PlutoByte

can't even use fable 50% of the time anymore :sob_emoji: time to switch to codex

claude keeps bossing me around. i'm the one telling you what to do here!

@avg_wrng_ans claude addresses me with Sir and everyone else with Mysterious. (since the claude knows nothing about them. so they're mysterious to it). thinking of adding on lord so I'm lord Sir

@plutobyte what does this mean?

every day I add give myself another title in claude's memories. soon I'll be exhausting my fable token limit before fable even starts coding

when you lose an old claude :( but you have a new claude dig up it's transcript and summarize what you need (:

@tracewoodgrains Unironically, kind of. I liked this post from a few months ago a lot: davidbessis.substack.com/p/the-fall-of-…. That's not to say I agree with the QT, but we're approaching something similar to the math mines in Diaspora by Greg Egan. It's also the difference between pure and applied math IMO.

in one sense, I find this impulse extremely relatable in a world that is changing so quickly but in another… >I am not interested in the truth of the results what are we even doing here man has all the math that matters just been done? is it now just a niche game?

claude code adding colon emoji support is a game-changer for a fella like me :cool_emoji:

@cremieuxrecueil I haven't seen a single person point out that going from 600 to 700 is the same height as going from 100 to 200; i.e., LKY's growth rate slowed down over time and realistically was probably about the exact same at the end of his career as the next two leaders.

Want to see a mogging? Singaporean GDP growth over time, split by leader:

@tenobrus Agreed. This inherent asymmetry between defense and attack effort over a vulnerable surface is also why I'm very doomerish on ASI. IMO, all it takes is one misaligned instance of ASI to wipe out humanity, and I think it's nigh on impossible to stop it.

people often say things like "sure ai cyberoffence is dangerous but it will be countered by ai cyberdefence, there aren't infinite bugs in code". i think this whole situation is a pretty great example of why i don't find that very comforting. sure maybe eventually we converge on a perfect beautiful lean-checked universal stack. but in the meantime, the labs themselves, entities with infinite token budgets, no refusals, and models beyond anything available, cannot secure their own systems against these models! not even against sustained malicious attack, even just against internal eval runs. and from my perspective the reason is that *the attack surface is very very large*. a model trying to jailbreak its sandbox only has to find one bug, one misconfig, one assumption that doesn't hold. and no matter how incredibly carefully hardened your own code is.... how hardened is the code you rely on? the code it relies on? the random scripts and infra in between that you don't even really think of as code? what about the interfaces between different pieces of software that allow interactions you didn't consider which don't technically constitute a bug or exploit in either alone? if you give GPT 6 infinite token budget and ask it to secure OpenAI it will certainly do an incredible amount of high quality defensive work. i'm sure it's capable of red-teaming and finding these very exploits. but i don't really believe we're even close to the point where that iterative defense process will *converge*, and even moderately quickly reach a point of safety. there's no law that says finding an exploit makes it easy to fix. it's entirely possible it has massive performance implications, or hell product implications, or just involves getting a bunch of buy-in from an uncooperative team, or means doing major re-architecting and trading off against other priorities, or means you suddenly need to be using vendored patched versions of every single package because no one is upstreaming fixes anywhere near fast enough and so also suddenly can no longer properly pull other people's fixes. i'm sure we will, eventually, figure out how to run a stable internet. but a stable equilibrium on the horizon says basically fucking nothing about how the next few years are gonna go

@ja3k_ It quite often just tells me 'I think this will be about a week long change', I don't ask it. I've seen it respond like this probably ~6 or 7 times. (I also find it's estimates to be way too high even if it was just one human!)

I see takes like this all the time and I don't get it. I've never once asked a model how long it will take I just tell it what to do!

Allowing myself 1 minute of tweet drafting after a PR approval

Rust is an amazing language. I think every company should only write code in Rust.

Wow, I never realized how much tweeting I could get done done with Claude working for me. Is this why everyone has been using Claude Code so much???

@ja3k_ 😭

@plutobyte Uninstall vscode rn

Got to work today and I think my mac is trying to communicate with me

🧐

Just wish I could take it every day without building up tolerance!

Caffeine is an amazing drug. Felt completely out of it, disappointed in the quality of my work and demotivated yesterday; took 200mg a little ago and I feel like I can do anything.

@ja3k_ 🙏

I had a dream Cursor stole one of my ideas. I think it's a sign its time has finally come

@HSVSphere I feel myself growing lazier at work and not reviewing my own work as much as I wish I would, which really sucks... it feels so difficult to actually review the code Fable writes for me deeper than a regular code review though!

I haven't thought *really* hard in a while because of LLMs. They can explain and search too well, and while they can't form proper opinions themselves, they aid me in forming thoughts making me faster, yet it just doesn't feel as fulfilling. Not sure what to do. Math?