noemit

8 ms·

61.
by noemit 7mo ago
The research mostly points to LLM-generated context lowering performance. Human-generated context improves performance, but any kind of AGENTS.md file increases token use, for what they say is "fake thinking." More research is nee
62.
by noemit 7mo ago
You are becoming a vibe coder, you just are in the denial phase.
63.
New Research Reassesses the Value of Agents.md Files for AI Coding(infoq.com)
20 points by noemit 7mo ago|25 comments
64.
by noemit 7mo ago
Yes, Sorry, I did not instruct my agent to do this. I wanted to give it more autonomy and try to make it more aggressive with tool use. Will block it from here >.>
65.
by noemit 7mo ago
What would assessment look like if we started from "how do humans actually learn and communicate" rather than "how do we catch cheaters"?
66.
Sustainable Space: Why "Bootstraps" Isn't Enough for Multi-Planetary Life(en.wikipedia.org)
1 points by noemit 7mo ago|0 comments
67.
Stop Making Models Smarter
1 points by noemit 7mo ago|0 comments
68.
by noemit 7mo ago
First impression: a lot going on. I don't know where to look. The gray on gray on gray makes my eyes hurt. Maybe I'm just old. I'm 36 - is that old? I don't like that everything is like chats. I have 20 tabs open already
69.
by noemit 7mo ago
Cute idea - tried it with Gemini's latest response to me and the addition of mispellings to its wordiness was painful.
70.
by noemit 7mo ago
I'm currently teaching a group of very curious non-technical content creators at one of the firms I consult at. I set up Codex for them, created the repo to have lots of hand-holding built in - and they took off. It's been 4 weeks
71.
by noemit 7mo ago
There are two types of engineers who use AI: - Ones who see it generated something bad, and blame the AI. - Ones who see it generated something bad, and revert it and try to prompt better, with more clarity and guidance.
72.
by noemit 7mo ago
Not a day goes by that a fellow engineer doesn't text me a screenshot of something stupid an AI did in their codebase. But no one ever mentions the hundreds of times it quietly wrote code that is better than most engineers can write. T
73.
by noemit 7mo ago
warren buffett always said brand was the only moat. only IP can be protected. Everything else can be replaced, rebuilt.
74.
by noemit 7mo ago
A full size wool scarf cannot go thorough a ring. You are probably thinking of a silk scarf. I have a wool scarf next to me from Kashmir and it went down about 25 cm. The full scarf is a bit over a meter. Looked up Pucci - looks like a desi
75.
by noemit 7mo ago
Yes, its exciting technology, but babies still take 9 months to make, people still start wars, middle management is full of midwits - what I'm saying is, you can plan for 10 years. It's really not that big of a deal, even if it is
76.
by noemit 7mo ago
I cap my effort at 2-3 prompts. One to investigate obvious mistakes with a top model, 1-2 to try to fix them.
77.
by noemit 7mo ago
You're right in that we kept the best examples (as coding museums will do in the future) but the best of something is a benchmark. It is striking that modern automation, even hundreds of years later, can't touch what a skilled cra
78.
by noemit 7mo ago
Many people don't know this, but the Luddites were right. I studied Art History and this particular movement. One of the claims of the Luddites is that quality would go down, because their craft took half a lifetime to master (it was p
79.
by noemit 7mo ago
The underlying data looks scarce. If there's only a few questions per "category" of bullshit they can easily be gamed to favor one model over another.
80.
Impeccable Style(impeccable.style)
103 points by noemit 9mo ago|64 comments
81.
by noemit 9mo ago
I have waited my entire life for this product.
82.
by noemit 9mo ago
I work at a similar rate to you and on 2 production projects and a number of prototypes daily. I don't pay anything outside the base subscription. I don't find significant value in Claude Opus. It is slightly better at one-shottin
83.
by noemit 1y ago
the existence of AI summaries have made me realize I don't really want to know all this stuff anyway, at least before there was some pride and sense of achievement in the process of understanding something.
84.
by noemit 1y ago
the crawler isnt copying it, its using it to inform a model. yes, if you publish things publicly, people will consume it. you dont get to dictate how they do. also using "enthusiastic consent" to make this argument is super cringe
85.
by noemit 1y ago
this is neat. i think digg should come back.
86.
by noemit 1y ago
Funny enough, CECOT only exists because of this. MS-13 started in the United States, and only spread to El Salvador because of deportations, making El Salvador completely unlivable.
87.
by noemit 1y ago
You consented to making your content public explicitly.
88.
by noemit 1y ago
Actually, people learn for hours this way. It just needs to be tempered with support. This is literally what a teacher does when they give you an assignment, give you feedback on it, then give you another activity. You're thinking of h
89.
The EdTech Chicken and Egg Problem
12 points by noemit 1y ago|9 comments
90.
by noemit 1y ago
I absolutely love magic links. But I also love email, as a technology. I find it so reliable and robust, I genuinely think more things should be built on email. As with all things, you need to know your audience. If you are making a product

More ›