simonw

11 ms·

301.
by simonw 2mo ago
Problem is that apparently the US will occasionally elect a corrupt president, so we need stronger measures in place for when that happens.
302.
by simonw 2mo ago
That was a different issue from the US government preventing Anthropic from selling general access to their Fable 5 model for a few weeks.
303.
by simonw 2mo ago
httpx2 is sponsored by Pydantic and has involvement from the new maintainer of Starlette, so it's off to a very strong start.
304.
by simonw 2mo ago
Anthropic made the same change a few weeks after OpenAI did: https://github.com/anthropics/anthropic-sdk-python/releases/... The problem with httpx as a dependency is that it's currently working towards
305.
by simonw 2mo ago
If this is a dedicated transcription model does that completely eliminate the risk that it might do weird things based on the audio it's transcribing? The nightmare scenario with an LLM-based transcription system is that someone says o
306.
by simonw 2mo ago
I had a conversation with someone about this the other day where they asked what AI was useful for, and I said programming, and they asked what programming was, and I eventually stumbled into "it's how you make apps". I'
307.
by simonw 2mo ago
That's a very convincing answer. I hadn't considered how many advertisers need video ads now and don't have the skill or resources to create them.
308.
by simonw 2mo ago
Interesting that OpenAI abandoned Sora entirely but Google are continuing to invest heavily in their own video generation. Maybe because they see video generation as key to developing "world models"?
309.
by simonw 2mo ago
Your hypothetical high schooler would have to learn what architecture is, and then what self review is, and iteration, and manual testing... and what "code" and "software" is first.
310.
by simonw 2mo ago
It's important news for the members of this community independent of the politics.
311.
by simonw 2mo ago
I imagine it's on the HN front page because it directly affects the technology careers of a large number of the people who are members of this community - or are coworkers of or employees/employers of members of this community.
312.
by simonw 2mo ago
The trick there is to use a subagent to read the HTML page and extract the relevant information, than dumping all that HTML into your top-level session. That's effectively using an LLM as an HTML to markdown converter, which is both a
313.
by simonw 2mo ago
... and once more with UD-IQ4_XS https://tools.simonwillison.net/markdown-svg-renderer#url=ht...
314.
by simonw 2mo ago
That used to matter to me back in the days when the best models still only accepted ~32,000 tokens, but these days even the models that run on my laptop are happy with ~100,000 and the hosted models I use take ~200,000 or more.
315.
by simonw 2mo ago
Tried again with a different quant, UD-Q2_K_XL: https://tools.simonwillison.net/markdown-svg-renderer#url=ht...
316.
by simonw 2mo ago
By chance you have a browser extension that might block fetching data from raw.githubusercontent.com ?
317.
by simonw 2mo ago
Good bicycle, good pelican: https://tools.simonwillison.net/markdown-svg-renderer#url=ht...
318.
by simonw 2mo ago
I ran some pelicans at the four different reasoning levels (none, low, medium, xhigh - apparently high and xhigh are aliases of each other) on a DGX Spark using Unsloth's unsloth/Qwen3.8-Flash-Next-GGUF (UD-IQ1_S): https:/&#
319.
by simonw 2mo ago
> More generally, it's not clear to me that there was any technical benefit to the rewrite in the first place. The Bun 1.4 announcement claims: https://bun.com/blog/bun-v1.4 > It reduces idle CPU usage by 5x
320.
by simonw 2mo ago
I think of it as similar to how SEO encouraged a lot of sites to start using more accessible markup and headings, which resulted in accessibility improvements even though the developers weren't thinking about that at all.
321.
by simonw 2mo ago
Which ones?
322.
by simonw 2mo ago
I wasn't particularly interested in WebMCP until I had a conversation with a screenreader-using accessibility engineer who pointed out that WebMCP is an incredible accessibility technology disguised as an AI thing. Imagine if we can tr
323.
by simonw 2mo ago
I think the understanding is the most important piece here. I've seen this called "cognitive debt". If you don't understand the system you are building - if the AI gets too far ahead of you - then you lose the ability to
324.
by simonw 2mo ago
I don't mean to imply that was easy! I think it's hard, and a skill that needs active investment. It's one of the reasons I'm not afraid for my career.
325.
by simonw 2mo ago
I don't work with a large team, so I don't personally experience the hell of coworkers dumping thousands of lines of unreviewed slop on me. (I get a bit of exposure to that from my open source projects but it's much easier to
326.
by simonw 2mo ago
> That's called programming. The Bun tests and oracle are the result of years of programming. I completely agree. That's why I warned about getting distracted by the headline.
327.
by simonw 2mo ago
Yes, why do you ask?
328.
by simonw 2mo ago
This is a strong article with a distracting headline. Challenge for commenters: can you discuss the content without getting caught up in the headline? My favorite paragraph: > The fact that AI wrote 1M LOC and then refined it over the c
329.
by simonw 2mo ago
> How much do you understand about statistics and p-values? Is that a well designed survey question? I would expect the use of a specific jargon term in that question to affect the results in a significant way.
330.
by simonw 2mo ago
"decided against buying one or more software products or features" I suspect the "or features" there is doing a lot of work.

More ›