lisperati Profile Banner
Conrad Barski Profile
Conrad Barski

@lisperati

Followers
7K
Following
12K
Media
576
Statuses
12K

Clojure Coder - Medical Doctor - Illustrator - Land Of Lisp Author - Lisp Alien Creator - AI notkilleveryoneist

Mill Valley, CA
Joined April 2009
Don't wanna be here? Send us removal request.
@lisperati
Conrad Barski
19 hours
product idea: USB-D
1
0
10
@lisperati
Conrad Barski
19 hours
I've stayed there before. We would bow to the orb every morning, after leaving our room. (ht @denik)
@alexkehr
Alex Kehr
2 days
hot take: the hyatt regency hotel in san francisco might be one of the most beautiful buildings in the city
1
0
7
@lisperati
Conrad Barski
2 days
The hardest parts of programming are no longer "naming things and cache invalidation" For all the downsides of AI, it clearly is very good at picking stupidly boring, good names for variables/functions A big fault of programmers is that we get fancy with names
0
0
10
@lisperati
Conrad Barski
3 days
old man yells at cloudflare
0
2
25
@lisperati
Conrad Barski
4 days
working on my 2d polygon game engine One thing you learn very quickly when working with such an engine is that arbitrary polygon intersection detection is expensive Most of your CPU budget will go to poly intersect, and most of your time will be spent optimizing poly intersect
1
0
9
@lisperati
Conrad Barski
4 days
still can't get over how Google once started a sentence earlier this year: "While it may be that some of the strict prohibitions that were in the AI principles don’t jive well with those more nuanced conversations we’re having now ..."
1
0
2
@lisperati
Conrad Barski
5 days
Since it seems to become easier all the time to make curved/oddly shaped LCD and LED screens, I wonder how long until someone makes a retro LCD monitor that is indistinguishable from a CRT
0
0
4
@lisperati
Conrad Barski
5 days
my local coffee shop has keypad locks on the bathroom. when you ask for the code, they say "all odd numbers, then pound" there’s gotta be some small % of people for whom that’s a nontrivial puzzle I feel kinda sad for that hypothetical person, when they have to use the bathroom
4
0
7
@NemecekPanda38
Miroslav Nemecek
7 days
I created #AntCalc programmable wrist calculator. It uses RPN notation with 10-level stack, CH32V002A4M6 processor, precission 21 BCD digits, exponent +-99. Trigonometric, exp and log functions. 10 memories and 10 macro-programs with storage in flash. https://t.co/S6ryRdCmdU
11
55
431
@lisperati
Conrad Barski
6 days
(for those who aren't tech news addicts: he just bought a ton of Google)
0
0
0
@lisperati
Conrad Barski
6 days
It's insane that Warren Buffett, of all people, is making a big bet right now against the AI bubble narrative
1
0
2
@lisperati
Conrad Barski
6 days
I can tell Amazon/Audible has recently started relying on llms for their recommendations, because I love books about ancient Sumer, and now I'm getting lots of book recommendations with "Summer" in the title The shape rotator algo has become a wordcel algo
0
1
19
@alth0u
alth0u🧶
7 days
@zebulgar
delian
7 days
just read this in an investor update "older engineers who graduated from college pre-GPT are actually the best-suited for our purposes. They have fundamental programming ability that's lost amongst most of the current-gen." the AI-induced thinking/skills decay has begun wild
68
4K
61K
@lisperati
Conrad Barski
7 days
when the actual car arrived it was like "duck this ship, I'll just stop on the other side of the road"- the wait time went from 15min to 0min instantaneously It must be really, really hard to get the central coordinating computers to handle this as well as the local computers
0
0
0
@lisperati
Conrad Barski
7 days
it would be interesting to analyze robotaxis from the standpoint of local knowledge economics (aka "Hayekianism") I had a waymo that was supposed to pick me up in 15 minutes, but it was going to pass me in 1 minute, then spend 14 minutes to find an opportunity to turn around
1
0
2
@lisperati
Conrad Barski
8 days
My understanding is that self-driving systems currently use neural nets for decisions shorter than 30 seconds, versus traditional computer code for decisions longer than that. What neural nets are good at is subjective judgment calls, which are hard to program manually. This
0
0
0
@lisperati
Conrad Barski
8 days
so instead what waymos and Tesla FSDs currently do is they just turn on their blinker and slow down awkwardly and annoy the hell out of all the other drivers on the road. it's not a safety issue but just socially uncomfortable
1
0
1
@lisperati
Conrad Barski
8 days
it's hard for full self-driving systems because it's the most common action you need to do as a driver that has a time horizon larger than 30 seconds, and current car AIs run on shorter time horizons
1
0
1