my Anth number is definitely higher, probably 2T
OpenAI, 1T is right about where I would wait and see

praying for you that this doesn’t break following-feed containment

jerry wtf I thought you were just a shitposter; now you’re introducing me to new browser database libraries? you really can do it all

But you mentioned writing 1000s of lines of codes to get to your final 100 — do you find AI helps you in that stage at all?

The value I’ve gotten from tools like Cursor so far is that I can write more exploratory code than ever before, because the barrier to sketching up a new idea is so, so low.

which sucks because I think it's a meaningful question that specifically every SWE should start composing an answer for

personally, I think I sign up for community college woodworking or welding classes and start looking for a place where I can practice

How China Works: An Introduction to China’s State-led Economic Development How China Works: An Introduction to China’s State-led Economic Development [Lan, Xiaohuan, Topp, Gary] on Amazon.com. *FREE* shipping on qualifying offers. How China Works: An Introduction to China’s State-led Economic Development

"How China Works" by Xiaohuan Lan is pretty great. Speaks a lot to where the CCP accepts (and doesn't accept) capitalism/free-market ideology, which is pretty interesting

Super nitty gritty book but a casual read is worthwhile

www.amazon.com/gp/product/9...

he’s a bad guy, but there are gonna be more bad guys on this site, and this is a piss-poor way to deal with them

any game soundtrack lacking kazoo is flirting with failure, honestly

yeah, totally agree that just drawing it out over and over is just playing into the hands of bad actors

Like, okay, comparisons to SA are great and all, but Bluesky is already easily seeing 100x, probably 1000x more daily post volume than SA ever did.

And FB/X/Insta do 1000x more than that on a slow day!

How could you possibly apply "human judgement" to all of that, all the time?

How do you moderate at scale without *some* kind of flowcharts, though?

You literally couldn't hire enough people to apply extended, thoughtful, nuanced judgement to every single contested post/account. Ignore money; there aren't enough people capable. And you probably disagree with a lot of them!

how many monitors do you work with?

I've used two 24", one vertical + one horizontal, for a few years. recently been feeling like it might be nice to go down to just one screen, for (I think) similar reasons to you

> Chatbots may sound easy to use, but they really aren’t—and they’re getting harder to use all the time.

Really excited to see all the different UIs people come up with in the coming years to ease LLM interaction. Artifacts and Canvas are cool but can't wait for other takes; still feels wide-open.

Image of two blog posts with footnotes -- the second footnote is reusing the same content as the first one.

fyi that footnotes are a little wonky on your main page! seems like they work as intended on the individual post pages, though

what kind of stuff do you include in anki? been feeling a similar way but never really dedicated any time to building an anki deck/second brain/etc

Code reviews != team having mutual understanding of changes, though. *especially* mandatory, perfunctory code reviews on every change, which I think is maybe more what @skamille.bsky.social talking about.

I think the "mandatory" bit can turn people's brains off in a way that reduces understanding.

current Discover algorithm seems to do a pretty good job of serving me posts which are the top of good conversations, not just super high-like/rt

I wonder if this could be highlighted in UI to reinforce it more? "Good conversation below" banner sounds super cheesy but something like that