mirroring

@_ontologic When AGI and asi o clock?!

@mirrorsrmirror AI is on tha clock

Nobody’s been checking their watch, so they didn’t know what time it was, not me tho 😮‍💨

@_ontologic But what time is it haha

I think the idea the gov't doesn't pay well enough to attract good talent is a great move if you want people to underestimate your capabilities. Remember when a bunch of people followed Thiel/Musk to work on DOGE or other gov't things? There are plenty of people who like to work on software for the purpose of it (opposite camp of calebhearth.com/dont-get-distr…) Am I avoiding finding that exact number? Sorta. The intelligence community has reason to be secretive over where funds go and I imagine the same applies to data. If I guess an incorrect number, nothing matters. If I guess a correct number that gives away information about the data itself, then I don't think I'd like that (classified information isn't about the information itself but how they got it, if the amount of data is less than X tech company's total platform activity then you may be able to skirt around eyeballs by botting there instead of somewhere the NSA would reliably be overseeing) Is there also a perpetual cat-and-mouse game between the "good guys" and the "bad guys"? Totally, think that can turn into an irrelevant circular thing tho The floor is zero data and the ceiling is the entire internet's worth of data. The question isn't so much "how much data does the NSA gather" or "how much data does the NSA hold onto and for how long" but "which of the data do they gather/hold onto"

@pwnies ahh i was going to resend it when i sent it i meant to say "why not branches on different *local* VMs?"

@mirrorsrmirror That's kinda what the promise of cloud agents is, and it does work. It's just costly spinning them up / spinning them down. There's a lot of resource wastage. Would be better if the system was able to distinguish what elements it could share / what elements need separation.

I'm convinced worktrees aren't the solution. I'm onboard for a new take for git - we need something that allows for parallel work while managing the diffs between different branches of the same codebase seamlessly.

people are using LLMs to invent new math we're using it to find undiscovered TUI spinners

@thepanta82 @MatthewSchrager @jamonholmgren i dunno man. i personally haven't stopped learning as much as possible and using ai as much as possible since dec 1, 2022

@mirrorsrmirror @MatthewSchrager @jamonholmgren ChatGPT was impressive, but not at all clear how far it could go IMO.

I've been spending a lot of time talking to experienced software engineers lately and when you get in a 1:1 candid conversation, everyone is rather shell-shocked by how quickly and drastically things have changed.

@HotAisle you should suggest that to their team somehow?

@mirrorsrmirror if it was me, i'd give credit to 3rd parties who submit issues as well as the person who did the PR.

Latest beta of cloudflare vinext has a fix for a problem that I discovered and reported, but I get zero credit. Since 1991, I've been making the internet a slightly better place, and few know it.

@tszzl what is the opposite of "AI is hitting a wall" in your most eloquent words that you craft and compile

year after year I’m like “better models are coming” and ai twitter screenshots as tho it’s news w something like “GPT8 confirmed” in a samsaric loop since chatgpt. we are all like babies without object permanence when it comes to exponential progress. but better models are coming

@HotAisle do they usually give credits for PRs?

@mirrorsrmirror github.com/cloudflare/vin…

@thepanta82 @MatthewSchrager @jamonholmgren x.com/owenthcarey/st…

@mirrorsrmirror @MatthewSchrager @jamonholmgren what happened then?

@jamonholmgren publicly it's optimism. privately it's mostly people recalibrating.

@jamonholmgren I was genuinely debating whether to DM you when I saw this post... A bit... relieving... to see that it's not a tiny subset of engineers across all levels trying to figure stuff out

@jamonholmgren i don't understand... we are in an industry literally known for change. the frontend for example was known for having a new framework every year why is change surprising

@jamonholmgren Shell shocked is a great way to put it, with a bit of emptiness

@jamonholmgren Six months ago i couldn't believe people said they don't write code by hand anymore. In the past week I think the only code I wrote by hand was changing a single tailwind class

@jamonholmgren Shell-shocked is exactly right. I hadn't tried Codex or Claude Code myself until a week ago. Suddenly I'm dual-wielding both and still maxing out my usage credits. I'm not worried. On the contrary: I'm excited for software development now more than ever.