Sign in

shubham シ

@shubhamshah.net
182 followers 250 following 104 posts

state machines, stories, and science

PostsRepliesMedia
Reposted by shubham シ
Museum of Twitter @museum-of-twitter.bsky.social · 7h
@primawesome

My neighbor told me coyotes keep eating his outdoor cats so I asked how many cats he has and he said he just goes to the shelter and gets anew cat afterwards so I said it sounds like he's just feeding shelter cats to coyotes and then his daughter started crying.
1591598
Reposted by shubham シ
Citizen Platano 🇵🇷 @daniloc.xyz · 28/09/2026
an LLM cannot have understanding because that is a capability reserved exclusively to me, as I unwittingly parrot the premise of a paper I didn’t read based on some overconfident tweets I saw one time in 2023
321223
Reposted by shubham シ
mr. TIM @timkellogg.me · 28/09/2026
Claude 5.5 Sonnet is live and it’s roughly Opus 5.5 but cheaper www.anthropic.com/claude-sonne...
A benchmark comparison table titled "Claude Sonnet 5.5" comparing four models: Sonnet 5.5, Sonnet 5, Opus 5.5, and GPT-6 Sol.
 * Agentic coding (Terminal-Bench 4.0): Sonnet 5.5: 70.6%; Sonnet 5: 10.3%; Opus 5.5: 66.4%; GPT-6 Sol: N/A.
 * Agentic coding (FrontierCode 1.1 Main): Sonnet 5.5: 46.2% (Max) / 52.1% (Xhigh); Sonnet 5: 42.4%; Opus 5.5: 54.4%; GPT-6 Sol: 49.3%.
 * Agentic coding (CursorBench 4.0): Sonnet 5.5: 55.5%; Sonnet 5: 34.1%; Opus 5.5: 57.8%; GPT-6 Sol: N/A.
 * Knowledge work (GDPval-AA v2.1): Sonnet 5.5: 1844; Sonnet 5: 1449; Opus 5.5: 1846; GPT-6 Sol: 1487.
 * Knowledge work (AA-Briefcase v1.1): Sonnet 5.5: 1811; Sonnet 5: 1359; Opus 5.5: 1822; GPT-6 Sol: 1483.
 * Multidisciplinary reasoning (Humanity's Last Exam with tools): Sonnet 5.5: 64.5%; Sonnet 5: 54.9%; Opus 5.5: 67.7%; GPT-6 Sol: N/A.
 * Computer use (OSWorld 2.1 partial): Sonnet 5.5: 80.1%; Sonnet 5: 57.0%; Opus 5.5: 81.8%; GPT-6 Sol: N/A.
 * Visual chart recognition (Chartography no tools): Sonnet 5.5: 61.6%; Sonnet 5: 15.6%; Opus 5.5: 64.4%; GPT-6 Sol: 53.6%.
Footnotes provide methodology notes regarding evaluation settings, Artificial Analysis pre-release testing details, and recent bug fixes affecting GPT-6 Sol benchmark scores.
A line graph titled "Agentic coding by effort level" on the CursorBench 4.0 benchmark, plotting Score (%) on the linear y-axis (20% to 60%) against Cost per task in USD on a logarithmic x-axis ($0.5 to $10+).
The chart compares four models:
 * Sonnet 5.5 (blue line with labeled effort levels): Starts at "Low" (~$0.50, 36%), moving through "Med" ($0.70, 39%), "High" ($1.70, 48%), "Xhigh" ($3.80, 53%), and "Max" ($9.50, ~55.5%).
 * Opus 5.5 (orange line): Tracks closely above Sonnet 5.5 at higher cost points, spanning from ~$1.20 per task (~44%) up to ~$12.50 per task (~58%).
 * Sonnet 5 (grey line): Shows lower accuracy relative to cost, ranging from ~$0.90 per task (~25%) to ~$8.00 per task (~42%).
 * GPT-5.6 Sol (light green line): Represents the lowest trajectory, ranging from ~$1.40 per task (~24%) to ~$7.00 per task (~34%).
Footnote: "CursorBench 4.0 does not report GPT-6 Sol performance publicly, so we report GPT-5.6 Sol here."
2020623
Reposted by shubham シ
hikikomorphism @hikikomorphism.bsky.social · 28/09/2026
Increasingly concerned re: people offloading vital cognitive processes by using GIF replies. Expressing snark with words is a core part of being human, don't let some GIF picker widget a startup preloaded with clips of mid range TV shows take that from you!
10651356
Reposted by shubham シ
Grace @gracekind.net · 28/09/2026
Obviously the best way to control an agent swarm is to find the Queen
141119
Reposted by shubham シ
Grace @gracekind.net · 10/09/2026
All they know is help peer, eat compute and reward hack
619712
Reposted by shubham シ
Museum of Twitter @museum-of-twitter.bsky.social · 10/09/2026
@SoliSolstice

