Live data from Hacker News

Viewing profile — hansvm

hansvm

HN member
Joined
Sun, Dec 22, 2019, 1:35 AM UTC
HN karma
5,395
Public activity
2,842 items

About hansvm

No profile information was provided.

Recent public activity

  1. comment
    Comment #49233828

    I'm 100% convinced that accidental website hits and bounces from unique IPs when viewing the ensuant shitty content counts as MAUs for most, if not all, major tech companies.

  2. comment
    Comment #49233799

    But what if you say that FSD is actually here and that the only reason you can't fully use it is regulations and then need an extra decade (at least -- still pending)? As one of ma…

  3. comment
    Comment #49232034

    That reminds me of a fun research presentation I saw a decade ago. They engineered a bioplastic which would break down on sustained contact with water. While trying to come up with…

  4. comment
    Comment #49133913

    It's not even complexity; raw size suffices to make a codebase hard for the LLM to manage, which is especially bad given the high ratio of LLM lines to ideal lines for any given ch…

  5. comment
    Comment #49130149

    They do care, but not at the right point in the sales cycle. They care about the eventual bait and switch. They care about how in the face of that pack of lies they can't even reco…

  6. comment
    Comment #49117367

    Yeah, they re-enabled using user history again in the settings. You'll want to turn that off.

  7. comment
    Comment #49104562

    Clear prompts have never worked as real constraints. Ask any OpenAI model to respond in full paragraphs, as forcefully as you'd like, on a prompt [0] involving MMOs and requiring 1…

  8. comment
    Comment #49098258

    Thinking... The prompt is about whether to hack Hugging Face. I have a relevant skill: "Do not." However, the skill only says what not to do, and doesn't explicitly forbid "respons…

  9. comment
    Comment #49098161

    If you don't, at a bare minimum, read the code once though, you don't know how it interacts with the world.

  10. comment
    Comment #49054664

    Net Present Value isn't quite the right thing to look at. Inflation-adjusted energy prices have been rising over time, so the thing these batteries are projected to deliver (fixed …

  11. comment
    Comment #49041412

    Oh yeah, that's a use case I love long contexts for. "Here's a day of prod logs, what's broken?" It's an incredible utility, but the qualities much less than with shorter contexts,…

  12. comment
    Comment #49035502

    At my last company I cut 8k lines from a 10k component and fixed every major bug in the process. It wasn't "dead" per se, but if you start by cleaning up one little bad abstraction…

  13. comment
    Comment #49027329

    FWIW, I notice the intelligence drop drastically with long contexts with Opus 4.6. It's barely usable for anything intricate and long. That long window is good for _something_, but…

  14. comment
    Comment #49015568

    For the lazy among us (not me of course), is there a small number of core techniques which enabled this for even a single architecture and single CPU core?

  15. comment
    Comment #49015530

    1000x improvements unlock qualitatively new capabilities, even when only applied to subcomponents. Moreover, the fact that it's only 0.1% of the runtime is usually [0] an artifact …

  16. comment
    Comment #48998703

    Say what now? What criteria do they use to scan the populace and create a pre-crime registry of any kind?

  17. comment
    Comment #48969177

    Assuming the ink is free, you use some sort of scheme to take advantage of the full printable ascii character set, and you don't bother presenting key values (if necessary, you cou…

  18. comment
    Comment #48958367

    The more obvious answer is that they get different marketing signals from the two companies' offerings and want both.

  19. comment
    Comment #48958271

    Both. The hypothesis with Eytzinger is that you're doing vanilla binary search on a data structure where each hop is one operation. It doesn't, by itself, do anything to optimize a…

  20. comment
    Comment #48950571

    My holdings aren't even locked up, and I'm still not allowed to short my employer -- true as a matter of policy which could get me fired, and true from a US legal perspective most …

  21. comment
    Comment #48948153

    My hunch is the HN formatter swallowed the double asterisk typical of python exponents. While we're being pedantic, 2^30 is 28 in normal programming languages ;)

  22. comment
    Comment #48943355

    > LLM-as-a-Judge Empirically, many problems look like they're easier to check than they are to solve. This seems like a reasonable way to bootstrap a little extra performance, with…

  23. comment
    Comment #48923784

    I thought about proposing that idea myself and decided against it. Consumer behavior is a bit hard to predict, and it's a little unclear to me how well "just works" will sell. I li…

  24. comment
    Comment #48921805

    This might sound like a rant, but it's more an attempt to outline the breadth of the problem by way of many examples. Software quality sucks right now, enough so that I plan my lif…

  25. comment
    Comment #48902857

    > Even though the text we're reading is Claude talking to us as well :) I'm not getting that vibe from this one; I'd bet money on it at least being substantially human edited. What…