aparker.io - Profile | ThreadSky | a Reddit-style client for Bluesky

aparker.io

the only thing worse than my code are my jokes governance @opentelemetry.io dir. open source @honeycomb.io av by @extinctinks.net

12,403 posts 7,603 followers 644 following

Posts 24 Comments 25

guess i have to clean up my act now

submitted 39 minutes ago • 0 comments

oof, right in the priors

submitted 5 hours ago • 1 comment

"Simulating human society requires more than generating plausible conversations. It requires simulating the structure of human reasoning." Debatable claim, because e.g. it's not clear we have great access to human reasoning. But this interesting position paper does try to address those objections.

submitted 7 hours ago • 4 comments

jellyfish aren’t either of those things

submitted 1 day ago • 16 comments

thinking of what sporting events would be fun if they’d happened in the Twitter/Bluesky era, and I just thought of Cal Ripken Jr. breaking the Ironman record while everyone posts “sorry, this fucking sucks, I’m not celebrating someone going to work every day”

submitted 21 hours ago • 86 comments

new LFX Insights is out in beta and open sourced for community feedback #kubecon #cloudnativecon insights.linuxfoundation.org

submitted 21 hours ago • 2 comments

plaudit is one of those words you really don’t read any more

submitted 22 hours ago • 0 comments

Hey ☁️ #cloudnative ☁️friends!! ✨ big news from Apple today ✨ Want a native way to build Linux containers on Apple hardware? Now you can ✅ — check out the new open source #containerization framework: 👏 github.com/apple/contai... 👏 github.com/apple/contai... Fun fact: it’s written in @swift.org 🙌

submitted 1 day ago • 2 comments

who am I gonna see at @renderatl.com this week?

submitted 1 day ago • 1 comment

lol this is what happens when your licensing changes wind up fucking over a multinational development org i guess

submitted 1 day ago • 1 comment

hope you like how this design language looks because it's all you're gonna see for the next fucking half a decade #wwdc

submitted 1 day ago • 2 comments

starting off your developer conference for an ad for your movie arm is, uh, very 2025

submitted 1 day ago • 0 comments

chappel roan chappel roan as administrator

submitted 1 day ago • 1 comment

shots fired www.honeycomb.io/blog/its-the...

submitted 1 day ago • 2 comments

there’s a really interesting tension at the end of this piece that I also dont have an answer for — inference is the cheapest part of an agent, but the value is all in adjacent services, and those services optimize for humans. www.snellman.net/blog/archive...

submitted 1 day ago • 3 comments

Every time I see a thread on here about the "possibilities of derivative functions on an inverse Browns-Johnson curve" or whatever.

submitted 2 days ago • 1 comment

I've fixed the app

submitted 2 days ago • 1 comment

switch 2 this switch 2 that you should switch 2 good posts

submitted 2 days ago • 2 comments

some days you really do need a miter saw

submitted 2 days ago • 1 comment

this fits my mental model — LLMs do learn procedures. But it’s the same mechanics as what’s learning facts. So of course it would also hallucinate procedures but also: what does procedure hallucination look like? i don’t think i have a grasp on that machinelearning.apple.com/research/ill...

submitted 2 days ago • 2 comments

guess i have to clean up my act now

oof, right in the priors

"Simulating human society requires more than generating plausible conversations. It requires simulating the structure of human reasoning." Debatable claim, because e.g. it's not clear we have great access to human reasoning. But this interesting position paper does try to address those objections.

jellyfish aren’t either of those things

thinking of what sporting events would be fun if they’d happened in the Twitter/Bluesky era, and I just thought of Cal Ripken Jr. breaking the Ironman record while everyone posts “sorry, this fucking sucks, I’m not celebrating someone going to work every day”

new LFX Insights is out in beta and open sourced for community feedback #kubecon #cloudnativecon insights.linuxfoundation.org

plaudit is one of those words you really don’t read any more

who am I gonna see at @renderatl.com this week?

lol this is what happens when your licensing changes wind up fucking over a multinational development org i guess

hope you like how this design language looks because it's all you're gonna see for the next fucking half a decade #wwdc

starting off your developer conference for an ad for your movie arm is, uh, very 2025

chappel roan chappel roan as administrator

shots fired www.honeycomb.io/blog/its-the...

there’s a really interesting tension at the end of this piece that I also dont have an answer for — inference is the cheapest part of an agent, but the value is all in adjacent services, and those services optimize for humans. www.snellman.net/blog/archive...

Every time I see a thread on here about the "possibilities of derivative functions on an inverse Browns-Johnson curve" or whatever.

I've fixed the app

switch 2 this switch 2 that you should switch 2 good posts

some days you really do need a miter saw

this fits my mental model — LLMs do learn procedures. But it’s the same mechanics as what’s learning facts. So of course it would also hallucinate procedures but also: what does procedure hallucination look like? i don’t think i have a grasp on that machinelearning.apple.com/research/ill...

Happy three year anniversary to Xbox saying Silksong would be out within 12 months!

I just write them I don't explain them.

meanwhile yuds appears to be sephiroth posting about the apple/potato thing