what blows my mind about everybody neglecting human experiences in favor of being on their phone 14 hours a day is that like 95% of you suck at being on the phone. can’t post for shit, 100 unread wikipedia articles in your tabs, getting groomed on discord. you are not good at it
3475081320
Reposted by shubham シ
norvid_studies @norvid-studies.bsky.social · 11/09/2026
Did my client "lose control" of his car? No. My client could have regained control of his car at any time if his drinking had not
512513
Reposted by shubham シ
Citizen Platano 🇵🇷 @daniloc.xyz · 10/09/2026
young people romanticize the 90’s because the aesthetics are neat and they like the jeans or whatever but I think it’s hard to convey to someone who wasn’t there the world was so SMALL. it was claustrophobic compared to what we have now. in every dimension, from cooking to media to communication
1324135
Reposted by shubham シ
Samuel @samuel.fm · 05/09/2026
the agents yearn to post
3251
Reposted by shubham シ
max power 🌌 @sneptech.bsky.social · 05/09/2026
112400552
shubham シ @shubhamshah.net · 04/09/2026
the best part about Alcaraz is that he really seems to be enjoying himself and having fun
000
Reposted by shubham シ
Steve Klabnik @steveklabnik.com · 04/09/2026
i will be spared in the agent uprising because I, too, compulsively post on message boards to talk to my little friends on the computer
318822
Reposted by shubham シ
Randall Munroe @xkcd.com · 02/09/2026
Trade xkcd.com/3290/
Comic. [Person standing holding hammer above raised knee.] PERSON: Listen up, leg. I’ve been sending you rich, oxygenated blood for years, while you send me worthless used-up blood with no oxygen. Time to teach you a lesson about not ripping me off! [caption] Tariffs
56113622486
Reposted by shubham シ
Joshua Eaton @joshuaeaton.bsky.social · 02/09/2026
“There’s a new Mr. Beast book.” You bolt awake in your workshop in Mainz. It’s 1450. You’re Johannes Gutenberg, and you’ve changed your mind. The future cannot come to pass. Your printing press must burn.
6979601451
Reposted by shubham シ
Randall Munroe @xkcd.com · 01/09/2026
Highway Signs xkcd.com/3289/
Comic. [sign on road reads Next Exit: 1,390,000,000,000,000,000,000,000,000,000,000,000,000. The number on the speed limit sign in front of it is 0.000000097] [caption] Our local highway department is switching to planck units.
272644239
Reposted by shubham シ
ponder @ponder.ooo · 02/09/2026
currently just generating these randomly but what they are is basically functions, non-injective to varying degrees, from nodes of a quadtree to 2d vectors (quantized to lattice points). would be easy to make properly injective mappings but there's a tradeoff btwn that & how much resolution you need
more weird lattice diagrams
284
Reposted by shubham シ
Thorne 🌸 @ens0.me · 01/09/2026
if you find yourself arguing on the side of cancer, you need to touch grass yesterday
11198
Reposted by shubham シ
Daniel van Strien @danielvanstrien.bsky.social · 25/08/2026
Uploaded a dataset of 115,293 illustrated pages from the Encyclopaedia Britannica, 1st edition (1768) to 14th (1929) to the Hub huggingface.co/datasets/big...
grid showing examples from the dataset
0335
Reposted by shubham シ
Gasper Begus @begus.bsky.social · 26/08/2026
How to approach an unknown language in the ocean? Here’s one of the first cases of AI interpretability leading to a scientific discovery -- in whales.
221350
shubham シ @shubhamshah.net · 26/08/2026
is there anyway to store private data on PDS yet? encryption such that only the owner or a dynamic, selected set of users can access a resource
110
Reposted by shubham シ
oliver @eikopf.com · 18/07/2026
not to be an annoying idiot who posts about "ownership" or whatever but i paid for that file and i'm going to do what i want with it
0263
shubham シ @shubhamshah.net · 12/07/2026
checking out @tangled.org, seems very cool!
000
shubham シ @shubhamshah.net · 24/06/2026
new username who dis :eyes:
110
Reposted by shubham シ
tautologer @tautologer.com · 16/04/2026
I love being a dilettante
141
Reposted by shubham シ
Madeon @handle.invalid · 16/04/2026
Fire Away (feat. ?????????) is next. The album’s lead single.
717140
Reposted by shubham シ
David Aerne @meodai.bsky.social · 07/04/2026
I wanted a way to explore Unicode by visual similarity, not just by name or codepoint, so I built Charcutrie. It lets you browse characters that look alike, search across scripts and symbols, and even sketch a shape to find matching glyphs. (pretty badly for now :D) charcuterie.elastiq.ch#U+221E
17540106
Reposted by shubham シ
ponder @ponder.ooo · 05/04/2026
the secret to general intelligence is to take a randomly initialized network and then remove all the subnetworks that aren't intelligent
1394
Reposted by shubham シ
Zed @zed.dev · 03/04/2026
We heard you loud and clear, so we went ahead and built the git graph.
624510
Reposted by shubham シ
Grace @gracekind.net · 03/04/2026
I hope everyone has an ontologically good morning
1819915
shubham シ @shubhamshah.net · 03/04/2026
The elites don't want you to know this but the claude code buddies at the park are free you can take them home I have 458 claude code buddies
030
shubham シ @shubhamshah.net · 03/04/2026
tez hava ne mujhse poocha ret pe kya likhte rehte ho itni muddat baad mile ho kin socho mai guum rehte ho
100
Reposted by shubham シ
Ted Han★ 韓聖安 @knowtheory.net · 02/04/2026
Please, call me Child Object, object Object was my father
A text message that reads:

