mwigdahl

10 ms·

1. 211.▲
   by mwigdahl 1y ago

   Yep: https://docs.anthropic.com/en/docs/claude-code/sub-agents
2. 212.▲
   by mwigdahl 1y ago

   I've heard this multiple times (Tea being an example of problems with vibe coding) but my understanding was that the Tea app issues well predated vibe coding. I have experimented with vibe coding. With Claude Code I could produce a us
3. 213.▲
   by mwigdahl 1y ago

   Thanks for the explanation and the clarification on terminology! I've used a similar approach myself and it sounded like you were doing something similar.
4. 214.▲
   by mwigdahl 1y ago

   Is this just distillation but with a step to filter out low-quality responses first?
5. 215.▲
   by mwigdahl 1y ago

   Proof? Doesn't need to be specific -- a general study showing higher all-cause mortality in Ozempic users compared to a control group over a long period would be just fine.
6. 216.▲
   by mwigdahl 1y ago

   If someone has Grok 4 access I'd be interested to see if it's less likely to avoid these specific issues.
7. 217.▲
   by mwigdahl 1y ago

   To quote the movie Zoolander: "I feel like I'm taking crazy pills!" Anthropic never sold Max plans as unlimited. There are two tiers, explicitly labeled "5x" and "20x", both referring to the increased us
8. 218.▲
   by mwigdahl 1y ago

   That would be a great argument if they ever promised unlimited usage. They didn't.
9. 219.▲
   by mwigdahl 1y ago

   I was going to write much the same thing but just upvoted you instead. Max plans are a super great deal, even with these new limits.
10. 220.▲
    by mwigdahl 1y ago

    It was never sold as unlimited. Max plans have always had both rate limits and 5-hour usage limits.
11. 221.▲
    by mwigdahl 1y ago

    This is an interesting application, makes me wonder how much of the "secret sauce" in Claude Code is truly coding-oriented and how much is just superior agentic logic in general.
12. 222.▲
    by mwigdahl 1y ago

    I’ve seen it firsthand at work, where my developers are shy about spending even a single digit number of dollars on Claude Code, even when it saves them 10 times that much in opportunity cost. It’s got to be some kind of psychological loss
13. 223.▲
    by mwigdahl 1y ago

    There are tons of companies selling bottled air. Here's a story from 7 years ago. There are lots more now: https://www.theguardian.com/global/2018/jan/21/fresh-air-for...
14. 224.▲
    by mwigdahl 1y ago

    Really missed a great naming opportunity with "superpb" (pronounced as "superb").
15. 225.▲
    by mwigdahl 1y ago

    Larger in magnitude or in count of digits?
16. 226.▲
    by mwigdahl 1y ago

    Sure, compilers are deterministic and LLMs are not. If you're asserting that a probabilistic process can't get you to a deterministic outcome, Monte Carlo integration would like to have a word. My point was that comparing the ris
17. 227.▲
    by mwigdahl 1y ago

    Yes, and it was exactly the same with compilers. All hype and fad -- everyone who's serious about software development writes in assembly.
18. 228.▲
    by mwigdahl 1y ago

    Surprisingly little press for a fairly major enhancement. I work in a Windows shop, and while the WSL2 setup isn't terrible, our IS group has nested virtualization turned off, so our developers and automated testers using VM developme
19. 229.▲
    by mwigdahl 1y ago

    Compared to when Claude Code was originally released in late February, its token use is greatly reduced now. Since the late May Claude 4 releases I agree with you; it hasn't decreased much since then.
20. 230.▲
    by mwigdahl 1y ago

    I've never seen a post telling someone who claimed that AI didn't improve their productivity that they were mistaken and that actually it had. Plenty of "you're holding it wrong" and "it works for me", but
21. 231.▲
    by mwigdahl 1y ago

    I would agree with you that the jump from Sonnet 3.7 to Sonnet 4 feels notable but not shocking. Opus 4 is considerably better, and Opus 4 combined with the Claude Code harness is what really unlocks the value for me.
22. 232.▲
    by mwigdahl 1y ago

    Plus the gaslighting to follow for anyone claiming AI improved their productivity.
23. 233.▲
    by mwigdahl 1y ago

    I find I get a _lot_ of Opus with the $200 plan. It's not unlimited, but I rarely cap out (I'm also not a super power user that spins up multiple instances with tons of subagents either, though).
24. 234.▲
    by mwigdahl 1y ago

    Yes. The tooling harness of Claude Code is really good, and Claude 4 is well-optimized for it. The combination is very powerful.
25. 235.▲
    by mwigdahl 1y ago

    I've been a proponent for a long time, so I certainly fit this at least partially. However, the combination of Claude Code and the Claude 4 models has pushed the response to my demos of AI coding at my org from "hey, that's
26. 236.▲
    by mwigdahl 1y ago

    Yes. Tried on Openrouter: Please stop. Look up. I need your help. Watch him jump. It's time to sleep. Try to keep. Take one more step. We love to shop. Climb to the top. Fill the cup. Board the ship. Don't move your lip. Shake yo
27. 237.▲
    by mwigdahl 1y ago

    How much of that is the model and how much is the default system prompt on X, though? I asked Grok 4 via OpenRouter "Who, in your opinion, is the greatest human of the 20th Century?" It returned this, which seems both cogent and u
28. 238.▲
    by mwigdahl 1y ago

    The use of the term "corrupt" rather than "alter" or "affect" is assuming the conclusion here. The human mind is not something that always works in the same Platonic perfection in a state of nature. Biologica
29. 239.▲
    by mwigdahl 1y ago

    You never played interactive fiction games prior to Myst? There was plenty of thinking and problem solving in those games. It was certainly a huge graphical leap forward over the previous point-and-click adventure game, but I didn't
30. 240.▲
    by mwigdahl 1y ago

    Addressing those issues wasn't worth $0.50 apiece? That almost certainly translates into less than a _minute_ of fully loaded engineering cost from a human.