itsgrimetime

i'm also confused. my speculation is they trained a lot of different versions/revisions of models. maybe they're different base architectures, maybe they're different pretraining sets, or maybe they're different RL/post training regimes, and we're just seeing a few of them. still not sure what the names would mean internally though even in that case

or I'm dumb and its just elements with names starting with A, B, C

If it's like any other matching (game) decomp, you compare the output byte-for-byte to the retail binary (e.g. any of the projects on https://decomp.dev/). Hard part to getting started is making sure you have a comparable compiler, flags, & toolchain

yeah but knowing if something is impossible or not is pretty hard to determine, right? The line between “takes weeks of trial and error and lots of out-of-the-box thinking” is indistinguishable from “literally impossible” until it’s been done. And they’re trying to get these models to do things that people haven’t been able to do. some would say these are/were “impossible”.

I don’t have a problem with labeling them. I would love to see more engagement with the content regardless of how it was written, though. I know a lot of folks do but I feel like I’ve been seeing more and more instant dismissal/criticism of how it was written rather than what the submission is actually about.

I don’t think the idea is that these companies are going to make their $ selling their services, that’s just step one. they’re betting that they’ll have their own “country of geniuses in a data center” to put toward whatever thing they think will make them the most money.

“These AI tools are garbage and can’t create anything worth creating”

“These AI tools are so powerful they can steal your ideas with nothing but a sentence”

I know that’s not exactly what OP is saying but the pretentiousness of the “we knew better” got to me a little bit. I think it’s a cool and unique analogy but I’m not as pessimistic.

Ideas have become so cheap to try/experiment with, more people are able to try 10x more or whatever, and that may keep increasing, I think there are way less hunters than hunted

a lot of the replies on here (not just yours, I just picked yours to respond to) make it clear I didn't articulate what I was meaning to very well - I'm still doing the "engineering". I used "programming" in a more general sense: building stuff with computers. I still go through the same motions. I try something, hit some failure mode, have to think of and (with the help of claude) execute on that, evaluate it, decide if its better or not, identify when the agent is off track or deviating from the vision I have, etc.

it seems you and others took my words a bit more literally than I intended for them to come across. it's not like I'm just one-shotting all my ideas directly into existence, I still need to understand how to use the tool to do it. it's just a different tool. one that's allowing me to build way more than I ever have, while having a ton of fun doing it.

and sure, your analogy seems reasonable if I was simply buying the code w/ my tokens. that wouldn't be fun or fulfilling at all - it's more like there is some new "cooking" tool that immediately spawns 90% of the ingredients pre-cut & prepped (maybe the other 10% isn't exactly what I asked for but I can improvise with it) and gives me a decent recipe based on the idea of what I wanted to cook in the first place that fills in (and gives me a starting point to learn about) the gaps that I didn't even realize I was missing. I see it more as: "All this time I thought I loved chopping onions and setting up the grill, but actually I just loved cooking".

you weren't wrong about the mcdonalds though. I do love mcdonalds

All of this new capability has made me realize that the reason i love programming _isn't_ the same as the OP. I used to think (and tell others) that I loved understanding something deeply, wading through the details to figure out a tough problem. but actually, being able to will anything I can think of into existence is what I love about programming. I do feel for the people who were able to make careers out of falling in love w/ and getting good at picking problems & systems apart, breaking them down, and understanding them fully. I respect the discipline, curiosity, and intellect they have. but I also am elated w/ where things are at/going. this feels absurd to say, but I finally feel like I'm _good_ at programming, which is insane, because I literally haven't written a line of code myself in months, but having tools that can finally match the speed my ideas come to me is intoxicating

I've done this! Given the right interface I was surprised at how well it did. Prompted it "You're controlling a character in Old School RuneScape, come up with a goal for yourself, and don't stop working on it until you've achieved it". It decided to fish for and cook 100 lobsters, and it did it pretty much flawlessly!

