MuffinFlavored

I wish people would pay more attention to:

* Anthropic is in some way trying to run a business (not a charity) and at least (eventually?) make money and not subsidize usage forever

* "What a steal/good deal" the $100-$200/mo plans are compared to if they had to pay for raw API usage

and less on "how dare you reserve the right to tweak the generous usage patterns you open-ended-ly gave us, we are owed something!"

How big of a handicap on performance is the external enclosure for something like an RTX5090?

> Running DeepSeek V3 (685B) requires 8×H100 GPUs which is about $14k/month. Most developers only need 15-25 tok/s.

> deepseek-v3.2-685b, $40/mo/slot for ~20 tok/s, 465 slots total

> 465 users × 20 tok/s = 9,300 tok/s needed

> The node peaks at ~3,000 tok/s total. So at full capacity they can really only serve:

> 3,000 ÷ 20 = 150 concurrent users at 20 tok/s

> That's only 32% of the cohort being active simultaneously.

Can you help me understand why devenv is needed instead of a shell like this/what is gained?

    { pkgs }:
    
    pkgs.mkShell {
      nativeBuildInputs = with pkgs; [
        # build tools
        cmake
        ninja
        gnumake
        pkg-config
      ];
    
      buildInputs = with pkgs; [
        # java
        jdk8
    
        # compilers
        gcc
        clang
        llvmPackages.libcxx
    
        # libraries
        capstone
        icu
        openssl_3
        libusb1
        libftdi
        zlib
    
        # scripting
        (python3.withPackages (ps: with ps; [
          requests
          pyelftools
        ]))
      ];
    
      # capstone headers are in include/capstone/ but blutter expects include/
      shellHook = ''
        export CPATH="${pkgs.capstone}/include/capstone:$CPATH"
        export CPLUS_INCLUDE_PATH="${pkgs.capstone}/include/capstone:$CPLUS_INCLUDE_PATH"
      '';
    }

really good question.

right now I have bought into the Nix koolaid a bit.

I have NixOS Linux machines and then nix-darwin on my Mac.

I use Nix to install Brew and then Brew to manage casks for things like Chrome what I'm sure updates itself. So the "flake.lock" probably isn't super accurate for the apps you described.

The world doesn't need a dozen JS runtimes.

The world doesn't need a dozen JS engines.

The world doesn't need many dozens of Linux distros.

The world doesn't need a handful of BSD distros.

The world doesn't need many dozens of package managers.

The world doesn't need hundreds of JS frameworks.

The world doesn't need dozens of programming languages or chat protocols or CI/CD systems.

The world doesn't need dozens of init systems, service managers, display servers, audio stacks, universal app formats, build tools/bundlers.

Deno may have dragged the JS runtime space forward, fully agree. Maybe it served its purpose and it is time to say goodbye.

> What’s next for Deno?

Who cares? Why does the world need so many fringe tools/runtimes? So much fragmentation. Why does every project have to be a long-term success? Put some stuff out if its misery. Don't waste the time of the already few open-source contributors who pour hours into something for no good reason.

any reason why you did

    const { rows } = client.query(
      "select id, name, last_modified from tbl where id = $1",
      [42],
    );

instead of

    const { rows } = client.query(
      "select id, name, last_modified from tbl where id = :id",
      { id: 42 },
    );

Is trading even a real thing? Is there really a job title "trader"? Entities really think they can outperform DCA SPY?

> Your machine runs a little slower, your bandwidth gets a little thinner, and someone halfway around the world is routing traffic through your home IP.

I wish in 2026 the default on new computers (Windows + Mac) was not only "inbound firewall on by default" but also outbound and users having to manually select what is allowed.

I know it is possible, it's just not the default and more of a "power user" thing at the moment. You have to know about it basically.

I always wondered this, is this true/does the math come out to be really that bad? 6x?

