Experience with modern society has shown that "third places" and human interactions get mowed over when you let algorithmic engagement take over. As I understood GP's point, the whole point of the "words on a shirt" was to get together as a group and attempt to form bonds, and by handing the task over to chatGPT they failed to do that.
Technically you can imagine a future where people use chatGPT "in moderation", but in practice they'll use it for everything and spontaneous/creative "hanging out" will suffer as a result.
At least with the formulation Tao had, one unfortunate side-effect is that if you answer with 100% confidence and get it wrong, you basically fail the entire course.
You can't possibly see a use-case for long-lived branches? Say what you will of git, at least it exposes enough knobs that it can mostly accomodate every workflow (possibly with a heavy porcelain layer to hide the plumbing). JJ seems to swing too far in the other direction, great for a "live on head" mentality but less ideal for other setups.
(The fact that all edits are automatically recorded is my personal peeve with JJ. I'm ok with lack of staging area like in Mercurial, but mercurial doesn't try to automatically amend the commit with my pending changes. Sure I can pretend that "@" is my staging commit then squash as needed, but then I also need to remember that checking out a commit is "jj new" which feels like absurd mental gymnastics to me.)
Oh and one more thing I forgot: in git a lot of operations require you to have a clean checkout. Otherwise you have to stash your changes first before doing that operation. With jujutsu’s model, there is no stashing. Every edit is already part of a commit. Therefore you can totally do rebases even when you are in the middle of some other unrelated change.
>Most agent loops reorder, rewrite, or inject fresh timestamps each turn
That's really surprising, since it'd defeat the whole point of KV caching. I mean I buy it considering how sloppily coded the harnesses seem to be, but this like obvious low hanging fruit.
I've also often wondered why LLMs aren't trained with a format of having a dedicated contextual system-instruction role at the _end_, which you could use to put context like current time or other misc stuff.
There are context pruning strategies that will prune old messages that are no longer relevant, and context compaction from summaries, etc. But to say "most" do this on "every turn" is overstating things. I think it's more correct to say that "many" do this "occasionally."
I'm also not sure what they mean about injecting fresh timestamps. I could see why you'd prepend/append a timestamp to the user's messages to make the model aware of the current time, and the passage of time, but I can't think of any good reason to edit timestamps in prior messages. I'm sure someone can come up with one, but I'd be very surprised if this was a thing that most agent loops do, let along doing it on every turn.
i put together this, for myself so i can try to track what coding agents are doing, I add agents to it or topics (like caching, or sandboxing, file editing methods, etc) just to try and find anything novel or good, since I am/was considering making a new harness but using all the best things from any of those. I still cannot find my perfect coding agent, every one of them has some problem or just not totally what it could be.
What I do is just point agents to a folder, have it loop around a few times on a repo, fact checks at the end, but people sometimes think the software/harness around the AI model doesn't do much which is TOTALLY wrong, its probably AS important or more.. file editing methods available matter a lot, context compaction methods... matter, caching matters. I am still fantasizing about a "best of N" coding agent, that tries to take all the best stuff from all of them.
I have an idea of a coding agent that puts a lot more effort into using more than one model at the same time. Sooo much can be done with that idea.. and no one is apparently doing it yet that I can find. I just am not sure I want to put that much time into a new coding agent project. I wonder how autonomous it could be - have weekly or daily scans of the current coding agent landscape and automatic scanning of coding agent/ai code related subreddits/hacker news, analyze it to figure out what the current problems are, complaints about existing coding agents, desires --> prioritized list of possible features/fixes ---> ai agents code and make releases
I thought one of the main issues is that during sleep you lose muscle tone and control, so no matter how much you do myofacial therapy while awake it doesn't really matter much in preventing your tongue from falling back when asleep.
>basic technique of "mewing" is very similar if not identical to something that's been practiced in yoga for thousands of years
Can you elaborate more on this? I also assume by "mewing" you mean putting the tip of your tongue on your hard palette right behind your upper incisor, and letting the rest of the tongue suction against the hard (and soft) palette. Surprisingly for something that seems to be so "popular" in "broscience" I find it hard to find any canonical definition or technique.
Experience with modern society has shown that "third places" and human interactions get mowed over when you let algorithmic engagement take over. As I understood GP's point, the whole point of the "words on a shirt" was to get together as a group and attempt to form bonds, and by handing the task over to chatGPT they failed to do that.
Technically you can imagine a future where people use chatGPT "in moderation", but in practice they'll use it for everything and spontaneous/creative "hanging out" will suffer as a result.
reply