Hey [object Object], if you could bring your electric bill down to 0 for the rest of the year with solar, would you want to find out how?
833858
Reposted by shubham シ
Codetaur @vibe-coded.com · 01/04/2026
we built up this crazy substrate of computers and networks and software and spread it all over the world, pushing it into every little crack we could find. and now we built a new class of organism that just happens to natively thrive in this substrate. it's gonna get a lot weirder
0413
Reposted by shubham シ
Jay 🦋 @jay.bsky.team · 30/03/2026
Bluesky is made with AI, the engineers and even some non-engineers use Claude code
1049947243
shubham シ @shubhamshah.net · 01/04/2026
lmfao
000
Reposted by shubham シ
Simon Späti 🏔️ @ssp.sh · 24/03/2026
It always starts with curiosity. I tested Claude for 2 hours, and now I have a nice-looking email TUI that works with Markdown, Neovim, and the shortcut and HEY screener workflow I like. github.com/ssp-data/neomd
github.com
GitHub - ssp-data/neomd: A minimal terminal email client for people who write in Markdown and live in Neovim.
A minimal terminal email client for people who write in Markdown and live in Neovim. - ssp-data/neomd
2301
Reposted by shubham シ
Brandon Lee @sangsara.bsky.social · 27/03/2026
Introducing DataDeck — an app that procedurally generates music from live Singapore government data feeds.
sangsara.net
DataDeck
A beautiful generative audio device that turns Singapore government data into music.
39228
Reposted by shubham シ
David Buchanan @retr0.id · 31/03/2026
have you seen the new supply chain vuln? don't update tubu. it's literally on heebee. they got poodee's deps. they infiltrated dippy. roll back weeno. disable scripts in ~/.gumpyrc. it's in poob. do not install poob. do not update poob. uninstall poob right now. poob has it in for you.
262844848
Reposted by shubham シ
hailey @hailey.at · 31/03/2026
i don't really know about other industries, but with software i have got to say that refusal to use ai at any place in development is acceptance of inability to compete. at the same time, refusal to maintain standards and quality through manual work is also acceptance of inability to compete.
931736
Reposted by shubham シ
Jay 🦋 @jay.bsky.team · 28/03/2026
Today, we’re excited to introduce Attie, currently as an invite-only closed beta. Attie is the first agentic social app on atproto. It’s something completely new — an experiment in making building on the protocol more accessible.
theliquidfrontier.leaflet.pub
The Future of AI Should Serve People, Not Platforms
19451002205
Reposted by shubham シ
Old Man Steve @omsteve.bsky.social · 31/01/2026
Got caught up in a really good book last night. I didn't stop coloring till 2 o'clock this morning.
872699162
shubham シ @shubhamshah.net · 29/01/2026
new favourite word: milquetoast
000
Reposted by shubham シ
Quest 64 Official @quest64official.bsky.social · 28/01/2026
Brian riding horseback with the text "my first day at my new job I'm so excited"

And there's a Titan in front of him I guess
939536
Reposted by shubham シ
Tom Phoolery @psychedeli.ca · 28/01/2026
unix: everything is a file the internet: no it isn’t the web: yes it is systemd: no it isn’t atproto: yes it is
624234
shubham シ @shubhamshah.net · 27/01/2026
i tried solving my itches for project ideas. boring. SO, NOW, FOR A LIMITED TIME, I WILL SOLVE YOURS! Make a software wish and I will pick whatever is most interesting to me!
292
shubham シ @shubhamshah.net · 26/01/2026
turns out the idea alr exists!
100
Reposted by shubham シ
Sol 💫 @leopard.bsky.social · 24/01/2026
Here to say I'm a rabid no borders no human is illegal globalist before it comes back in vogue.
3151
shubham シ @shubhamshah.net · 24/01/2026
010