and the content dm mono, but the cursive font is simply labeled as dm mono which isn't accurate

junior engineers aren't hired to get tons of work done; they're hired to learn, grow, and eventually become senior engineers. ai can't replace that, but only help it happen faster (in theory anyway).

this is where the advances in collaboration come in; it's a way out for them because existing editors are largely stuck with git and are patching ai on top of everything.

zed came at the incredibly (un?)fortunate timing where they were just able to build a solid base before the editor wars began. their only path now is to fully maximize the few advantages they do have:

* a fresh base that is far more flexible

* really good experience with performance, design, general craftmanship

* a buzzy community and fresh/boldness that attracts vcs

for zed to truly win (at least in sequioa's eyes) they will need to completely take over vscode as the new default, and that will require a big lead when it comes to collaboration and ai

for those who just caught the headline, the real gem here isn't the investment, but what zed's doing different: rather than using git for version control, they're announcing deltadb which incrementally stores all changes as crdt [1] while being interopable with git.

this feels pretty important; git is definitely not the right primitive for version control with ai and that pain is obvious with existing solutions. zed seems to be going all in on collaboration with realtime, git, and now something in between and it'll be interesting to see where they end up-to me three solutions feels overcomplicated but that may be necessary given how teams work now.

- [1]: https://zed.dev/blog/crdts