Is the writing on the wall for $100-$200/mo users that, it's basically known-subsidized for now and $400/mo+ is coming sooner than we think?

Are they getting us all hooked and then going to raise it in the future, or will inference prices go down to offset?

> Imagine if Siri could genuinely file your taxes

I do not like reading things like this. It makes me feel very disconnected from the AI community. I defensively do not believe there exist people who would let AI do their taxes.

Thank you. Could you give a tl;dr on "the full model needs ____ this much VRAM and if you do _____ the most common quantization method it will run in ____ this much VRAM" rough estimate please?

> You can shrink the model to a fraction of its "full" size and get 92-95% same performance, with less VRAM use.

Are there a lot of options how "how far" do you quantize? How much VRAM does it take to get the 92-95% you are speaking of?

Did you eventually move to a $20/mo Claude plan, $100/mo Claude plan, $200/mo, or API based? if API based, how much are you averaging a month?

and it has proven to not be a great hedge against inflation (lately, short term)

the digital gold narrative falls apart when non-digital gold outperforms it and nobody wants it for its digital properties (payments, blockchain, etc.)

Why does the news matter?

Sincerely, Somebody on "Hacker" News

I had a conversation (prompts) with Claude about this article because I didn't feel I could as succinctly describe my point alone.

Worth noting that the implied volatility extracted here is largely a function of how far OTM the strike is relative to current spot, not some market-specific view on $100. If NVDA were trading at $250 today, the options chain would reprice and you'd extract similar vol for whatever strike was ~45% below. The analysis answers "what's the probability of a near-halving from here" more than "what's special about $100." Still useful for the prediction contest, but the framing makes it sound like the market is specifically opining on that price level.

Trivial to build something that is opinionated and custom to you?

> the point is validating and advertising Polymarket on national news outlets will have ugly consequences.

Coinbase

Robinhood

DraftKings

FanDuel

How did Rhianna look him in the eyes and say "yes babe, good album, release it, this is what the people wanted after 7 years, it is pleasing to listen to and enjoyable"?

Is this the answer for citizens of countries other than America? I don't think it is.

I am surprised that... if they have a session ID of some sort for your chat that they make you re-send the entire message history each time? Not a single kind of cache or stateful proxy/buffering mechanism? Guessing that extra cost in bandwidth is cheaper than having to develop and maintain that? Seems kind of like an obvious optimization/design tradeoff they could eventually decide to change one day?

> He consistently communicated with shareholders of Berkshire in a straight-forward and transparent way in his letters and annual reports.

How often did he ever deliver "bad news"?

As in, you can now craft your own "update" and sign the bootloader/entire package and it will flash?

edit:

> You still won't get a jailbroken PlayStation 5 with this leak, but it will make it easier for hackers to compromise the console's bootloader.

nope?

I respect the work the Dolphin team (and any/every team for that matter) does.

I don't quite get, with 100,000+ video games to play across so many different platforms, why people choose to play specific ones in niche ways.

> OpenBSD uses 0xdb to fill newly allocated memory and 0xdf to fill memory upon being freed. This helps developers catch "use-before-initialization" (seeing 0xdb) and "use-after-free" (seeing 0xdf) bugs quickly.

Looks like this is the default in OpenBSD.

What % of effectiveness would you say is gained from these because... I am a pretty regular user of Claude Code in VS Code with no special goodies and I routinely hit "compacting" after 5-6 prompts in a single session. Then I need to validate it didn't slop all over the place. I can't imagine having 3-4 agents in the background (extra things to check) being a net-positive.

Skills and slash commands for sure but... I don't see them as necessary? "Review this code for ____ XYZ" as a skill. To me it's just something you could do as a prompt in your session?

What if I said "I don't use any of those and I think AI is very good at coding".

I'd be more interested in an article from you on how to go from "I use Claude Code out of the box" as a baseline and then how each extra layer (CLAUDE.md, skills, agents, MCP, etc.) improve its effectiveness.