Havoc

23 ms·

631.
by Havoc 5mo ago
> But you cannot practice kindness towards a computer program. And yet rubber duck debugging is a thing
632.
by Havoc 5mo ago
Partial. I suspect the unspoken issue here is that in the context of EVs efficiency doesn't translate to higher price at pump (which would be fine for rich people) but rather range anxiety & severe impracticality.
633.
by Havoc 5mo ago
What baffles me is the range of estimates. Operating profit is both post depreciation and fees paid to third parties for hire. So aside from shenanigans like RSUs and financing interest that's already somewhat close to actual economics
634.
by Havoc 5mo ago
>I should split my $200 claude max plan into two $100 CC max and $100 codex plan You may want to get one of them to check the math on that :p
635.
by Havoc 5mo ago
Surprised it doesn't mention heat - one of the big advantages. There is also a new generation of "invisible" fiber that is interesting for home use. Basically looks like fishing line so you can run it somewhere that you just
636.
by Havoc 5mo ago
Have encountered this too - we really need new social norms around this. Bombarding others with pages of slop that took you 10 seconds to generate (and not even read) yet take minutes to untangle for the recipient is obviously downright rud
637.
by Havoc 5mo ago
It’s a good way of circumventing the algo feed. Which naturally is why Google has been making it ever harder to actually use it > I never saw the point of subscribing Guessing you’ve never used it extensively? My home page and subscripti
638.
by Havoc 5mo ago
The interviewer certainly didn’t mean tell me about your failed romantic relationships That’s a misread of note
639.
by Havoc 5mo ago
And by "we" you mean what random cities making up a couple percent of accommodation globally?
640.
by Havoc 5mo ago
There is some chat as to why it looks like that shape below [0]. In short no engine space constraint & aiming for efficiency rather than downforce. https://youtu.be/K-o0r2zSgCE?t=2162
641.
by Havoc 5mo ago
The numbers miss a key aspect - risk management Landlords do whatever the hell they like with zero consequences. Thats not a game I’d like to play with a 40 year horizon of unknowns. With a mortgage the risk is interest rates. And on that I
642.
by Havoc 5mo ago
Same for online feeds like YouTube. Good to occasionally clear out anything that hasn’t delivered good vids in a while
643.
by Havoc 5mo ago
These AI mandates are quite hard to actually push in some job. I’m in a finance role and thus far it’s all been rather hand wavy „use copilot more“. Maybe some meeting summarization. Nothing like the programming space where token counts mat
644.
by Havoc 5mo ago
No particularly strong feelings either way. It suffers from the same issue as all the other JS frontend - it evolved along with browsers over years and was constrained by whatever browsers happened to support rather than from a cohesive des
645.
by Havoc 5mo ago
> From a model’s standpoint, there are simply too many ways to write any of this They seem quite good at figuring this out in my experience
646.
by Havoc 5mo ago
Don’t think we’ll see similarly logically behaviour from LLM users tbh. A sizeable portion of the user base seems to insist on through opus at every trivial task
647.
by Havoc 5mo ago
An admirable principled stance that just doesn’t fucking work in the real world. Government processes and staff have zero interest in such stances. Next time you go through a border control try refusing to be searched or scanned on grounds
648.
by Havoc 5mo ago
>not to engage with systems such as these Yoti is used by governments. Principled stances are all good and well for hn comments but eventually collide with reality
649.
by Havoc 5mo ago
>The weakness of Rust WRT LLMs is compilation times. That's a more tractable problem then basically anything else around LLMs and programming. We're definitely getting more cores in the avg machine judging by roadmaps & lea
650.
by Havoc 5mo ago
This definitely has "cruelty is the point" vibes
651.
by Havoc 5mo ago
Just checked the stats on my opencode/DS usage...looks like 70%ish hit rate. Pretty shaky datapoint though...don't use it as primary model
652.
by Havoc 5mo ago
Neat. I like DS for secondary checks on code. Sometimes spots things other models don't
653.
by Havoc 5mo ago
I think the assumption that we're looking for an equivalence here is fundamentally flawed and with it the entire post. For most people income is tied to selling their time. It doesn't scale at all. Unless the income comes from wea
654.
by Havoc 5mo ago
Inference costs absolutely did fall. And even more so when looking at intelligence it buys you. eg compare say gpt 3.5 to latest deepseek. Both cheaper and more at more capable
655.
by Havoc 5mo ago
To be fair electron to rust would probably work even with LLM ram overhead
656.
by Havoc 5mo ago
Amused by all the comments about optimisation. If anything the vibe coding era will lead to the opposite
657.
by Havoc 5mo ago
And if you upgrade to standard workspace it still tells you account not eligible for antigravity Idk what google is doing
658.
by Havoc 5mo ago
End of an era
659.
by Havoc 5mo ago
Uk border control has confiscated them in the past
660.
by Havoc 5mo ago
>This account isn't eligible for Google Antigravity And clicking on "Explore Google AI plans" takes me to...I kid you not...the storage settings page of google drive. Genuinely can't tell wtf google even wants me to u

More ›