Biggest downside was it's inability to see (literally), getting lists of interact-able game objects, NPCs, etc was fine when it decided to do something that didn't require any real-time input. Sailing, or anything that required it to react to what's on screen was pretty much impossible without more tooling to manage the reacting part for it (e.g. tool to navigate automatically to some location).

ive recently found some rayon shirts I really like, but how do you wash them without destroying them? everything I've read online says dry cleaning is the only way

Yep - this has worked well for me too. I do it a little differently:

I have a /review-sessions command & a "parse-sessions" skill that tells Claude how to parse the session logs from ~/.claude/projects/, then it classifies the issues and proposes new skills, changes to CLAUDE.md, etc. based on what common issues it saw.

I've tried something similar to DISCOVERIES.md (a structured "knowledge base" of assumptions that were proven wrong, things that were tried, etc.) but haven't had luck keeping this from getting filled with obvious things (that the code itself describes) or slightly-incorrect things, or just too large in general.

When you have 70+% browser market share, stopping support for something _is_ killing it.

Wish they would have included what the actual failure mode was. I’ve been having issues where Claude Code will just hang after running some tool call, was that caused by one of these bugs?

Anthropic is US-based - unless you meant something else by "foreign corporation"?

I have the opposite experience (although with sonnet) - I have to recklessly instruct Claude to make super verbose tool calls to even get close to chewing through enough tokens to use up limit before the 5 hour reset period or whatever it is. I don’t find enough performance increase w opus to justify it.

I’m not sure how to go about solving it at the architecture level but I would assume an LLM with access to a diff tool would get 100%, but I understand that’s not really the point

site does something really strange on iOS chrome - when I scroll down on the page the font size swaps larger, when I scroll up it swaps back smaller. Really disorienting

Anyways, never heard of oiiotool before! Super cool

00000001 00000001 00000001 00000001 = 16843009 in base 10 (concatenate each dot-separated 8bit number as one big base 10)

I agree - the truly curious will be rewarded while those who couldn’t care less will mindlessly copy and paste. Maybe that will give the rest of us job security?

> It’s only adversarial because you want to get as much pay as possible out of them for as little productivity as possible.

Or maybe pay that’s proportional to the value we provide

or someone from the hotel was in on it

IMO this is a bad take. I use LLMs for things I don’t know how to do myself all the time. Now, I wouldn’t use one to write some new crypto functions because the risk associated with getting it wrong is huge, but if I need to write something like a wrapper around some cloud provider SDK that I’m unfamiliar with, it gets me 90% of the way there. It also is way more likely to know at least _some_ of the best practices where I’ll likely know none. Even for more complex things getting some working hello world examples from an LLM gives me way more threads to pull on and research than web searching ever has.

Programming tasks, brain storming, recipe ideas, or any question I have that doesn’t have a concrete, specific answer.

> Why invest so much time and money in a feature that prevents such a small percentage of data breaches ...

Because it's a tractable problem that these devs can solve - and just because they're working on this doesn't meant they (or others) aren't also working on the other things.

> It doesn't matter that you can cryptographically verify that a package came from a given commit if ...

Sure, but just because it doesn't solve every single problem doesn't mean it's not worthwhile

That’s a bummer, I use it daily in SF and the tracker it has for upcoming buses/trains has always been super accurate, and the stop countdown is always dead on. It even tells me when I need to speed up my walking to make it to the next bus in case it’s running a little early.

CV pipeline to get some additional realtime stats for an annual Mario Kart 8 LAN tournament my friends and I run, hoping to be able to get real time race position tracking and other stats like boost/drift %, per-player item distributions, average time to 10 coins, and whatever else we can think of (and make work)

Just realized this is posted on the SeaTac website now: “ SEA is experiencing temporary issues with the system that populates flight and baggage information on in terminal screens and the flySEA app/website. Travelers are recommended to check with their airlines for current gate and baggage claim information. Check With Your Airlines”

I just landed at SeaTac an hour ago and the rideshare/app pickup was absolutely nutso. Like thousands of people standing around waiting for taxis and Ubers. The one person I asked what was going on said that the computer systems at all the regional hotels are down (not sure how that makes more people need cabs). Wonder if it’s from this

I just landed at SeaTac an hour ago and the rideshare/app pickup was absolutely nutso. Like thousands of people standing around waiting for taxis and Ubers. The one person I asked what was going on said that the computer systems at all the regional hotels are down (not sure how that makes more people need cabs). Wonder if it’s from this