ripvanwinkle

what plan level are you on

I've been using codex to design and build a product that is a fairly conventional looking app - think of a gmail like experience - with some delicate synchronization across devices.

I find that I need to invest a whole lot in high level design myself to get Codex to create a suitable architecture and make the right tradeoffs. It's more like I am designing and Codex is reviewing and occasionally we brainstorm. When I tried having it design based on requirements, it went wild with an unsustainable design / architecture.

And I use things like plan mode etc. My experience is unlike what I read in most vibe coding exploits.

I wonder if I am doing something wrong. Is there a good canonical example of a project built with Codex or even Claude Code that shows how the human and AI interact that I could use as a reference

Why would they do it. Is it for marketing ?

this is my favorite bit

>So I gave up and sent mail to Amir saying - where is this Moviemaker download? Does it exist?

>So they told me that using the download page to download something was not something they anticipated

fixed the url here

I've noticed ChatGPT Work does a much better job of editing google docs than the pure Chat. It also does a better job of handling long threads and doing the necessary compaction to get better results

This para caught my eye

>Frontier cyber models may push states and defense firms toward the opposite logic: security by obscurity, with closed software, closed tooling, closed firmware, and closed chips. If a model cannot train on the code and architecture of a target stack, it will usually have less context and less speed. That does not make systems safe, but it does raise the value of proprietary stacks all the way down to hardware.

Is this really true. Are there any experts who can weigh in on this.

Should we interpret this to mean that in the new world Windows is more resistant to attacks than say Linux.

TSLA not in the top 10- very interesting

I stumbled on this page and it seems to imply that Claude Code calls other models like Qwen, not just Anthropic's

Is that commonly done. Presumably this is from people customizing their installs - is that correct?

IF the feds are a sufficiently large market your viability as a business might depend on keeping them happy.

btw i am not making a judgement call on the ai usage issue itself, just saying that this and taxes are more equivalent than it might seem

One observation.

Having your work being used by the govt in ways you disagree with feels similar to having your taxes used in ways you disagree.

When you pay taxes you have no say in the bombs acquired with that and where they are dropped. The latter though doesn't seem to provoke the same push back

>>In our analysis we found that all frontier models we tested were able to reproduce the original, human-written bug fix used as the ground-truth reference, known as the gold patch, or verbatim problem statement specifics for certain tasks, indicating that all of them have seen at least some of the problems and solutions during training

this statement alone seems to invalidate the SWE-bench tests

The LLMs need documentation

I had this big guy, very big guy with tears streaming down his cheeks asking me "Sir can you take down that woke star that keeps me awake"

I'd say it's a closer race and the end is not a foregone conclusion yet. That one country currently exhibiting some very troubling tendencies also has more robust self-healing mechanisms in the form of democracy. It has gone off course in the past too but then found its way back. That said, those self-healing mechanisms are under attack as well so there is at least some suspense in the long term outcome

The surplus goes to the owners of the capital. Labor has been losing to capital for sometime now

I was wondering about that too. It shows 1.9M Software Developer Jobs and 122K Computer Programmer jobs.

Reason for hope

because it worked out for North Korea

Pick an activity that is accessible that catches your fancy. Even better if you already have an activity, just spend more time doing it and with people you enjoy hanging out with. At a minimum you'll start feeling less lonely and over time hopefully you'll start forming relationships outside the activity

I am a recent convert to pickleball and highly recommend it because it relatively easy to start with but also the wide range of people who participate in the sport - college kids to retirees

It's not just about protecting data in the old sense - typically from other corporate entities. It would suck if your information somehow made it into a generally available model that then leaked some of that anyone asking a question

I see mentions of Gemini as a fast growing alternative to ChatGPT. Isn't anyone troubled by the fact that for consumers there is no way to keep your data from being used for model training if you want to maintain history of your Gemini chats.

ChatGPT respects privacy and allows for maintaining history while also opting out of using ones data for model training

This shouldn't be a surprise - capitalism always overshoots. Anything worth investing in will generally receive too much investment in because very few people can tell where the line is.

And that's what causes bubbles but at this point it should be clear that AI will make a substantial impact - at least as great as the internet, likely larger

shouldn't the comparison be with gpt4o or 4.5 and not 4.1 or o3

a well written postmortem and it raised my confidence in their product in general

I think this is the best april fools article I've seen

Very cool to have his kids closely involved in his work

Its cyclical, a lot of companies over hired and are trying hard to cut down

about time. you also need a clawback provision since it can take a while for flaws to be detected and the execs could be in new jobs by then.

As a newish user of Apple (Macbook and the IPad mini) it was not as big a leap from Android and Windows as I had feared. I still live in Google services including Gboard on the IPad mini and apart from klutzing around with system settings occasionally the mini feels "not terribly different" from the android devices I use. The Macbook is a bigger challenge though.

I only picked the mini because I couldn't get the same performance with that form factor in Android.

I only entertained the mini because I was forced to use a Macbook for work and realized that apart from annoyances with keyboard shortcuts and system settings I could continue to live in a Firefox + Chrome + Edge + Google services ecosystem.

I will now definitely consider Apple hardware if I don't find a good fit in the Android + Windows world

It would be interesting to feed it a formal language specification of some language it hasn't seen and then ask it write code and see how it does.

That could be a test of reasoning and reading comprehension