HNHacker News
TopNewBestAskShowJobs

logickkk1

1,973 karma · joined April 28, 2026

submissionscomments
logickkk1··on What's the largest software project AI can complete on its own?
Yeah, it always wants to add another layer or append another fix. It almost never goes back and fixes the original thing in place.
logickkk1··on Claude Is Down
Fable’s been getting more stuff wrong than Opus for me lately. Now the whole thing is down too. Well, at least they’re consistent now.
logickkk1··on Same model, same Q4_K_M label: 5.02, 5.07 and 5.27 bits per weight
Yes, picchio monitors a running llama-server on a timer and flags when the prefill/decode ratio goes CPU-shaped. I hot-swapped one mid-run. probe 4 caught ENGAGED -> NOT ENGAGED.
logickkk1··on Norway imposes near ban on AI in elementary school
fwiw, Alpha School is the supervised version. the New York campus is $65k/yr and not legally a school.

private school money with homeschool paperwork and an app doing the teaching.

https://www.wired.com/story/alpha-schools-new-york-city-camp...

logickkk1··on Tell HN: Dont use Claude Design, lost access to my projects after unsubscribing
tbh, backups matter. but nobody would accept Word deleting your files when you cancel Office. somewhere along the way we stopped distinguishing backup from custody.
logickkk1··on Accelerating Gemma 4: faster inference with multi-token prediction drafters
i think this is mixing two separate ideas. MTP is the training-side piece. speculative decoding is the inference trick. DeepSeek V3 used MTP as an auxiliary loss. the 2022 Google paper is speculative decoding. now Google is combining them. https://arxiv.org/abs/2404.19737
logickkk1··on IBM didn't want Microsoft to use the Tab key to move between dialog fields
Funnily enough, IBM had already published this. CUA explicitly says tab and backtab move between fields.

so they spent seven layers of management escalating against their own standard: https://archive.org/details/ibmsj2703E/page/n13/mode/2up

logickkk1··on What I'm Hearing About Cognitive Debt (So Far)
fwiw, understanding was never really the goal. calling it "debt" assumes all of it needs to be repaid, but most of the time this just feels like brain OOM. imo the skill is knowing what actually needs to stay in your head.
logickkk1··on How OpenAI delivers low-latency voice AI at scale
IMO this probably isn't just about latency. keeping people in voice gives them training data text never will. is that why they were fine going transceiver over sfu and mostly ignoring multi-party?
logickkk1··on Stop big tech from making users behave in ways they don't want to
"just delete your account" assumes the exit was designed to work. this is the company that called its own prime cancellation flow "the iliad."
logickkk1··on Does Employment Slow Cognitive Decline? Evidence from Labor Market Shocks
I don't think this is "work is medicine." It's that too much of normal life depends on having a job, so policy lands on working longer.
logickkk1··on Agentic Coding Is a Trap
"Don't vibe code" but here's a deadline that's impossible without it. classic
logickkk1··on Why TUIs Are Back
if ai makes native rewrites cheap, "write once run anywhere" matters a lot less. tuis stick around for dev workflows, not for shipping to users.
logickkk1··on VS Code inserting 'Co-Authored-by Copilot' into commits regardless of usage
I already added a pre-commit hook to strip this out. Having to defend myself from my own editor is absurd.
logickkk1··on AWS stops billing Middle East cloud customers as repairs to war damage drag on
"Stops billing" makes it sound generous. If those regions can't run customer apps, not charging for them is just the minimum.
logickkk1··on Anthropic Joins the Blender Development Fund as Corporate Patron
I don't think it's just placating artists. Most Blender sponsors want more people getting good at Blender. Anthropic benefits if fewer people have to.