Live data from Hacker News

Viewing profile — jrflo

jrflo

HN member
Joined
Thu, Mar 19, 2026, 2:23 PM UTC
HN karma
957
Public activity
188 items

About jrflo

festudio.net

Recent public activity

  1. comment
    Comment #49222727

    This is the thing I've become increasingly radicalized about. I built an app (link below if you're curious, it's $5 [1]) to cure my particular brand of phone addiction. I've been u…

  2. comment
    Comment #49215263

    Might not actually be that much cheaper, we don't know what margin OpenAI is charging on Luna API. Open models likely have much less margin.

  3. comment
    Comment #49214814

    I'm using the term "one" loosely, it's a chain of exploits rather than a single weakness, but the argument is the same: it's much harder to find every chain than a single chain.

  4. comment
    Comment #49214180

    Do you really need a KV cache if inference is that fast though?

  5. comment
    Comment #49213987

    Not a security guy but my understanding is: you only need to find one flaw to exploit a system, to make a system totally secure you need to find them all. It's inherently easier to…

  6. comment
    Comment #49213167

    https://spectrum.ieee.org/orbital-data-centers-heat This analysis calculates 80 square meters of radiator surface area needed per rack, or 200,000 square meters per 100 MW. It's no…

  7. comment
    Comment #49212580

    This is a nothingburger article, 23,000 jobs lost is essentially flat.

  8. comment
    Comment #49212517

    The Fed is not going to cut rates. Looking at Fed Rate futures they only predict increases: https://www.cmegroup.com/markets/interest-rates/cme-fedwatch...

  9. comment
    Comment #49211308

    > Cost of compute on Earth Power build Permitting Operating costs Land Energy costs Infrastructure Cooling > Cost of compute in space Launch cost Operating costs Satellite build In…

  10. comment
  11. comment
    Comment #49211000

    Thank you for this, I'm going to just accept this theory and close the thread because it makes me so upset, lol

  12. comment
  13. comment
  14. comment
    Comment #49196941

    You clearly haven't spent much time around young adults. I work with college students a fair amount and they use chatbots for 100% of their questions, including the ones you listed…

  15. comment
    Comment #49188254

    Their entire business is predicated on search dominance, which is now on much shakier footing. Agreed that those comments were ridiculous pre-ChatGPT, but now they have a genuine c…

  16. comment
    Comment #49177695

    Realistically it's "too much 'environmental' review", aka NIMBYs killing any project that will actually implement good public transit because it will ruin the view from their kitch…

  17. comment
    Comment #49171255

    Totally agreed. Capitalism is just a game of competition, and if you want to win the game you should take all advantages that you have. Pretty much everyone else you're competing a…

  18. comment
  19. comment
    Comment #49160134

    I think we previously assumed that AGI would need to come before superintelligence, but it kind of seems like that's wrong? Super intelligence in a narrow field has arguably alread…

  20. comment
    Comment #49159691

    I think that HN is particularly negative towards AI because the vast majority of users here will have their prestigious CS careers disrupted by AI advances. So, there's an inherent…

  21. comment
    Comment #49113947

    The local LLM argument never really held water tbh. You can get surprisingly good performance for lightweight tasks locally, but you're just fighting economies of scale if you're g…

  22. comment
    Comment #49113922

    It's marketing hyperbole, but Luna is more intelligent per dollar than deepseek-v4 pro. Cost means nothing without the associated capability

  23. comment
    Comment #49113824

    Very interesting. Can you share more about your hypothesis/research pipeline? I have been using Sol for those types of task because I figured you'd need more reasoning for getting …

  24. comment
    Comment #49113799

    Burning the weights into silicon would be many orders of magnitude increase, not just 10x. It's kind of crazy that this hockey stick the AI hype bros talk about seems more and more…

  25. comment
    Comment #49113127

    I agree. I think part of the problem is how recommendation algorithms have conditioned people to act online. Recommendation algorithms optimize for overall platform engagement to i…