bensyverson

21 ms·

1.
by bensyverson 3d ago
The non-alphanumeric characters are just asking for trouble. Agents will fail to escape them correctly, and the $ could expand a variable. Base62 (or 64, with -_) is totally sufficient.
2.
by bensyverson 5d ago
Looks about as effective as Robots.txt
3.
by bensyverson 5d ago
Yeah, that's like someone in 1998 saying we're at peak internet hype
4.
by bensyverson 5d ago
And 20% YOY compounds quickly… If the speed doubles in 4 years and the RAM situation gets better, we could be looking at some very capable on-device models in 2030.
5.
by bensyverson 6d ago
The text is clearly not written by AI… people need to chill on this accusation. Yes, the blog design itself looks Claudish, but if you actually read the post, no AI writes like that.
6.
by bensyverson 6d ago
Our devices are always listening, if you have them set up to listen for hail words like "Siri" or "Hey Google," so I think the Overton window on listening already shifted about 10 years ago. The device actually recording
7.
by bensyverson 6d ago
The stock /frontend-design skill really loves an all-caps mono caption
8.
by bensyverson 8d ago
Not anymore! [0] [0]: https://www.mturk.com
9.
by bensyverson 8d ago
Completely agree. How many people over the past 20 years have been musically gifted, but found ProTools or Logic impenetrable? The AI unlock is about enabling that creativity, not replacing it.
10.
by bensyverson 8d ago
It's already possible, but in my opinion it will never be popular. In some ways this has already been proven through other media; ChatGPT can easily generate a novel, but how many people are reading pure-AI novels?
11.
by bensyverson 8d ago
Yes, I'd be really interested to see Fable / Opus / Sonnet. GLM 5.3 Flash Max had an interesting showing. Its Codex version was bad [0], and it completed in 9 minutes. The OpenCode version was much richer [1] and more detaile
12.
by bensyverson 8d ago
I have videographer friends, and this is exactly what they've been asking for. Their dream is to be able to say "sync the A and B camera footage, chunk the whole interview down into segments, and then pull together the highlights
13.
by bensyverson 9d ago
I'm building something similar with `Agents`[0] (think of it as a package manager for AGENTS.md snippets). I'm getting good results by including detailed instructions for certain tasks in the repo, and providing hints in AGENTS.md
14.
by bensyverson 11d ago
For context, this post seems to be a reaction to MacStories deciding to return to X [0], which has been quite controversial in the Apple developer community. There has been the predictable blowback, then the blowback to the blowback, which
15.
by bensyverson 12d ago
Yes, this makes little sense. It looks like it's a way to avoid having Claude read or write your code. And why stop at 90%? I have this one weird trick to reduce Claude Code token use by 100%: use a different harness and model!
16.
by bensyverson 13d ago
I do. Do we think economists are actually generating this understanding? I'd be curious to know about policy changes which have been the result of non-obvious insights from the field of economics.
17.
by bensyverson 13d ago
> Economists aren’t naive, so they don’t really believe these things entirely. But they assume they are mostly correct over the medium to long term. Spoiler alert from the article: turns out economists are wrong. This will sound inflamma
18.
by bensyverson 14d ago
It’s better than for real, it’s for LinkedIn
19.
by bensyverson 14d ago
It probably depends on what you're trying to do; I needed Markdown styling and Markdown-aware behaviors (e.g. when you hit return after a list item, indent the same amount and add a new list marker), context-aware nesting (hit Command-
20.
by bensyverson 14d ago
Dupe: https://news.ycombinator.com/item?id=49535835
21.
by bensyverson 14d ago
An SEO tale as old as time
22.
by bensyverson 14d ago
In case you find this site difficult to read: document.getElementsByTagName('main')[0].style.maxWidth = '39em';
23.
by bensyverson 14d ago
Yes… I don't think the author is thinking on a long enough time horizon. I'm curious what happens to these landfill sites in 100, 1000 and 10k years. Do they inevitably break down and leak into the water table? Do they eventually
24.
by bensyverson 14d ago
Thanks—I might return to it and open source it soon… It was a pre-Claude Code project, and it was hard to feel any momentum. It's basically a note-taking tool for capturing quick one-off notes in Markdown, but it stores a lot of surrou
25.
by bensyverson 14d ago
I implemented a pretty simple iOS/macOS text editor using TextKit 2, and it was astonishing how many hacks, workarounds and manual implementations were needed to support even the most basic features. That codebase is the polar opposite
26.
by bensyverson 15d ago
Ah yes, have you read Seth Godin's "1000 unwitting accomplices?" In Marketing 101 classes, the first session is usually "Recruiting Useful Idiots." You seem to have product marketing all figured out!
27.
by bensyverson 15d ago
Yes, basically. [0] [0]: Obligatory https://xkcd.com/386/
28.
by bensyverson 15d ago
Flawless logic! I can’t believe I got busted again… The other lizard people are going to be so disappointed. I’m already on probation for leaking the fact that the Earth is flat and bungling the JFK thing. Now they’ll probably pull me off t
29.
by bensyverson 15d ago
True, but so what? Is the idea that they sponsored the creation of OpenClaw and Claude Cowork, CIA-style? Or is the idea that this (presumably lower-level, given that they're working on the mini) marketing employee had a large discreti
30.
by bensyverson 15d ago
You're right, Apple has a long history with guerrilla marketing, especially gray-area tactics like bots and paid commenters. Even though there is no evidence of this, it "feels" true, so it must be true, right? Why wouldn&#x2

More ›