randomtoast

8 ms·

31.
by randomtoast 7mo ago
That's why I often ask for "Source?" — because sometimes people seem to make up numbers. However, whenever I do this, I receive a large number of downvotes. Maybe it's not common on HN to back up claims with sources.
32.
by randomtoast 7mo ago
It could but then it would show that the effective tax was not 30% but 3% and there is a strong lobby in Washington against that.
33.
by randomtoast 7mo ago
Reads like an LLM response.
34.
by randomtoast 7mo ago
But it is not an em dash.
35.
by randomtoast 7mo ago
> the magic isn't in the input, it's in the system around it. It isn't [this], it's [that]. Is AI slop, just saying.
36.
by randomtoast 7mo ago
> If a model flips 3/10 times on a trivial constraint, that’s a reliability issue, not a reasoning ceiling. I have reviewed your previous comments, and you have consistently written: that's instead of that’s . So what I read
37.
by randomtoast 7mo ago
> But unfortunately the article doesn't mention whether they used the reasoning model or not. You can run the test yourself if you ask GPT-5.2 with reasoning effort high or xhigh, it will always answer correctly. So if the got 0 fro
38.
by randomtoast 7mo ago
> This doesn’t look like a reasoning ceiling. It looks like a decision reliability problem. This doesn’t look like a human comment. It looks like a LLM response.
39.
by randomtoast 7mo ago
Enable reasoning effort and the results are completely different.
40.
by randomtoast 7mo ago
This is because it is without thinking enabled. Of course the results are disappointing.
41.
by randomtoast 7mo ago
The problem is you haven't used the API, but you have used your ChatGPT subscriptions with personality, memories and possible customization. I can see for instance that your ChatGPT answers with emojis, while my ChatGPT subscription ne
42.
by randomtoast 8mo ago
Except for a few models, the selected ones were non-reasoning models. Naturally, without reasoning enabled, the reasoning performance will be poor. This is not a surprising result. I asked GPT-5.2 10x times with thinking enabled and it got
43.
by randomtoast 8mo ago
The entire Factbook Archive was written by Claude Code, as you can see in the GitHub commit history: https://github.com/MilkMp/CIA-World-Factbooks-Archive-1990-2... AI is present everywhere these days. I wouldn’t be su
44.
by randomtoast 8mo ago
I think you could make it work for a fraction of the price if you buy a bunch of low-cost e-ink screens and combine them to create a larger display. The main challenge would be on the software side, as you'd need to control the content
45.
by randomtoast 8mo ago
0.2 tok/s is fine for experimentation, but it is not interactive in any meaningful sense. For many use cases, a well-quantized 8B or 13B that stays resident will simply deliver a better latency-quality tradeoff
46.
by randomtoast 8mo ago
This will lead to natural selection. As AI becomes increasingly integrated into all areas, companies that manage it less effectively than others will face greater selection pressure.
47.
by randomtoast 8mo ago
> Do you think we should just do nothing and let casual bad behavior go because some people are gonna be abusive? No. I refuse to accept that. I'm with you, and I also give a clear: No. My criticism is about how it is handled. If yo
48.
by randomtoast 8mo ago
You just need to use this list as a prompt and instruct the LLM to avoid this kind of slop. If you want to be serious about it, you can even use some of these slop detectors and iterate through a loop until the top three detectors rate your
49.
by randomtoast 8mo ago
I’ve built fairly large OTP systems in the past, and I think the core claim is directionally right: long lived, stateful, failure prone "conversations" map very naturally to Erlang processes plus supervision trees. An agent sessio
50.
by randomtoast 8mo ago
Okay, then we need to distribute the database as torrent. Maybe Anna's Archive can help.
51.
by randomtoast 8mo ago
"So instead of destroying those clothes in Europe, we'll just add an unnecessary shipping step to the process, producing tons of unnecessary CO2." > The world being as it is you're likely correct and your cynicism m
52.
by randomtoast 8mo ago
I think part of the ArchWiki’s strength is that it treats documentation as first-class infrastructure. There is a shared expectation that if you solve something nontrivial, you upstream it into the wiki in a reasonably neutral, upstream-ori
53.
by randomtoast 8mo ago
I think we will see more and more games and applications be fully reversed engineered in the coming month due to increase coding agent capabilities.
54.
by randomtoast 8mo ago
> but I haven’t been to get them to do something totally out of distribution yet from first principles Can humans actually do that? Sometimes it appears as if we have made a completely new discovery. However, if you look more closely, yo
55.
by randomtoast 8mo ago
I mean you can takedown a website hosting the JavaScript/WASM Version of Vice City, but if you just host the JavaScript/WASM on localhost:8080 then TakeTwo cannot take it down.
56.
by randomtoast 8mo ago
I think it should also be possible to mod it into a browser version. I think even GTA Vice City runs in browser.
57.
by randomtoast 8mo ago
> This post is surprisingly shallow considering the author. I think a low effort post from time to time is okay.
58.
by randomtoast 8mo ago
Heck the LLM accusations get a bit out of hand lately here on HN. I could delve into it now ... but I want to safe our time.
59.
by randomtoast 8mo ago
I’ve seen this play out a lot. People say they “write games in C” and then quietly rebuild half of C++ anyway with vtables in structs or giant switch statements, just without the compiler helping. That’s fine if it makes you happier, but it
60.
by randomtoast 8mo ago
The main motivation of the author seems to be that Tor is slow, but you can significantly speed it up by manually choosing Tor nodes. For instance, if you pick your three fast nodes within a short distance of you location, then the speed is

More ›