blinkingled

Yes 27B - not so much of a surprise but I was hoping for a usable coding model that's local and fast - this is fast but maybe not usable. I used Gemma4 26B before and it wasn't good in my testing.

I asked it to implement the simplest todo in ghostty codebase and it failed badly - kept going in circles fixing one compile error and introducing another. Qwen 3.8 was slower (40 tok/s compared to 75) but did get it done.

Ugh stop posting links dude - answer what do you mean by your previous claim for fixing everything you were complaining about earlier - "you should use open-weights models that run locally and "harnesses" built by members of the community" - like how is that not contradicting what you are saying - if open weights and local harnesses can do it Anthropic and OpenAI will do it faster and better. That doesn't make sense at all in this conversation.

I wasn't talking about fixing the tools or using open weight models - I don't need to watch anything to know that's an option. You are also now contradicting your earlier stance somewhat by saying what I was saying - LLMs can and will get better - you have now introduced an irrelevant open vs close element which doesn't really matter. There's no way only open weight models and harnesses get better - if you understand even a little, proprietary models with access to same open information but with billions to burn on research and hardware will get better faster.

I am not lowering myself or lionizing anyone - I am being practical about this. If we as a society have a workable stance for/against LLMs that is beneficial to all of us I am very open to hearing about it - It would however be fair to say that I am pessimistic given a lot of our collective history. And on oop of that nobody seems to be talking about any real solutions either this time around. Just being dismissive or being all-in are equally bad approaches.

Even after acknowledging that LLMs are fuzzy matchers and have been hyped, I am in the camp that rationally believes they will get better at some things including hunting bugs / chasing security vulnerabilities - my point also is that you and me can do whatever we can but on the grand societal scale we are not going to make a difference as far as slowing LLM adoption down - it just makes sense for some things and there's a lot of people with time and money that will make it do the rest - that's how it works until everything is saturated in the market.

So no I have no way to address anything you said - that was the point, I don't believe you can - not with regulation and not with voting with your money (I am sure some people tried to vote with their money to slow down mega stores and keep the mom and pop shop alive - there maybe some of those still there, but largely it's big chains occupying most of the market) - that stuff hasn't worked - heck LLMs work better today that that stuff has ever.

Just for the 'record' - I am totally with you on the hype and just in general the normalization of sleazy behavior surrounding it but I'm not sure we as normal people have any say anymore including where our money is going to be invested.

> he repeats some myths in the video, or at least confidently states some things that are not demonstrated to be true

I am genuinely curious what the myths/unproven things he states - I watched the video and it's repetitive sure but not much felt controversial to me.

Fair point, but he was talking about Mythos in particular and the point wasn't so much that LLMs will always have false positives rather he was saying they are causing a lot of them right now and how to deal with it.

Also as other replies said Linux kernel process is to assign CVE to everything - some of them may be just DDOSes, very hard to exploit and everything in between. All of them are bugs so they all get fixed and it's not a bad thing if distros ship those fixes and people update their kernel.

It's great to hear about $topic from someone no-nonsense and in-the-know like Greg KH. You can verify all of this too - since, well Linux kernel. (As opposed to what Microsoft or Apple claims to fix as far as LLM finds.)

Mythos may not be great today but it is not far fetched to imagine bug discovery, analysis and fixes can be made much quicker, accurate and even newly possible with specialized models trained on say Linux kernel specifics - with codemap/coding standards/threat models, good and bad coding patterns, tools to validate etc. an LLM can be much more relentless than humans and if it has the help to be accurate it will be worth the electricity burned. Oh and another model trained on triage data to validate the first one's findings would be good.

(I think Microsoft is doing this internally - different models trained internally alongside Mythos - there was some talk about it on the tubes, don't recall where exactly.)

> this work is focused on laptops with Snapdragon X2 Series. It does not currently cover desktop form factors, earlier Snapdragon X platforms, or other development boards. Readiness also varies by OEM design and Snapdragon X2 Series variant.

So same ARM BS as usual - AMD and Intel x64 are still the kings when it comes to out the door Linux support.

I'll buy a ARM PC when I can swap its motherboard CPU and GPU to a newer gen and my existing Linux install boots up without having to flash a new bootloader and custom vendor kernel with proprietary GPU blob.

As a KDE Connect user for a few years I know it doesn't always work without reconnecting especially with sleep cycles on the phone and the laptop but it does everything on Android at least - clipboard sync, notifications, sms, calls, remote control and media playback control. Must be iOS restrictions that made the author go through the admirable effort.

IOW Linux ecosystem has better vertical integration without any direct ties with Android :)

Different programming languages are very obviously not the same thing - different cp command implementations are similar conceptually to having different linker implementations that all do the same thing. But you knew that so not sure if there was a point you were trying to make there.

Writing linkers must be incredibly rewarding - go has its own, there's mold, there's LLD, there's the OG GNU bfd LD and now Zig has one too! I am sure there's a Rust one too - Wild!

Every one of them is faster than the others too lol! Mold for one tries really hard to be GNU ld and to be useful as an independent linker most have to - I guess Zig/Go ones are purpose built so at least those don't duplicate GNU ld compatibility.