baq

28 ms·

781.
by baq 8mo ago
Ideally all of them.
782.
by baq 8mo ago
XML is better than yaml. …note this doesn’t really say much. Both are terrible.
783.
by baq 8mo ago
yeah gemini is dumb when you tell it to do stuff - but the things it finds (and critically confirms, including doing tool calls while validating hypotheses) in reviews absolutely destroy both gpt and opus. if you're a one-model shop
784.
by baq 8mo ago
TACO isn't enough, Iran must also withdraw, this isn't a given if they feel they have nothing to lose
785.
by baq 8mo ago
you tell your clanker to do it obviously
786.
by baq 8mo ago
esim has you covered.
787.
by baq 8mo ago
Don’t confuse php the language with php the way of webmaster 2006 vintage.
788.
by baq 8mo ago
Multiple agents with different frontier models for best results. Claude code/codex shops don’t know what they’re missing if they never let Gemini roast their designs, code and formal models.
789.
by baq 8mo ago
Thank you!
790.
by baq 8mo ago
yes, exactly this, multiple -o. I sometimes have two or three branches which I keep a single merge branch on top of and being able to switch out the parents is super convenient.
791.
by baq 8mo ago
one thing I'm missing in jjui which jj cmdline does natively is rebase onto multiple heads - using this for quickly testing my branch on some other pr and latest main. other than that agreed, helps a lot with tedious noting down of cha
792.
by baq 8mo ago
jj/jjui should have you covered
793.
by baq 8mo ago
everything of importance ever done in EU was in response to a major crisis, in no small part because these exact voter emotions are dampened in such times.
794.
by baq 8mo ago
These aren't even their thoughts, it's just a bot let loose. I remember the first time I suspected someone using an LLM to answer on HN shortly after chatgpt's first release. In a few short years the tables turned and it&#x27
795.
by baq 8mo ago
You’re talking to someone’s clanker
796.
by baq 8mo ago
While I appreciate the morality and ethics of this choice, the current trend means projects going in this direction are making themselves irrelevant (don't bother quipping at how relevant redox is today, thanks). E.g. top security rese
797.
by baq 8mo ago
And what is there to replace it? macOS Tahoe?
798.
by baq 8mo ago
Docs and code work together as mutually error correcting codes. You can’t have the benefits of error detection and correction without redundant information.
799.
by baq 8mo ago
Start a defacto book club in a local cafe, that is, become a regular, show up with a book at approximately same day, same hour
800.
by baq 8mo ago
There’s a surge of demand for sure, but I’m not at all convinced that it’s at OpenAI’s expense. My bet is the non-swe folks caught wind the things got seriously good at a lot of boring office work, i.e. we’re seeing diffusion of AI into the
801.
by baq 8mo ago
Gemini schools the other two when doing code reviews. I used to think tokens are a commodity, but it’s becoming clear that the jagged frontier is different enough even for the easiest use case of SWE that there’s room for having two if not
802.
by baq 8mo ago
this is important from an operational security perspective and hence will probably pass. if they cared about actual insider trading they'd close that gap years ago.
803.
by baq 8mo ago
> major dieoffs in warm latitudes, even for humans, due to exceeding wet bulb limits my extremely pessimistic position is nothing will happen systemically even after the first few such events, and they'll take tens if not hundreds o
804.
by baq 8mo ago
tariffs on goods are mostly noise. if there were tariffs on services, though...
805.
by baq 8mo ago
Exactly, the $20 codex is so good value it’s irresponsible to not give it to everyone. Claude code $20 is otoh pointless, the limits are good enough for 10 mins of work twice per business day.
806.
by baq 8mo ago
Cursor sub from $DAYJOB.
807.
by baq 8mo ago
Somebody on Twitter used Claude code to connect… toys… as mcps to Claude chat. We’ve seen nothing yet.
808.
by baq 8mo ago
Gemini 3.1 slaps all other models at subtle concurrency bugs, sql and js security hardening when reviewing . (Obviously haven’t tested gpt 5.4 yet.) It’s a required step for me at this point to run any and all backend changes through Gemin
809.
by baq 8mo ago
such hopeful naivety was passable early 2025. having seen 1% of the epstein files, you'd have to be acting in bad faith to say there was no collusion.
810.
by baq 8mo ago
Lying implies knowing what’s true

More ›