experquisite
experquisite.bsky.social
experquisite
@experquisite.bsky.social
Quant/dev/trading, enjoy Rust, FPGA hacking, and more
owning exactly the hardware I want is a cozy feeling, like throw pillows on a sofa.
October 25, 2025 at 12:52 PM
I’m back.
October 19, 2025 at 3:43 PM
100% agree. The type system and error messages of Rust, for instance, make 80% of the issues of LLM nonsense irrelevant - it either works, or it doesn't, and you can iterate on the error messages productively until you get it to work or figure out why it's impossible.
I'm increasingly optimistic for the future of agentic coding for languages with strong static guarantees. Type systems just provide an awesome form of feedback for the RL process.

This came up a little in my conversation last year with Richard Eisenberg.

signalsandthreads.com/future-of-pr...
Signals and Threads Podcast
Listen in on Jane Street’s Ron Minsky as he has conversations with engineers working on everything from clock synchronization to reliable multicast, build systems to reconfigurable hardware. Get a pee...
signalsandthreads.com
March 2, 2025 at 2:44 PM
one my of my biggest challenges is learning how to drop what I am doing and do something else, particularly when the first thing is, "this bit of deep work that I am juggling in my mind", and the second thing is, "my 5yo son wanting to play magnatiles".

it's so hard, but I must get better at it.
March 2, 2025 at 2:29 PM
First thing I did after waking up in 2025 was make some art and do a demo for my nieces and nephews. Hope this trend continues.

(It won’t)
January 1, 2025 at 6:52 PM
Reposted by experquisite
If you are going to POPL 2025 in Denver consider coming to my talk at #PEPM2025 on the missing diagonal.
The computing community has produced many high level languages and tools for programming high level systems (e.g. Java for user interfaces)
popl25.sigplan.org/details/pepm...
The Missing Diagonal: High Level Languages for Low Level Systems (Invited Talk Abstract) (PEPM 2025 - The 2025 ACM SIGPLAN Workshop on Partial Evaluation and Program Manipulation) - POPL 2025
The ACM SIGPLAN Workshop on Partial Evaluation and Program Manipulation (PEPM) has a history going back to 1991 and has been held in conjunction with POPL every year since 2006. The origin of PEPM is ...
popl25.sigplan.org
December 13, 2024 at 12:09 AM
Reposted by experquisite
December 4, 2024 at 10:35 PM
I’m thankful for our family, our careers and our health. I’m thankful I quit smoking, drinking and drugs over ten years ago. I’m nevertheless thankful I had those experiences, though I could have done with fewer of those lost years.
November 27, 2024 at 11:44 PM
Reposted by experquisite
A long time ago I wrote an article for Communications of the ACM called "Computing without Processors" which promoted a heterogenous world of computing. dl.acm.org/doi/pdf/10.1...
dl.acm.org
November 22, 2024 at 5:26 PM
so what ever happened to mastodon
May 30, 2023 at 1:21 AM
thank god HD VR didn’t exist when I was playing WoW 60 hours a week or I would have almost flunked out of college.

shit, wait
May 11, 2023 at 1:03 AM
Followup question:

Why is all the software you have to maintain shitty hacker code instead of perfect clean code?

Hint:
May 2, 2023 at 12:59 PM
Software takes so long if it’s carefully planned by many people, but can be so fast if it’s 1-3 hackers. Is the former better code? Maybe.

But what is the opportunity cost of doing one thing perfectly vs 10 things shitty? Or how shitty will it be after you redo it 10 times?

Was it even perfect?
May 2, 2023 at 12:46 PM