i wonder what the app team is working on.
though it's also partially my fault with past prioritization, the app still feels like a less polished and incomplete replica of old twitter. the genuinely novel parts (feeds, labelers) feel poorly bolted on. and the default feed is still not good enough.
i try to imagine i’m directing an intern.
i give it manageable isolated tasks, try to separate research from execution, restart when it stops doing good work.
i don’t micromanage the code but i’m very picky about the layering, and heavily steer towards constraints i want it to satisfy.
RE:
) so that it doesn't wait for the whole stream to resolve before enabling stream step through
kind of a basic thing to have missed, anyway it's fixed now
ahhhh spent a few days doing super complicated prototypes that didn't feel right and then a throwaway two sentence "how about we try this thing" prompt gave me the first good one
yea honestly i tried reading about publishing lexicons in the official docs and i felt like the docs just want me to reverse engineer whatever they're saying back into human language. hard pass. same feeling about permission sets docs.
RE:
