smcleod

5 ms·

1. 31.▲
   by smcleod 2mo ago

   Japanese software and software technology culture is amongst the worst in the world. They're famous for failing to adopt modern ways of working, tools and as an industry is seriously struggling. There's fantastic video in this fro
2. 32.▲
   by smcleod 3mo ago

   You know your phone is OLED right? I'm assuming you left a still image on without the inbuilt screen refresher enabled or something? I remember seeing that way back in the early OLED days but not on good TVs in the last 8~ years.
3. 33.▲
   by smcleod 3mo ago

   Really? I've had many updates over the 7 years since my c9 was released, still getting them as recent as earlier this year I think.
4. 34.▲
   by smcleod 3mo ago

   7 years into my C9 and it never fails to impress.
5. 35.▲
   by smcleod 3mo ago

   If you care about home cinema it's important. There's a pretty big jump down. You can look up the specs and reviews of the C and G series from LG. Anecdotally quite a few times when people see my 7 year old LG C9 I've had the
6. 36.▲
   by smcleod 3mo ago

   I suspect the entire 3.x family is fundamentally problematic and we'll need to see an architecture change before they're half decent like back in the 2.5 days again.
7. 37.▲
   by smcleod 3mo ago

   I'm aware of that setting, but this is the Samsung baseline. The panels they export for OEMs / other manufacturers don't seem to have the same hyper saturation. Even just standing next to someone using a Samsung phone or tabl
8. 38.▲
   by smcleod 3mo ago

   Their panels are often over-saturated with that fake HDR like look, similar to their phones.
9. 39.▲
   by smcleod 3mo ago

   Unfortunately they make arguably the best OLEDs and webOS is pretty decent to use (especially compared to the terrible OS that Samsung and Sony run). I think the best thing people can do is update their firmware then disconnect it from the
10. 40.▲
    by smcleod 3mo ago

    I haven't used Codex in some time, but in Claude Code I force my context size down to 280k on purpose as in my experience your best work is done within the 30-250k~ range regardless of a models maximum context size.
11. 41.▲
    by smcleod 3mo ago

    It's 2.8T, I'm sure they will open the weights but it will only be able to be run on very high end machines.
12. 42.▲
    by smcleod 3mo ago

    And American providers, not sure if it's still the case but OpenAI were doing this.
13. 43.▲
    by smcleod 3mo ago

    Soulseek is still going hard.
14. 44.▲
    by smcleod 3mo ago

    I suspect there's some cargo culting, and some folks that are generally more likely to table flip than understand things that challenge their workflow.
15. 45.▲
    by smcleod 3mo ago

    Yes their docs cover how to properly prompt it to ensure you don't get hit by the classifier, you can distil those down to a skill to help build prompts for the more challenging situations, e.g. https://github.com/sammc
16. 46.▲
    by smcleod 3mo ago

    I've been working with it heavily since its first release. I use it for software architects, complex debugging and some development and I have not had it refuse or downgrade even once.
17. 47.▲
    by smcleod 3mo ago

    Yeah the 256GB/s bandwidth is really very limiting
18. 48.▲
    by smcleod 3mo ago

    ultracode in Claude Code kicks off a dynamic workflow.
19. 49.▲
    by smcleod 3mo ago

    This looks very much like many slopped up UIs I've seen over the past few years. Not saying the code is, but the design itself looks vibe coded?
20. 50.▲
    by smcleod 3mo ago

    Weird they talk about their 31B dense model but haven't actually released it anywhere.
21. 51.▲
    by smcleod 3mo ago

    But does anyone actually need a Windows?
22. 52.▲
    by smcleod 3mo ago

    They canned the moved to make -p commands API billable.
23. 53.▲
    by smcleod 3mo ago

    Material design looks pretty dated these days. I'm wondering why people would still be using it? Is it just a taste thing or something people have been working with for a long time?
24. 54.▲
    by smcleod 3mo ago

    Qwen as well.
25. 55.▲
    by smcleod 4mo ago

    Looks decent but $250 AUD for a font? Even for local and personal use? That's... a lot. I was thinking if it is paid and it was around $25 I'd consider it, then I saw the price!
26. 56.▲
    by smcleod 4mo ago

    3.5's 122-A10B is still great!
27. 57.▲
    by smcleod 4mo ago

    35b-a3b is only 3b active parameters, it's a MoE.
28. 58.▲
    by smcleod 4mo ago

    Handy, Cotypist, Markedit, Ghostty Tip, BetterTouchTool, GoodLinks, Pixelmator Pro, Handbrake, Zed, Onyx, LittleSnitch (or LuLu), Chezmoi, Stats, Shottr, llama.cpp, oMLX, LM Studio, Parcel, Things, OrbStack, Msty Studio.
29. 59.▲
    by smcleod 4mo ago

    Try 27b, it's significantly smarter than 35b-a3b (although it is slower, it's not so bad with MTP).
30. 60.▲
    by smcleod 4mo ago

    It's sort of a form of corrective antithesis (or "negation-antithesis") I think. A bit like "it's not x, it's why". Really grinds my gears.