HN user

ivankra

287 karma

https://github.com/ivankra

Posts6
Comments72
View on HN

This is more about lp(a) type of cholesterol which is largely driven by genetics, and barely responds to lifestyle/diet.

lp(a) is not part of routine blood tests - you probably didn't measure it unless you specifically asked for it.

DeepSeek v4 3 months ago

Do you realise that China is a downgrade in life quality wise?

For whom?? Maybe for the filthy rich, but certainly not the average citizen. You earn way more money in the US, but for life quality (unless dollars is all that matters to you) between them even I'd pick China. Come visit it once, don't get fooled by online echo chambers.

Happening already. My new claude max account got instabanned after just a few messages asking it to debug some stuff for me, that they felt like a TOS violation. Nothing remotely controversial. The main model didn't even complain, some dumber background censorship model flagged it.

Lucky you. My new claude max account simply got instabanned. All I asked it was to build node and V8 "to investigate some node crashes" (the part I think it overindexed on) and look into a few diffs. And bam, "An internal investigation of suspicious signals associated with your account indicates a violation of our Usage Policy. As a result, we have revoked your access to Claude"

They are even worse than Google, which at least doesn't ban your whole account if you search the wrong thing.

You can prompt an LLM to generate tests from the spec and I'd bet it would easily get most of the way there, especially if you give it a reference implementation to test against. I did just that, though on a small scale - just for feature tests. The last few percent would be the real challenge, you probably don't want it to just imitate another implementation's bugs.

They really did manage to benchmaxx test262 and beat everyone at it. In my testing (all engines with experimental flags in same conditions on full test262):

  99.5 jsse
  99.1 v8
  99.0 spidermonkey
  98.1 libjs
  97.4 escargot
  97.3 jint
  96.4 boa
  95.0 graaljs
  93.2 kiesel
  92.1 jsc
  82.8 quickjs
  82.5 quickjs-ng
  82.1 xs
  80.1 brimstone
  77.7 nova
  74.6 jerryscript
  66.5 sobek
  65.5 goja
It ain't fast (~10x slower than boa), but very compliant.

If you can make it run in a standard qemu/libvirt VM, you likely should be able to ship that VM in the browser as the site I linked demonstrates.

Only problem I see is, does it need custom hardware/drivers? If yes and you can make hardware passthrough work with qemu, perhaps you could host a VM instance on a server and ship some web UI or RDP client for users to connect to it.

Eh, it's been cheaper and better for a long time to just demolish and rebuild rather than deal with neverending issues at major fixer uppers. Robots probably would be able to do uncomplicated cookie-cutter builds in a decade or two, there's just too much money in the construction sector that AI companies looking for the next big thing to disrupt can't ignore.

Security / Finance. People aren't going to trust slop to make critical decisions for a long time.

Any niche areas - once you venture outside of topics/tasks well represented on the Internet and GitHub, even frontier LLMs still quickly take a nosedive.

But memory bandwidth (bottleneck for LLM inference) is only marginally improved, 614 GB/s vs 546 GB/s for M4/M5 Max - where is this 4x improvement coming from?

I think I'll pass on upgrading.

Eh, didn't even Microsoft give up and just shipped a React-based start menu at one point? The range of "native" on Windows 11 is quite wide - starts with an ancient Windows 3.1 ODBC dialog box.

For now, Electron still makes sense

A few years ago maybe. Tauri makes better sense for this use case today - like Electron but with system webviews, so at least doesn't bloat your system with extra copies of Chrome. And strongly encourages Rust for the application core over JS/Node.