marcus

@nbaschez it's plan is already inherently lossy. like, if you plan something naturally, you usually pressure over it multiple times to reduce entropy. equally, if you ask it to find flaws, it will - that's it following your direction. if there's no off-ramp in your prompt - this happens

My biggest challenge with vibe coding / agentic engineering lately has been getting stuck in what I call a "plan doom loop" - have AI write a plan - review myself, seems good - have AI review plan, it always finds something - repeat It drains my time and energy to determine how important the "findings" really are Who has solved this

Y'all clowning on GitHub but the real lesson is that agents suck at scaling infra. Even the labs are struggling with their DBs. We're seeing the highest workloads in history but good architecture is currently still a human-bottlenecked activity. Make good infra choices.

@khushkhushkhush intellectual leverage

@sohan_zhang a long time ago, there was this one partner who kept rescheduling. i honestly just had fun just by continuously saying “yeah sure np”, even long after we finished raising. ofc it was a massive sign of no confidence, but it was alarming it lasted 6-8 months

one of the most frustrating experiences about the fundraising process is how decent respect gets thrown out the window one VC fund scheduled 3 meetings with myself and the team over 3 weeks, the last of which included an in person meetup that I took the courtesy of meeting them in the hotel they were staying at in SF. after a conversation, they mentioned that they would get back to me at the beginning of next week. cue 5 days later and i ask for an update. ghosted. i sent follow-up updates, as requested, over the span of a week, and still no response. i completely understand rejection, but it's very frustrating to take multiple meetings over weeks and get completely no response. future YCombinator founders, check the Bookface for investor reviews.

@MegaBasedChad db in shambles

would you like to run «incredibly specific command that is only useful in this instance» > 1 - yes > 2 - yes, always allow «incredibly specific command that is only useful in this instance»

Dynamics in software engineering is changing and it’s disruptive. Corporate empire builders are in the same position. The playbooks that worked for decades don’t work anymore. It’s the best time to be an engineer if you have some capital and autonomy.

here's how your company is rotting right this moment: - your senior devs stopped writing code - they ask Claude to generate it, they check that it mostly works, they ask a junior to approve the new PR - a junior who never had a chance to learn about architecture or read the docs can't really explain what you are doing wrong, so they blindly LGTM it - your senior devs stopped thinking - instead they "consult" Claude on making a bunch of strategic decisions; they ask the PM/principal to approve the new architecture - your PMs and principals are too busy (re)discovering the joy of producing 10k LOC, so they don't care if what you are doing is wrong, so they blindly LGTM it

this and $60 waymos each way

The people I want to hear from right now are the security teams at large companies who have to try and keep systems secure when dozens of teams of engineers of varying levels of experience are constantly shipping new features

new age hiring looks like 4-6 talented people clustering together to build a feature and getting acquihired

AI can make work faster, but a fear is that relying on it may make it harder to learn new skills on the job. We ran an experiment with software engineers to learn more. Coding with AI led to a decrease in mastery—but this depended on how people used it. anthropic.com/research/AI-as…

Remember the Pre AI days when tech was totally dead and the big new thing was a WebRTC wrapper

pro tip: never ever defend yourself online. it’s pure negative ev theater. self defense signals insecurity. pride is unreadable thru a screen. context collapses. mobs don’t update priors. you gain nothing & bleed attention. best move is to call yourself a moron, shrug, & move on. see my bio on here. online discourse is not a court, it’s a colosseum. it is not suitable for any sort of litigation.

@Kazanjy this is totally insane and runs against almost everything I consider best practices when it comes to running engineering operations. if your system can go down because of one engineer making a mistake, your system sucks and you have failed.

Medium-spicy take: whimsy is an excellent product principle BUT it has to happen during periods of mental *deload* One company that does this well is Anthropic with Claude Code Their choice of whimsy is brilliant. Claude's funny, weird thinky words while it's working on your query are delightful That's because the placement is after you've done the hard work of crafting a prompt, and before the hard work of evaluating an answer If Claude were whimsical while I was reading through it's answers, I'd hate it, probably

ship something shitty fast and fix it later culture just does not produce anything great there's mountains of rationale as to why you should do that but i've never seen it work and it always turns the company into a painful place to work nothing going on right now changes this

pov you rsvp’d “maybe” to the partiful

can someone help this poor man

why is pizza in the Bay Area such absolute trash like wtf, we nerds can cook up a recipe for a gajillion-parameter model no problemo but can't figure out a mildly appetizing ratio for some flour, water, and cheese