Ernest Ng

@Keleesssss Oh neat! I didn’t realize these were the slides for the paper

@ngernest2 Ah, I think I accidentally linked the presentation version of that! (wiki.haskell.org/wikiupload/7/7…)

New post! Everyone knows about the lava lambs, but very little cares about what we do with the randomness we acquire. I talked about pseudo random number generation, and then random structure generation via PRNGs. What is random generation?👇🏻 alperenkeles.com/posts/what-is-…

@krismicinski I was lucky to see Yihao’s talk in-person, it was great!! Super engaging and very easy to follow (cc @StarGazerMiao)

Photo someone sent me with Yihao's advisor's advisor in the front row.

I guess it’s a good time as any to announce it, our work on building a random testing system evolving with the database itself has been accepted to DBTest 2026! We’ve built a Property-Based Testing infrastructure for Turso in collaboration with its developers, brought existing properties from the literature into the system to the degree that when we stopped contributing, the testing infrastructure has continued to evolve as the database changed as several Turso developers kept developing the system. In the paper we analyze 23 bugs found by the testing tool we’ve worked on, but I presume the number is much higher today. Sincerest thanks to @glcst, @penberg, @superchicao10 , @LeMikaelF, @el_yawd and any other part of the Turso team!

I’m giving a talk at NJPLS (5/22 at Penn)! I’ll be presenting a DSL for specifying hardware communication protocols & some cool things you can do with it. Joint work w/ Nikil Shyamsunder, Francis Pham, Adrian Sampson & Kevin Laeufer njpls.org/may2026.html

I defended my PhD at MIT and will be joining Cornell CS as an assistant professor in January 2027, based at the Cornell Tech campus in NYC. I'll be living in Kazakhstan and Armenia this fall and would love to get connected with the local CS/tech community there!

@c20 Welcome to Cornell!! Excited to have another PL/systems faculty join the group!

@StarGazerMiao Congrats!!!

Synthesizing Backward Error Bounds, Backward Laura Zielinski, Justin Hsu arxiv.org/abs/2604.15633 [𝚌𝚜.𝙿𝙻]

Presenting 3 papers at @iclr_conf in Rio this week - main conference + 5 workshops - spanning both sides of LLM research: improving fundamental reasoning with RL, and deploying LLMs for AI for Science. See you in Rio! 🏖️

@krismicinski Looking forward to watching & learning from them!

Going to do some YouTube videos on building reasoning engines, theory, implementation, etc.

More new compiler hackers coming out of Cornell this spring. Reply here if you want to hire some of these talented folks!

Only one month left to register and propose talks for NJPLS this May at Penn: njpls.org/may2026.html

@aaroneline I am so sorry Aaron, my condolences to you and your family. May his memory be a blessing.

He was a filmmaker, a musician, and a magician. He ran a funeral home, and wound a town clock. I miss him.

I lost my dad this week. There's still some unreality to it. We were incredibly close. I took from him, among other things, a curiosity about the world. That there was nothing it wasn't worth wondering about. That the world, and its people, are fractally interesting. I miss him.

@nassaujuan thanks for making it!! (And for bringing it all the way from Baltimore!)

@ngernest2 I had planned to make it for over a year and finally got the chance!

The new frontal was used for the first time today :)

It's been a minute, but, time for another Signals and Threads, this one with Will Wilson about the testing tools he's building at @AntithesisHQ. And, in a bow to modernity, we have a video version of this one too!

In a topical sidequest, today's OCaml app is fuel.mint.caelum.ci.dev - using the UK's new fuel finding API to find the cheapest gas nearby (and detect price gougers?) tunbury.org/2026/03/13/fue…

@IoannesSilentio knowledge is justified true belief, or maybe what an LLM justifies to be true

me when I’m asked to do philosophy of AI

In the new Interlude, we focus on teaching. To help us, we are joined by Jamie Willis who is a Teaching Fellow at Imperial College London. The episode explores the benefits of live coding, and why Haskell is the best language for teaching programming. haskell.foundation/podcast/78/